Improvement of Buildings' Structual Quality by New Technologies Final Conference Proceedings January 2005

IMPROVEMENT OF BUILDINGS’ STRUCTURAL QUALITY BY NEW TECHNOLOGIES

PROCEEDINGS OF THE FINAL CONFERENCE OF COST ACTION C12, 20–22 JANUARY, 2005, INNSBRUCK, AUSTRIA

Improvement of Buildings’ Structural Quality by New Technologies Edited by

Christian Schaur Austria

Federico Mazzolani Italy

Gerald Huber Austria

Gianfranco de Matteis Italy

Heiko Trumpf Germany

Heli Koukkari Finland

Jean-Pierre Jaspart Belgium

Luis Bragança Portugal

A.A. BALKEMA PUBLISHERS

LEIDEN / LONDON / NEW YORK / PHILADELPHIA / SINGAPORE

This edition published in the Taylor & Francis e-Library, 2005. “To purchase your own copy of this or any of Taylor & Francis or Routledge’s collection of thousands of eBooks please go to www.eBookstore.tandf.co.uk.”

Copyright © 2005 Taylor & Francis Group plc, London, UK All rights reserved. No part of this publication or the information contained herein may be reproduced, stored in a retrieval system,or transmitted in any form or by any means, electronic, mechanical, by photocopying, recording or otherwise, without written prior permission from the publisher. Although all care is taken to ensure the integrity and quality of this publication and the information herein, no responsibility is assumed by the publishers nor the author for any damage to property or persons as a result of operation or use of this publication and/or the information contained herein. Published by: A.A. Balkema Publishers, Leiden, The Netherlands a member of Taylor & Francis Group plc www.balkema.nl and www.tandf.co.uk

ISBN 0-203-97084-5 Master e-book ISBN

ISBN 04 1536 609 7 (Print Edition)

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Table of Contents Foreword

XI

Session 1: Mixed building technologies Steel truss to concrete connection at the ski jump in Innsbruck C. Aste, A. Glatzl & G. Huber

3

Mixed advanced technologies for seismic upgrading of RC buildings F.M. Mazzolani

11

A new design proposal for timber/concrete-composite beams M. Kaliske & J. Schmidt

21

Renovation and modernization of historical buildings – structural problems ´ ˛czka A. Kozłowski & L. Sle

35

Behavior of masonry members confined with steel tying elements A. Mandara & F.M. Mazzolani

45

Retrofitting of RC structures taking as an example multi apartment gallery-access block of flats building K. Wróbel & W. Kubiszyn

57

Full refurbishment of an office building in Innsbruck G. Huber & C. Aste

65

Flexibility of building structures R. Blok & F. van Herwijnen

73

Session 2: Structural integrity 1 Seismic upgrading of RC structures by means of composite materials: a state-of-the-art review G. Della Corte, E. Barecchia & F.M. Mazzolani Structural integrity of buildings with precast load bearing walls under gas explosion G. De Matteis, I. Langone & F.M. Mazzolani

81 91

Seismic response of light-gauge steel stick-built constructions: a research project L. Fiorino, G. Della Corte & R. Landolfo

105

Shaking table test of efficiency of ALSC base-isolation system Lj. Tashkov, A. Antimovski & M. Kokalevski

115

Behaviour of wood skeleton – OSB cladding shear-walls under monotonic and cyclic loading L.A. Fülöp, S. Bálint-Major & D. Dubina Seismic retrofitting of steel and concrete structures using low-yield strength shear panels G. De Matteis, E.S. Mistakidis, A. Formisano & S.I. Tsirnovas V

125 135

Session 3: General methodology Autosystems in probability-based durability prediction A. Kudzys

149

Real-time model for earthquake prediction S. Radeva & D. Radev

155

Nonlinear response analysis of different retrofit strategies F. Aras & G. Altay

165

The behavior of different eccentrically braced frames with short links H. Köber & B. S¸ tef a˘ nescu

177

Cable material consumption depending on the geometrical parameters of hierarchic roof L. Pakrastinsh & K. Rocens

185

Wind and snow loads: some methodological problems of normative regulations B. Snarskis, R. Simkus, V. Doveika & A. Galvonaite

193

Session 4: Research and development concerning mixed building technologies On the performance based-design of a class of “adaptive” fiber-composites for applications in building structures Y.M. Haddad & J. Feng

201

Non linear procedure for the analysis of FRP reinforced frames S. Coccia, U. Ianniruberto & Z. Rinaldi

207

Stiffness of beams prestressed with FRP tendons Z. Plewako

213

Strengthening timber beams with prestressed artificial fibres: the delamination problem M. Brunner & M. Schnüriger

219

Time depended behavior of steel – reinforced glue – laminated timber beams, regarding rheology D.N. Partov & V.K. Kantchev

225

Timber–concrete-composite with an adhesive connector M. Brunner & M. Schnüriger

233

Composite of board stacks and concrete with integrated slim-floor-profiles J. Schänzlin & U. Kuhlmann

239

Composite of board stacks and concrete J. Schänzlin & U. Kuhlmann

247

Session 5: Robustness Stresses in steel columns under natural fire Z. Sokol & F. Wald

259

On the fire resistance of aluminium alloy structures B. Faggiano, G. De Matteis, R. Landolfo & F.M. Mazzolani

267

VI

Nonlinear stress–strain behavior of RC elements exposed to fire M. Cvetkovska & L. Lazarov

277

Fire resistance of garage plate-wall prefabricated structure ˇ R. Cajka & P. Mateckova

285

Performance based design of steel frames F. Dinu, D. Dubina, D. Grecea & A. Stratan

291

Flexural cyclic behaviour and low-cycle fatigue of cold-formed steel members B. Calderoni, A. Formisano & A. De Martino

301

Resistance and ductility of stainless steel bolted connections A. Bouchair

311

About the configuration of long links in eccentrically braced frames H. Köber & B. S¸ tef a˘ nescu

323

Session 6: Exceptional actions A capacity approach to the design of buildings to resist terrorist attack M.P. Byfield The collapse of WTC twin towers: general aspects and considerations on the stability under exceptional loading of columns with partial-strength connections A. De Luca, E. Mele, A. Giordano & E. Grande

333

341

Collapse analysis of timber and steel sway frames under increasing fire temperatures B.W.J. van Rensburg

351

On the structural effects of fire following earthquake G. Della Corte, B. Faggiano & F.M. Mazzolani

359

Soil–structure interaction in case of exceptional mining and flood actions ˇ R. Cajka

369

Gas explosion effects on buildings C. Bob & C. Badea

377

Session 7: Urban design 1 Current trends in tall building construction S. Kind

387

Aesthetic of historical towns and innovative constructive techniques M. Fumo & M. Naponiello

393

New living concepts for late 19th and early 20th century town housing B. Pahl, T. Hobusch, S. Kruger & C. Pluto

401

Using MBT in transformation of multi-family prefabricated buildings A. Rybka

409

The architecture of the buildings in Düsseldorf Harbour Region as a master guide for all Y.K. Aktuglu

417

The structural features of Millennium Bridge in London, connecting St. Paul and Tate Modern, as a very successful Urban Design Project Y.K. Aktuglu VII

421

Improving the performance of buildings H. Koukkari & P. Huovila

425

Effective use of cold-formed steel structures for low-story urban buildings E.L. Airumyan, O.I. Boiko & S.V. Kamynin

431

Session 8: Research and development concerning mixed building technologies On design of composite beams with concrete cracking J. Bujnak & J. Odrobinak

441

Design methodology of profiled steel sheets for composite slabs by FEM M. Ferrer, F. Marimon & F. Roure

447

Crack propagation at headed shear studs in composite beams M. Feldmann & H. Gesella

455

Load-carrying capacity of anchor plates with welded studs U. Kuhlmann & M. Rybinski

463

An experimental study of the strength and stiffness of concrete-filled steel tubular column connections with weld and stiffener angles S. De Nardin & A.L.H.C. El Debs Load-deformation and vibration-behaviour of new types of composite slim-floor slabs C. Butz, O. Hechler & H. Trumpf

473 481

Session 9: Urban design 2 Sustainable design in construction sector L. Bragança, R. Mateus & H. Koukkari

495

Aesthetics in urban design seen from the perspective of sustainability C.M. Ravesloot, L. Apon & E.M. Boelman

503

Social demands and stakeholders participation in Dutch sustainable housing policy C.M. Ravesloot

511

Energy neutral retrofitting of apartment flats – modelling and detailing with consent of inhabitants C.M. Ravesloot, L. Apon & E.M. Boelman

519

Sustainability assessment of new construction technologies: a comparative case study H. Gervásio, L.S. da Silva & L. Bragança

527

Aluminium – a sustainable building material? C. Radlbeck, D. Kosteas & M. Schlinz

537

Sustainability by LCCA of aluminium structures E. Dienes, C. Radlbeck & D. Kosteas

547

Functional assessment of lightweight construction solutions in view of sustainability L. Bragança & P. Mendonça

555

Comparative assessment of exterior walls construction solutions’ sustainability L. Bragança & R. Mateus

565

VIII

The impact of climate parameters on air movement in ventilated roofs air gap E. Monstvilas, V. Stankevicius & R. Bliudzius

573

The effect of thermal resistance value to the external deterioration due to climate impact J. Šadauskiene, V. Stankevicius & E. Monstvilas

579

Session 10: Robustness 2 Retrofitting of complex wooden structures by means of mixed reversible technologies: a study case B. Faggiano, A. Marzo & F.M. Mazzolani

587

Moment-resisting timber frames with densified and reinforced beam-to-column connections under seismic loads A. Heiduschke, P. Haller & B. Kasal

599

Experimental analysis on mechanical connections for ancient chestnut beams B. Calderoni, G. De Matteis, C. Giubileo & F.M. Mazzolani

607

Injection renovations of cracked joints A. Kudzys & R. Simkus

615

Seismic vulnerability assessment of RC structural walls M. Fischinger & P. Kante

623

Ductile design of CBF steel structures A. De Luca, E. Mele & E. Grande

629

Influence of the hole clearance on the bolted joints rotational characteristics A. Kozłowski & Z. Pisarek

639

Session 11: Structural integrity 2 Identification, repairing, strengthening and revitalisation of existing buildings structures in seismic prone areas Z.Lj. Bozinovski & K. Gramatikov

649

Micro-modeling of RC frames with masonry infill L. Krstevska & D. Ristic

657

Design of low-yield metal shear panels for energy dissipation G. De Matteis, A. Formisano, F.M. Mazzolani & S. Panico

665

Applicability of variable stiffness seismic isolators based on magnetically controlled elastomer T. Isakovi´c & M. Fischinger Numerical analyses of the wall structure under conditions mining subsidence L. Szojda

677 683

Session 12: Research and development concerning mixed building technologies Composite joints – rotational capacity U. Kuhlmann & M. Schäfer

693

IX

Characterization and qualification of reinforcement by composite of the reinforced concrete labs D.T. Nguyen, J.F. Georgin, A. Limam & J.F. Reynoard

701

Component model for steel to concrete joints D. Gregor & F. Wald

709

New technique of improving the cracking resistance of concrete walls in early age S. Wolinski

717

Investigations of the lap-joints with blind bolts of cold-formed sections W. Wuwer

725

Evaluation of behaviour of hybrid composite cable in the saddle-shaped roof D. Serdjuks & K. Rocens

733

From tree trunk to tube or the quadrature of the circle P. Haller

741

Author index

749

X

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Foreword

The COST C12 Action has been launched in 2000 and all along its whole duration, about 100 scientists from 21 of the 35 COST member countries contributed to its success. During four years, efforts have been devoted to the achievement of the main objectives of the Action which may be summarised as follows: – – – –

to develop, combine and disseminate new technical engineering technologies; to improve the quality of urban buildings; to propose new technical solutions to architects and planners; to reduce the disturbances of the construction process in urban areas, and finally improve the quality of living in the urban habitat.

The present publication includes the proceedings of the COST C12 Final Conference held in Innsbruck, Austria, from January 20 to 22, 2005. It reflects the outcome of the cooperative activities within C12 but also the views from external international experts on various topics relevant to the three following thematic fields: – mixed building technology; – structural integrity under exceptional actions; – urban design. On behalf of the scientific committee in charge of the edition of the present book, I would like to thank warmly Dr C. Schaur for the organisation of the Innsbruck conference and all the authors of the papers included in the present proceedings. The quality of their work and the involvement of all the C12 members all along the Action have been highly appreciated. Special gratitude is also addressed to Mr I. Samaras from COST offices, Mr J. Spousta and Mrs I. Silva-Ballesteros from ESF (European Science Foundation) as well as to Mr L. Bijnsdorp from BALKEMA for their support and help, respectively in administrative and publication matters. Finally, the financial support from the European Commission has to be underlined. Jean-Pierre Jaspart C12 Chairman

XI

Session 1: Mixed building technologies

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Steel truss to concrete connection at the ski jump in Innsbruck C. Aste & A. Glatzl aste konstruktion, Innsbruck/Austria

G. Huber Goetzens/Austria

ABSTRACT: The crucial challenge in MBT (Mixed Building Technology) are the connection elements between the different construction systems. A typical example of an MBT realisation is the ski jump tower in Innsbruck where a steel cage of hollow section trusses is cantilevering up to 12,5 m from the central concrete tower in a height of about 35 m above ground. The transfer of the localised truss forces into the concrete box section was solved by special prestressed docking devices. KEYWORDS: Steel-concrete construction, mixed building technology, cantilevering steelcomposite platforms, docking connection devices between steel hollow section trusses and concrete tower. 1 INTRODUCTION The ski jump in Innsbruck known for the famous annual New Year “Four-ski-jump-tour” was renewed. The original jumping tower (built for the 1976 Olympic winter games) was pulled down and a new landmark similar to a lighthouse was erected. Zaha Hadid (London) won the international architectural competition for this significant building. The constructional realisation was ordered from aste konstruktion and was honoured with the “Austrian State Award for Consulting 2002”. A speciality of this MBT are the high tension docking devices between the cantilevering steel cage and the concrete tower. 2 PROJECT OVERVIEW “Bergisel” – a glacier hill located to the south of Innsbruck (Fig. 1) – has a significant history: Offering place of the Celts, path marking for the roman–german emperors moving to Rome, battle field of the Tyrolean war of liberation, centre of the Olympic Winter Games in 1964 and 1976

Figure 1. Panoramic view of “Bergisel” with the new jumping tower.

3

with the two flame basins and – already since more than 50 years – scene of the international Four-ski-jump-tour at New Year. In the years 2001 and 2002 these facilities were fully renewed and extended by new buildings (Figs 2, 3): • Concrete tower and cantilevering tower top: described in detail in the following paper sections. • Approach ramp bridge: sagging fish-bellied steel truss girder with suspension over a direct span of 69 m, bend with a radius of 100 m, inclination of 35◦ , overall length including the take-off building 98 m, approach lane of trapezoidal composite slabs, erection with a temporary pier within four weeks. • Take-off building and ski-jump platform: concrete abutment looking like the knee of a ski jumper, length 24 m, fix support of the approach ramp bridge. • Front building: three-storey concrete building with bent roof below the jump platform – flownover by the jumpers, technical equipment, power supply, common rooms. • Landing hill: concrete fixing and border bead or retaining wall, transverse ribs, holding devices for the snow nets and the plastic mats for summer jumping, mat sprinklers. • Reporter cabin tower: four storeys for 31 cabins, steel tube frame. • Coach platform: grate platform close to the take-off building, steel tube frame. • Funicular railway: automatic cabin inclination corresponding to the slope. • Judge tower: redevelopment of the old timber construction.

Figure 2. Project overview. Office/Company Bergisel Management Assoc. Hadid aste konstruktion Pichler IMO-Bau Vorspann-Technik

City/Country Innsbruck/A London/UK Innsbruck/A Bozen/I Leipzig/D Oberndorf/A

Function architect design office steel construction sub-steel constr. bridge equipment

Figure 3. Construction board (tower).

4

Competence client ski jump incl. tower design calculations and detailing approach ramp and tower top erection of ramp and tower top pre-stressing, ramp suspension

3 CONCRETE TOWER With 49 m above ground the ridge height of the concrete tower reaches 791 m above sea level (Fig. 4). The foundation was solved with a plate of 20 × 20 × 1, 0 m at a level of −11 m below ground with three basement storeys. The standard cross section of 7 × 7 m and a wall thickness of 40 cm rises up for about 40 m above the foundation, stabilised with wall diaphragms to the base plate limits. It contains the two elevators, the stairway and the supply pit. From 29 m above ground the cross section tapers to 3,7 × 7 m making place for the jumping access stairway. Also at this level of cross section change the support girder for the ramp bridge cantilevers 4,5 m with a height of only 1,45 m. This slenderness was necessary to hide this girder in between the steel truss flanges of the bridge. The demand of fair-faced concrete in combination with the difficult access and supply conditions resulted in the choice of a climbing formwork. Concreting started in June 2001.

Figure 4. Concrete tower.

5

Figure 5a. Tower top – an architectural challenge.

Figure 5b. Tower top – sketch of the steelwork construction.

4 STEEL-COMPOSITE TOWER TOP The tower top is not ordinary – neither in architectural nor in constructional respect. A three-level steel cap with a rescue level, a restaurant and an observation platform is docked to the central concrete tower (Figs. 5a, 5b). Being 250 m above the city centre one has a fantastic view on Innsbruck and the surrounding mountains. 6

Figure 6a. Slotted hollow sections welded to the docking brackets.

Figure 6b. Cantilevering steel cage during construction.

The levels are cantilevering around the concrete core up to 12,5 m. Together with the steel hollow section frames and the diagonal suspension tubes anchoring back to the concrete core a steel cage is built (Figs. 6a, 6b). The horizontal stiffening to the core is realised by the trapezoidal composite slabs. The transparency and elegance of the facade is supported by the fact that diagonal bars within the front could be avoided and huge glass elements were placed into the facade.

5 DOCKING DEVICES The crucial challenge in MBT construction are the connection elements between the different construction systems. Thanks to the common, material-independent design and safety concept of the constructional Eurocodes the interface problems at the level of design methods and internal forces lost its deterrent effect and MBT solutions become more and more usual in daily design practice. The effect is a more economical use of different materials related to their constructional benefits (strength, stiffness, weight, prefabrication, strengthening, dismanteling, …) and more innovative architectural solutions. For the actual case of the Bergisel Ski Jump the considerable docking forces between the steel cap and the concrete core had been a crucial challenge which was solved by special pre-stressed steel brackets (Fig. 7). These elements of at maximum 550 kg weight were integrated into the formwork with a tolerance of less than 1 cm. The load transfer into the concrete walls was handled with mutual pre-stressing cables (interior tendons) from one docking point to the opposite one going through the tower. Additional concentrated rebars in the local load introduction zones were provided to cover the bursting forces and for crack distribution. The characteristic docking forces can be taken from Figure 8. Pre-stressing was applied from the opposite side of the fixed anchor after hardening of the concrete and before connecting the steel sections. Depending on the tension forces either one or two strands were provided. The conduits were then filled with injection grout against corrosion. 7

Figure 7. Pre-stressed docking devices.

Figure 8. Characteristic docking forces.

The resulting necessary welding length led to the geometry of these docking brackets. By the use of four longitudinal ribs which were welded on site to the slotted push-over hollow sections the total bracket length could be minimised. The eventual negative influence of the high welding temperatures on the end anchorage of the pre-stressing strands could be dispelled by a test specimen. The maximum heat increase was measured to be only 50◦ C. Attention has to be paid to the fact that the tendon head is no more accessible after positioning of the steel cage. Therefore this application type is limited to quasi-static loading. 6 CONCLUSIONS The new building at Bergisel proved to be an excellent combination of architectural shape and constructional design. Fair-faced concrete, steel and glass together with the harmonious longitudinal 8

Figure 9. Illuminated structure at night.

section and the top view are showing the worldwide appreciated style of Zaha Hadid. Construction and erection were based on modern steel-concrete mixed building technology: concrete core with climbing formwork, pre-stressed steel docking brackets for the steel frame cage on the tower top, a pre-stressed very slender concrete cantilever as upper support of the approach ramp, three-level widely cantilevering steel frame cage on top, approach ramp in the form of an organic fish-bellied and suspended trough bridge – all in all “Toccata and Fugue in major F” for a civil engineer and his orchestra. REFERENCES Aste, C., Glatzl, A., Huber, G. 2002. Ski jump “Bergisel” – A new landmark of Innsbruck. 3rd Eurosteel Conference, ISBN 972-98376-3-5, Lisbon. Aste, C., Glatzl, A., Huber, G. 2002. Schisprungschanze “Bergisel” – Ein neues Wahrzeichen von Innsbruck. Stahlbau Heft 3/02, S.171–177, ISSN 0038-9145, Ernst & Sohn, Berlin. Aste, C., Glatzl, A., Huber, G. 2003. Steel-concrete mixed building technology at the ski jump tower of Innsbruck, Austria. International Journal on Steel and Composite Structures, ISSN 1229-9367, Technopress, Korea. Aste, C., Glatzl, A., Huber, G. 2003. Innsbruck ski jump: a triumph of mixed building technology. Concrete Journal, The Concrete Society, Berkshire. Aste, C., Glatzl, A., Huber, G. 2003. Schisprungschanze “Bergisel” – Ein neues Wahrzeichen von Innsbruck. Zement + Beton, www.zement.at, Wien. Huber, G., Aste, C. 2004. Steel truss to concrete connection. COST-C12 – Output of the co-operative activities, Balkema.

9

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Mixed advanced technologies for seismic upgrading of RC buildings F.M. Mazzolani Dept. of Structural Analysis and Design, University of Naples “Federico II”, Naples, Italy

ABSTRACT: The paper presents the general planning of a physical/numerical testing of two fullscale existing reinforced concrete buildings. The research investigates the efficiency of different advanced mixed techniques for seismic upgrading/retrofitting of RC structures. The buildings under study were built at the end of 70s, for resisting only vertical loads and according to out-of-date technical codes. They are destined to be demolished. Thus, they are used as full-scale specimens to be tested up to collapse. The research involves several universities, national institutions and industrial companies. One of the building was divided into six separate sub-structures, in order to examine some seismic upgrading/retrofitting systems: 1. base isolation; 2. concentric buckling-restrained steel braces; 3. carbon fiber reinforced polymers; 4. eccentric steel braces; 5. concentric shape memory alloy braces; 6. steel shear panels. The experimental campaign is going to be finalized on the modified sub-structures of the first building and it is still in progress on the second one.

1 INTRODUCTION Seismic upgrading of existing gravity-load designed (GLD) reinforced concrete (RC) buildings is an important subject in the field of earthquake engineering. In fact, for this type of structures recent earthquakes (Northridge, U.S.A. 1994; Kobe, Japan, 1995; Izmit, Turkey, 1999; Athens, Greece, 1999; ChiChi, Taiwan 1999) demonstrated the lack of adequate protection against both damage and collapse. Hence the need to improve the current structural design codes for seismic zones, in order to safeguard human lives, but also to limit damage and loss of functionality in buildings and facilities after a strong earthquake, is highlighted. Several technical solutions are currently available for the mitigation of earthquake risks, going from active to passive dissipating devices as well as base isolation. In general, seismic repairing/ upgrading structural systems can be classified according to the following conceptual scheme: 1. Systems based on adding new structural elements, which directly operate at global level for improving the seismic response. 2. Systems based on repairing/upgrading existing structural elements, aiming at improving the global response by changing the local behaviour. Type 1 systems are useful for seismic upgrading in those situations characterised by the absence of purposely-designed lateral-load resisting structures, such as in the case of GLD buildings. A correct design of these systems is based on the idea to eliminate/reduce the plastic deformation demand to the existing structure by adding supplemental energy dissipating devices. Among type 1 systems, metal-based technologies are often considered as the most satisfactory technical solutions, because of the effectiveness, practicality and economy (Mazzolani, 1992, 1996, Mazzolani & Mandara, 2001). Metal solutions mainly consist in adding new structural elements in the form of braces, which collaborate with the existing structure, varying its static scheme and operating at global level as supplemental energy dissipation passive systems. The bracing systems are designed according to modern knowledge of earthquake engineering, so that they may significantly improve 11

safety of existing buildings against lateral collapse. Steel-based solutions are very affordable, but, as they rely on steel yielding for dissipation, they are also affected by the problem of residual deformations of the structure after the earthquake. This drawback could be overcome by means of innovative shape-memory alloys (SMA) solutions, which, based on the super-elastic properties of such materials, allow a self-centring capacity of the structure after the earthquake (Dolce et al., 2000). In the latter case, the energy dissipation could be integrated through the addition of viscous damping devices. The effectiveness of many of such systems in protecting both structural and nonstructural components has long been demonstrated by both theoretical and experimental studies. An overview of classic passive control systems is given in Kasai et al., 1998. Among type 2 systems, a multitude of seismic rehabilitation/reparation techniques is proposed and studied. With reference to the traditional systems, the following practices can be mentioned: (a) epoxy injections; (b) steel plating or concrete jacketing (Alcocer & Jersa, 1993). More recently, the use of fibre reinforced polymers is spreading (Cosenza & Nanni, 2001). Nowadays, a lot of both theoretical studies and experimental tests of reinforcing systems on structural elements and sub-structures have been performed. Laboratory experiences are valuable for studying the intervention techniques, but they present important restrictions, concerning the difficulty to well reproduce the actual boundary conditions for the single structural element or substructure, sometimes the scale-effect, the difficulty to reproduce the actual RC structure defects (e.g. reinforcing bars corrosion and/or concrete degradation). It is apparent as the opportunity to perform collapse tests on existing structures must be considered a precious and unavoidable occasion to improve the knowledge on both design and analysis methods. What is more, the major part of recent studies mainly examines each technical solution independently from the others. Therefore the surplus value of the current experimental investigation consists in both the analysis of two existing buildings and the examination and comparison of different technologies for building upgrading.

2 RESEARCH SUMMARY 2.1 The buildings The buildings under investigation are located in the Bagnoli district of Naples (Italy). In this site a very important industrial plant of the Italian steel producer, named “ILVA” (or “Italsider”), was installed at the beginning of the last century. After the decision of the European Community to reduce the Italian production of steel, many iron and steel mills were closed, including the Bagnoli plant, whose site was recognized as very attractive from the viewpoint of the tourist exploitation. So, the ex-industrial area of Bagnoli is now under reorganization, in order to convert it into a residential and tourist one. As a consequence, a lot of constructions located in Bagnoli were already demolished or they will be in few years. However, several of such buildings possess cultural value from the perspective of the structural engineering, because they represent the construction practice for RC buildings during the 60s–70s in the South of Italy. They were designed and constructed for resisting mainly vertical loads, according to old seismic codes for constructions. Moreover, they often underwent material degradation because of aggressive environmental conditions. For such constructions the urgency of evaluating the seismic vulnerability and identifying the appropriate upgrading/retrofitting structural system exists. In particular, the examined buildings were designed and constructed at the end of the 70s, few years before the occurrence of the Campano-Lucano earthquake (Irpinia, Italy, 1980). At that time, the city of Naples was even not considered to be exposed to significant seismic actions, so that the practice was to design either without or with only very small attention to the horizontal load-resisting structural scheme. Inspired from this occurrence, the acronym of the research project has been decided: «ILVA-IDEM (ILVA Intelligent DEMolition)». Figure 1 shows the buildings before the experimental investigation. The first building (Figure 1a) has a rectangular lengthened plan shape (41.6 m × 6.50 m) and it is on two floors with a first and second floor heights on the ground of 3.55 m and 6.81 m, respectively. Beams are set only along the building perimeter and they support hollow tiles mixed slabs. In order 12

Figure 1. Global views of the original buildings.

Figure 2. Building (a) – Techniques under study: 1. BI; 2. BRB; 3. C-FRP; 4. EB; 5. SMA; 6. SP.

to increase the potential number of specimens and to test different upgrading solutions, slabs were cut at the first and second floor, in such a way to divide the whole building into six separate simple structures to be analysed. Before the cutting of the slabs, both the internal partitions and the external claddings of the building were removed. Figure 2 illustrates the building at the end of such preliminary operations. The second building (Figure 1b) has a rectangular, more compact as respect to building (a), plan (12.00 m × 18.50 m) and it is on two floors with a first and second floor heights on the ground of 5.10 m and 9.45 m, respectively. In this case, the building is going to be tested entirely with all the completion elements. 2.2 Research planning The general planning of the research can be summarized by the following main 8 steps: Step 1 Step 2 Step 3 Step 4 Step 5 Step 6 Step 7 Step 8

As-built data collection. Identification of dynamic properties of existing structures by means of in-situ testing. Static pushover test of structures (to be used as reference response). Modelling of the original structures (using data coming from steps 2 and 3). Design and application of seismic repairing/upgrading systems. Identification of dynamic properties of upgraded structures by means of in-situ testing. Static pushover tests of the upgraded structures. Modelling of the upgraded structures (using data coming from steps 6 and 7).

The experimental campaign started on building (a). The seismic upgrading techniques, which have been selected, designed and applied, are shown in Figure 1. A summary of the 8 research steps previously listed are given in the remaining of the paper. For more details about the research activity carried out on the building (a) reference can be made to papers Mazzolani et al., 2004a, b, c, Della Corte et al., 2003, 2004a, b. 3 STEPS 1 AND 2 The geometry of each structure was measured in-situ and data about the structural member sizes, the slab arrangement and the steel reinforcements were acquired. As an example, Figure 3 shows 13

Figure 3. Measured geometry of structure n. 3 of building (a).

4,50 4,20

0,30

4,85 4,55

4,35

3,75 0,60

4,60

8,95

(+ 5.10)

(+ 5.10)

(+ 0.70)

(+ 0.60)

6,00

(+ 0.70) 6,30

(+ 9.45)

4,00

0,60

3,75

4,35

(20x60)

(15x60)

(15x60)

11 (20x60)

10

4,60

8,95

5,40

5,10

9 (25x60)

(20x60)

12,00

8

7

6

(+ 9.45)

5

4,00

4

3

(20x60)

0,60

18,50 4,55 4,25

2

0,60

TRANSVERSE SECTIONS

FIRST FLOOR (+5.10) 4,30 4,00

1

(+2.90)

12

13 (20x60) 4,00 430

14 3,45 375

15 2,00 230

16 2,70 300

17 4,55 485

Figure 4. Measured geometry of building (b).

some details of the survey of structure n. 3, as a typical sub-structure of the building (a); whereas Figure 4 shows some details of the survey of building (b). Material properties were measured both by means of in-situ non-destructive tests and laboratory tests on samples drawn out from the original buildings. Dynamic properties of existing structures were measured through in-situ tests, by means of three techniques: 1. harmonic vibrations by means of a vibrodine; 2. impulsive loading by means of a special hammer; 3. vibrations induced by an heavy body impacting the ground close to the structure. Results are carried out in terms of both frequencies and relative viscous damping (Mazzolani et al., 2004a, b).

4 STEPS 3 AND 4 At first, a static pushover test up to collapse was performed. A monotonically increasing lateral force was applied, through the use of hydraulic load jacks contrasted by means of the retaining structures shown in Figure 5. As it can be seen, the lateral load is applied by means of a single point load-actuator, contrasted against two vertical systems of beams, one placed on the testing structure and the other on the reacting frame. This lateral force is applied in such a way to have an inverted triangular lateral force distribution on the structure to be tested, in order to experimentally replicate the usual assumption of pushover studies. Lateral story displacements are measured by means of optical-electronic devices. For building (a) the static pushover test was performed on structure n. 3, whose response was destined to be used as reference behaviour (Mazzolani et al., 2004c). In Figure 6 the deformed configuration at collapse and some details of the damaged structure are shown. For building (b) the retaining structure is just been erected and the static pushover test is going to be performed. 14

(a)

(b)

Figure 5. The reaction frame systems for sub-structures of building (a) and for building (b).

Figure 6. Structure n. 3 of building (a) – Deformed configuration at collapse and plastic hinge details.

Numerical models of the original structures have been developed. Both plastic hinge models and fiber beam-column elements have been alternately used. The plastic hinge model was implemented in the well-known commercial software SAP 2000NL v.7.12 (Wilson, 1998). Concrete cracking was indirectly taken into account by fictitiously reducing the moments of inertia of beams and columns. The strength properties of plastic hinges were computed using the average measured strength for concrete. On the contrary, a reduced value of the steel yielding strength was required in order to match the experimental results. The fiber model was implemented in the recently developed software platform named OpenSEES (Mazzoni et al., 2003). In this type of model, the tension strength of concrete was assumed as equal to zero, while its compressive behaviour was assumed according to different hypotheses, namely both considering and neglecting the effect of steel stirrups confinement. Finally, the steel yielding strength was assumed as coming from the calibration phase of the plastic hinge model. The comparison among the different numerical assumptions and the experimental results was made, catching the best agreement between models and experimental results. Further details about the structure modelling and the calibration phase are given in Della Corte et al., 2004b.

5 STEP 5: DESIGN AND APPLICATION OF SEISMIC REPAIRING/UPGRADING SYSTEMS Within the planned experimental investigation on building (a) three seismic upgrading/retrofitting systems, which consist in different types of bracing, as well as a seismic repairing intervention by means of composite materials, are tested. Moreover dissipative steel shear panels and a base isolation system are also taken in consideration. More in detail the examined and applied techniques are: 1. A base isolation (BI) system, using a relatively new type of high damping rubber bearing, with a ring shape. The latter originates from the need to have a stable bearing under lateral loading even if the plan size is small due to the low value of vertical loads. A horizontal steel bracing system assures the diaphragm effect at the base level (Figure 7). 2. A concentric bracing system, based on buckling-restrained braces (BRB), obtained by inserting a steel rectangular plate between two rectangular hollow-section tubes, which have the only 15

Figure 7. The base isolation system (BI).

Figure 8. The concentric buckling-restrained steel braces (BRB).

Figure 9. Seismic repairing by means of composite materials (C-FRP).

Figure 10. The eccentric steel bracing system (EB).

function to stabilise the internal plate from lateral buckling, thus allowing the yielding of the steel plate under external actions (Figure 8, Della Corte et al., 2003). 3. A seismic repairing intervention, using Carbon-Fiber Reinforced Polymers (C-FRP) for reinforcing the damaged columns of the structure already tested under monotonic loading conditions (Figure 9, Della Corte et al., 2004a). 4. An eccentric bracing (EB) system, with the link vertically placed and directly attached to the slab of the existing structure. Links are properly designed for mainly undergoing shear deformation, i.e. “shear links” or “short links” (Figure 10, Della Corte et al., 2003). 16

Figure 11. The shape memory alloy braces (SMA).

Figure 12. The dissipative steel shear panels (SP).

Figure 13. The C-FRP-reinforced structure before and after the experimental test.

5. A concentric bracing system with braces made-up of shape memory alloys (SMA), which exploits the super-elastic properties of such materials for permitting the self-re-centering of the structure after the earthquake (Figure 11, Dolce et al., 2004). 6. Dissipative steel shear panels (SP, Figure 12). This part of the research is directly linked to another experimental activity, which is under development at the Department of Structural Analysis and Design of the University of Naples “Federico II” (Panico et al., 2003). Here, pure aluminum shear panels are being tested. Pure aluminum, when heat-treated, exhibits low strength and large ductility, allowing energy dissipation in shear activated by the interstory lateral displacements of the main structure. With reference to Figure 2, the BI system, the BRB system, the C-FRP system, the EB system, the SMA system and the SP system are applied to sub-structures n. 1, 2, 3, 4, 5 and 6 respectively. Figures 7 through 12 give a general representation of the BI, BRB, C-FRP, EB, SMA and SP systems, respectively. 6 STEPS 6 THROUGH 8 A summary of the progress state of the presented experimental activity is given in this section. Concerning the building (a), structures modified by the C-FRP, EB and SMA systems have already been tested (Figures 13, 14, 15) and now results are going to be elaborated. The BRB 17

Figure 14. The EB-reinforced structure before and after the experimental test.

Figure 15. The SMA-reinforced structure before and during the experimental test.

system, it has been already designed, constructed and erected in the existing structures. The design of both the base isolation system (BI) and the dissipative shear panels (SP) is currently in progress. These seismic protection devices should be soon constructed and erected. The above-mentioned seismic up-grading systems will be also tested under cyclic loading conditions. ACKNOWLEDGEMENTS The studied RC buildings have been offered free of charge by the Bagnoli Company. In the period 2000–2002 the research activity was developed on voluntary basis only. Recently, we acknowledge the financial support from the BagnoliFutura Company for the erection of a new retaining wall. This activity is partially included in the budget of the research unit “Development of behavioural models of innovative devices for the structural preservation” within the CNR-MIUR national research project titled “Analysis and protection of architectural constructions against the effect of earthquake and other natural calamities”. Finally, the financial support by Regione Campania (L.R. N. 5 28.03.2002) is acknowledged. The author would like to acknowledge all the research staff, which is composed by Barecchia E., Calderoni B., D’Aniello M., Della Corte G., De Matteis G., Faggiano B., Fiorino L., Formisano A., Giubileo C., Panico S., Landolfo R., for the precious contribution offered to the research experimental activity for the past, the present and the future.

REFERENCES Alcocer S.M., Jersa J.O. “Strength of reinforced concrete frame connections rehabilitated by jacketings.” ACI Structural Journal, American Concrete Institute, Detroit, MI, USA, Vol. 90(3), pp. 249–261, 1993. Cosenza E., Nanni A. (editors). “Composites in Construction: A Reality.” Department of Structural Analysis and Design, University of Naples Federico II, Naples, Italy, 2001. Della Corte G., Barecchia E., Mazzolani F.M. 2004a “Seismic upgrading of existing RC structures using FRP: a GLD study case.” Proceedings of the First International Conference on Innovative Materials and Technologies for Construction and Restoration, Lecce, Italy, accepted for publication, 6–9 June.

18

Della Corte G., Faggiano B., Mazzolani F.M. 2003. “Innovative steel bracing systems for seismic upgrading of reinforced concrete structures: planning of a testing program.” Proceedings of the XIX CTA National Congress, Genova, Italia, pp. 393–406. Della Corte G., Faggiano B., Mazzolani F.M. 2004b. “The ILVA-IDEM project: a full scale pushover test on an existing RC structure – Part II Modelling Issues.” Proceedings of the XI Congresso Nazionale “L’ingegneria Sismica in Italia”, Genova, Italia, 25–29 January. Dolce M., Cardone D., Marnetto R. 2000. “Implementation and testing of passive control devices based on shape memory alloys.” International Journal of Earthquake Engineering and Structural Dynamics, Vol. 29, N. 7, July. Dolce M., Cardone D., Marnetto R., Nigro D., Ponzo F.C., Santarsiero G., Mucciarelli M. 2004. Pro-getto ILVA-IDEM: sperimentazione in-situ dell’adeguamento sismico mediante controventi ricentranti con leghe a memoria di forma. Proceedings of the XI Congresso Nazionale “L’ingegneria Sismica in Italia”, Genova, Italia, 25–29 January. Kasai K., Fu Y., Watanabe A. “Passive control systems for seismic damage mitigation.” Journal of Structural Engineering, ASCE, Vol. 124, No. 5, May, 1998. Mazzolani F.M. “The use of steel in refurbishment.” Proceedings of the 1st World Conference on Constructional Steel Design, Acapulco, Mexico, 1992. Mazzolani F.M. “Strengthening options in rehabilitation by means of steel works.” Proceedings of the 5th International Colloqium on Structural Stability (SSRC), Brazilian Session, Rio de Janeiro, 1996. Mazzolani F.M., Mandara A., “Advanced metal systems in structural rehabilitation of monumental constructions.” Proceedings of the International Conference on Structural Engineering, Mechanics and Computation (invited lecture), Cape Town, South Africa, 2001. Mazzolani F. M., Della Corte G., Faggiano B. 2004a. Seismic upgrading of RC buildings by means of advanced techniques: the ILVA-IDEM project. Proceedings of the 13th World Conference on Earthquake Engineering, Vancouver, B.C., Canada, August 1–6, Paper No. 2703. Mazzolani, F.M., Calderoni, B., Spina, D., Valente, C. 2004b. The ILVA-IDEM project: structural identification of the existing building. Proceedings of the XI Congresso Nazionale “L’ingegneria Sismica in Italia”, Genova, Italia, 25–29 January. Mazzolani F.M., Della Corte, G., Calderoni B., De Matteis G., Faggiano, B., Panico S., Landolfo R., Dolce M., Spina D., Valente C. 2004c “The ILVA-IDEM project: a full scale pushover test on an existing RC structure – Part II Modelling Issues.” Proceedings of the XI Congresso Nazionale “L’ingegneria Sismica in Italia”, Genova, Italia, 25–29 January. Mazzoni, S., McKenna, F., Scott, M., Fenves, G.L., Jeremic, B. “Command Language Manual – Open System for Earthquake Engineering Simulation (OpenSees)”, 2003. Panico S., De Matteis G., Mazzolani F.M. 2003. Numerical Investigation on pure aluminium shear panels. Proceedings of the XIX CTA National Congress, Genova, Italia, September, Vol. 2, pp. 459–470. Wilson E.L. “Three dimensional static and dynamic analysis of structures.” Berkeley, California, USA: Computers & Structures, Inc., 1998.

19

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

A new design proposal for timber/concrete-composite beams M. Kaliske & J. Schmidt Institute for Structural Mechanics, University of Leipzig, Germany

ABSTRACT: Timber/concrete-composite systems are efficient and economical structures to reinforce timber ceilings. Novel non-linear finite element models enable reliable structural analyses of this class of constructions. Comparisons of finite element simulations and experimental results show the suitability of the numerical models to simulate the load-bearing behavior and long-term features (see Schmidt et al. 2002a, 2003a, 2003b). The design of timber/concrete composite beams is currently carried out according to DIN 1052 and EC 5, respectively, based on linearly calculated internal forces (approximated by so-called γ -procedure). However, the real load-bearing behavior shows considerable non-linearities. For this reason and because of simplifications of the γ -procedure, the safety zone between working load and failure load is not constant depending on the system parameters like the load-bearing behavior of the joints. A comparison between failure loads determined by finite element analysis and the working loads according to the current design rules will be shown. The influence of the non-linear behavior of the connectors on the load-bearing behavior of the composite beams is important. It is shown that the decrease of load-bearing capacity is smaller than assumed by current code of practice. Structures with different spacing of connections have the largest safety-factor (see Schmidt et al. 2004). These systems can be designed more economically. As result of the investigations, a new design proposal is presented, which takes nonlinearities into account and guarantees a constant safety-zone between failure load and working load. This proposal permits an economical design of timber/concrete composite beams.

1 INTRODUCTION A non-linear finite-element-model is presented (cp. Figure 1), which is verified by experimental tests (Schmidt et al. 2002a). In Schmidt et al. 2003a, it is investigated if E DIN 1052: 2000-05 is suitable for an economical design and if the distance to the failure load (safety zone) is always guaranteed. It is shown that the safety zone, which results from the design, is larger for composite beams with ductile behavior compared to composite beams with brittle properties. This is in contrast to the goal of a design approach. The discussed layout is based on the Eurocode safety-concept with partial safety factors γM . The design value of the strength Xd arises according to

from the characteristic value Xd and the modification factor kmod , which takes the influence of the surrounding climate into account. The field of application of timber/concrete composite structures for ceilings of residential buildings or office buildings necessitates the inclusion of climate conditions of heated inner rooms. Therefore, the modification factor for concrete is kmod = 1,0, but generally kmod = 0,7 is used. For the employed materials partial safety factors • timber γM = 1,3 (according to reference E DIN 1052), • concrete γM = 1,5 (according to reference DIN1045), • connection γM = 1,3 21

Figure 1. Schematic model of FE-geometry. Table 1. Geometry of standard composite beam.

Strength class b/h

Timber (index t)

Concrete (index e)

Intermediate layer

C24 according to E DIN 1052 140/260 mm

C25/30 according to DIN 1045-1 600/60 mm

E = 1000 N/mm2 Thickness  = 20 mm

are utilized. In the subsequent investigation, the loading-zone of the line load is located at the upper side of the concrete, vertical to the timber beam with the width bt . The partial safety factor for the connection is a suitable proposal. EC4 and DIN 18800 also use this value, if ductile load–displacement-behavior is ensured. Furthermore, Blaß et al. 1995 propose a global safety factor of γ = 2,5 between service load and failure load (2,5 ≈ γM · γF /kmod ). A linear design rule for timber-composite beams is presented in DIN 1052. The field of application is enhanced for timber/concrete composite beams in EC 5. 2 PARAMETER OF INVESTIGATION The load–displacement-behavior of the connection influences the load-bearing behavior of the composite beam significantly. Therefore, the influence of the joints is determined for standard timber/concrete-composite beams with a span of L = 5,50 m. The geometry is given in Table 1 (cp. also Figure 2). The characteristic load–displacement-behavior of the joints is assumed according to the typical multi-linear elastic/plastic joint characteristic of hexagonal head wood screws (cp. also Schmidt et al. 2002b). This load–displacement-behavior is as expected and has also been determined by other types of joints. Thus, the following investigation is valid for a large number of timber/concreteconnection types. The characteristic load–displacement-behavior, which is normalized with respect to the spacing e of joints, is given in Figure 3. The connection is defined by the carrying load Rv,k applied to the spacing of joints e

22

Figure 2. Cross section – symbol definition.

Figure 3. Normalized load–displacement-characteristic of connection 0,025 kN/mm ≤ rv,k ≤ 0,25 kN/mm, increment 0,0125 kN/mm.

Additionally, the influence of gradation with respect to the spacing of joints is investigated. The spacing is small near the support. With increasing distance to the support, it is possible to change the spacing of joints. The stiffness of joints and the load carrying capacity of joints reduce to the kc -fold value. Four types of gradation are investigated: • • • •

type A, without gradation (e = const.), type B, one gradation at L/4, kc = 0,5, type C, two gradations at L/6, kc = 0,67 and at L/3, kc = 0,33 type D, three gradations at L/8, kc = 0,75, at L/4, kc = 0,5 and at 3/8L, kc = 0,25.

The gradation of the spacing of joints is shown in Figure 4. 3 ULTIMATE LIMIT STATE At the ultimate limit state, the design value of action Ed is equal to the design value of resistance Rd . A timber/concrete-composite beam is loaded with a unit line load 1 and the maximum stresses 23

Figure 4. Composite characteristics along beam length taking symmetry conditions at L = 2750 mm into account.

σ (1), τmax (1) and the maximum joint force Tmax (1) are calculated. The design value of the maximum line load, which the structure can carry, results according to E DIN 1052 in

Line one in equation (3) is the combination proof of timber for tension (second index t) and for bending (second index m). Line two is the proof of compressive stress in concrete and, if necessary, in timber (in case of very small stiffness of connection). Line three services the proof of shear stresses in timber and line four is the proof of the connection. The nonlinear load-deflection-behavior is computed with the FE-Model (cp. Figure 1). Additionally, the load-bearing-behavior of the pure timber beam (140/260 mm) and the design value of the line load qmax,d , which the timber beam can carry, is calculated. These values and the design value of maximum line load according to equation (3) are shown in Figure 5. Composite beams with very strong connections do not reach a deflection of 80 mm. These structures reach their carrying capacity due to concrete compressive failure before 80 mm of deflection. Figure 5 reveals that the bending stiffness and carrying capacity of timber/concrete-composite beams increase with an increase of joint stiffness and joint capacity. This observation is in accordance with the expectation. Moreover, the design values of maximum line loads according to E DIN 1052 are increasing much more than the real carrying capacity of the structure (computed by FEM) compared in percent. Structures with strong connections rv,k > 0,20 kN/mm fail rather brittle. In contrast, the load-bearing behavior of beams with low or normal stiffness of joints rv,k < 0,15 kN/mm) is ductile. For the linear design proposal (see equation (3)) the proof of connection (line four) and the proof of timber (line one) is governing. The design values according to this proposal are shown in Figure 5 24

Figure 5. Load carrying behavior of standard timber/concrete composite beams without gradation of connections type A (black) and pure timber beam (grey) and the design values of maximum line load qmax,d according to the linear design concept of equation (3) (dashed).

Figure 6. Load carrying behavior of standard timber/concrete composite beams with one gradation of connections type B (black) and pure timber beam (grey) and the design values of maximum line load qmax,d according the linear design concept of equation (3) (dashed).

with dashed lines. In contradiction, the design values of the maximum line load of composite beams with rv,k < 0,10 kN/mm (long dashed) is less than the design value of the pure timber beam, although, the carrying capacity of composite beams is higher than the carrying capacity of the pure timber beam. Furthermore, the safety factor is larger for structures with ductile behavior than for structure with brittle behavior. These findings are in conflict with the goals of an optimal design (cp. Kersken-Bradley 1992). Thus structures with a moderate stiffness/load capacity of connection and ductile load–deflection-behavior are penalized with respect to structures with strong connection and rather brittle failure. This fact also has been found for structures with gradation of the joint spacing. Figures 6, 7, 8 show the load-bearing behavior of composite beams, the design value of maximum line load qmax,d 25

according to linear design proposal (equation (3)) (dashed) as well as the load–deflection-path of a pure timber beam (grey) and its design value of maximum line load (grey-dashed). The comparison of the figures shows that the carrying capacity decreases, but not as much as proposed by the linear design approach according to E DIN 1052. This observation yields the contradiction, that the majority of the design values of maximum line load for composite beams type C and type D is lower than the design values of the pure timber beam. Just some composite beams with very strong connections have a higher design value as the pure timber beam. But these structures fail rather brittle and, therefore, are not recommended for practice. The actually obtained

Figure 7. Load carrying behavior of standard timber/concrete composite beams with two gradations of connections type C (black) and pure timber beam (grey) and the design values of maximum line load qmax,d according the linear design concept of equation (3) (dashed).

Figure 8. Load carrying behavior of standard timber/concrete composite beams with three gradations of connections type D (black) and pure timber beam (grey) and the design values of maximum line load qmax,d according the linear design concept of equation (3) (dashed).

26

increase of carrying capacity cannot be used. The resulting safety factor is significantly higher than usually in timber engineering. The economical gradation of joints yields the largest safety zones, although, no decrease of ductility is determined. With a constant safety zone a more economical design is possible for the majority of these timber/concrete-composite structures. Therefore, with following the new design proposal leads to an economical design with constant safety factors. By this proposal, the real non-linear joint characteristic is taken into account.

4 NON-LINEAR DESIGN PROPOSAL The features subsequently mentioned are important for structures: redistribution of load-carrying effect as consequence of local failure, ductility and threat of collapse (see Kersken-Bradley 1992). The constructive conditions are chosen in order to guarantee ductile behavior. This goal is reached by the use of ductile connections. In this case the joints reach their maximum loadcarrying capacity significantly before timber and concrete. The stiffness of connection should have a sufficient initial value. Otherwise, the contribution of the normal force to the loadcarrying behavior is very small and the flexural rigidity is nearly equal to the sum of bending stiffness of timber beam and concrete plate. Then, the concrete is bending and will crack. Furthermore, “yielding” of joints has not an effect on the reduction of the flexural stiffness of the composite beam. In this case, the failure of composite beam is not ductile and the failure is not prefigured. A sufficient mount of joints ensure a robust structure. A local failure of one or some joints is not significant for the global system behavior. The new design proposal is in analogy with the concept of DIN 18800-5. During the construction phase of timber/concrete-composite ceilings, the timber beam is supported. The support is removed when the concrete is sufficiently solid. Thus, the consideration of composite effects for self weight is admissible. The internal forces are elastically calculated. The new design proposal assumes ductile joints. This assumption is correct, when the load– displacement behavior of joints justifies an elastic-plastic behavior of the connection (cp. DIN 18800). The loading of the connection along span depending on the normalized line load q/qu is shown in Figure 9 for a composite beam type A and rv,k = 0,15 kN/mm.

Figure 9. Saturation of connection T /Rv,k along beam with connection type A (rv,k = 0,15 kN/mm) depending on the load factor q/qu .

27

The saturation rate for connection with n joints

increases with an increase of the line load. The joint forces increase under-proportionally near support and over-proportionally near midspan. At the load factor of 0,82, the joints near the support are fully loaded. With an additional increase of loading, the joints closer to midspan are further loaded. At the ultimate load state, the saturation factor is kv ≈ 0,90. As the result, this connection type is really an economical approach. For high load levels, the non-linear load–displacement-behavior of joints yields on overproportional increase of stresses in timber and concrete. Cracking at the bottom side of the concrete plate is possible. The ultimate load of the timber/concrete-composite beam is reached, when the concrete fails under pressure or timber fails under tension and bending. To sum up, structures with a moderate stiffness of joints are efficient, because all components have reached their ultimate capacity. Furthermore, these structures have a ductile behavior and the system collapse will be signalized by large deflections. This publication proposes to waive the proof of joints according to line four of equation (3) because the joints are ductile and, therefore, the composite beam is also ductile. For this reason, the design value of ultimate capacity RVT ,d of the composite beam (index VT ) should be calculated. The value is to be compared to the design value of the actions EVT ,d

The safety factor γVT results from the required partial safety factor for resistance

Usually, timber/concrete-composite structures are loaded in bending. In this case, the proof according to equation (5) yields the proof of the design value of the maximum bending moment

where My,max,d is the design value of the ultimate moment which the composite beam can carry. As shown in Schmidt et al. 2002, 2003, the loading moment My is in equilibrium with the internal moments My,e and My,t and the normal forces couple N ∗

where a is the moment arm of internal forces

28

The bending stiffness and, consequently, the bending moment My,e of concrete plate yields only 10% of the timber beam stiffness resp. moment My,t . Hence, My,e will be neglected (My,e ≡ 0). A ductile behavior of the composite structure is guaranteed, if the joints reach their ultimate capacity before timber and concrete. This situation leads to the requirement

In the middle of span, the normal force N ∗ results from the horizontal equilibrium of forces as sum of the connection forces Ti at the span section 0 ≤ x ≤ L/2 with n joints

The saturation rate for connection kV according to equation (4) is depending on the material parameters and geometry. The factor is circa 0,9 for the investigated timber/concrete-composite beams with 0,07 kN/mm ≤ rv,k ≤ 0,2 kN/mm without gradation of joints (type A) under a constant line load q. By gradation of joints, the value of kV increases close to 1,0. But this value should not be used for design. We propose to set kV ≤ 0,9 for all structures. The influence of the gradation of joints is considered by a geometrical factor kg . kg results form the integral of the curves according to Figure 4 normalized by the span L

kg is 0,75 for type B, kg = 0,66 for type C and kg = 0,5 for type D. Consequently, the design value of the normal force at ultimate state Nu,d results in the sum of the load capacity Rv,d of the n joints along half beam multiplied with the factors kv and kg

The design value of stress in centre of timber at the ultimate state results in

A factor k is determined with this factor and the design value of timber tension strength ft,t,d using the proof of timber stresses according to line one in equation (3)

The factor k characterizes the normalized residual bending capacity. The design value of the maximum bending moment My,t,max,d , which the pure timber beam can carry, results from the moment of resistance Wt , k and the bending strength ft,m,d

29

The design value of the maximum bending moment My,max,d , which the composite beam can carry, comprises the “residual”-bending moment of timber beam My,t,max,d and the normal force couple Nu,d multiplied with the arm of internal forces a

Thereafter, the design value of a composite beam is at least equal to the design value of a pure timber beam. This result is an advantage of the proposal in contrast to the linear proposal according to equation (3) (compare Figures 5, 6, 7 and 8). Values of k > 0 have to be assumed for timber. This requirement is fulfilled for all structures excluding connection type A with rv,k > 0,2 kN/mm. If the requirement k > 0 is not fulfilled, the new design proposal may not be used because the assumed failure mechanism will not occur. In case k < 0, tension failure of timber or crushing of concrete occurs. The failure modes are brittle and the joints can not reach their ultimate load capacity. Furthermore, the proof of shear stresses in timber results in

Vd is the design value of maximum shear force at the support and fv,d the shear strength of timber.

5 VERIFICATION The load-bearing capacity of composite beams computed on the basis of FE-models and reduced by the global partial safety factor γVT = 1,86, depending on the connection, is shown in Figure 10.

Figure 10. Load carrying capacity, which is reduced by a global partial safety factor γVT (black thin), compared to the design values according to the new design proposal (black bold) and to the design values according to E DIN 1052 (grey).

30

These values are compared to the design data according to the new proposal and to the design values of E DIN 1052. The design values of the new proposal are higher than the results of E DIN 1052. Furthermore, the resulting safety factor is higher than the postulated partial safety zone (γVT = 1,86) derived from the modification factor kmod and the partial safety factor γt (cp. equation (6)). Deviations of the curves “global partial safety factor” and “new design proposal” result from the constantly assumed saturation factor of joints kV = 0,9 and the neglected bending capacity of the concrete plate. Finally, the new design proposal allows an economical design of timber/concrete-composite beams and ensures a sufficient and constant safety level with respect to ultimate state also for structures with gradations of joints. Prerequisite of the new design proposal is the assumed failure mechanism. The joints must reach their capacity before timber and concrete. Concrete crushing or timber cracking has to be excluded to ensure ductile behavior with a good saturation of joint capacity. Therefore, it is subsequently researched which type of structures ensure this behavior. The parameters of the composite beam are manifold like geometry parameters (L, bt , ht , be , he , ), connection type (stiffness, capacity, gradation) and material parameters (strength class of timber and concrete, reinforcement). The number of possible combinations is nearly unlimited. Often, timber/concrete-composite beams are used to reinforce existing timber ceilings because the serviceability is not given (for example large deflection due to creep or insufficient stiffness). But the proof of the load-bearing capacity for loads of self weight and service load of dwelling rooms is mostly fulfilled. To reduce the number of possible combinations, it is assumed that the timber beams are designed this way, that a constant area load of qd = 6,0 kN/m2 yields saturation of bending stress for pure timber beams of σt,m,d /ft,m,d = 1 (σt,t,d = 0). Furthermore, strength classes for timber C24 according to DIN EN 338 and for concrete C25/30 according to DIN 1045 are assumed. The area of crosssection of concrete should be equal or larger than the area of timber. Further, a minimum thickness of 40 mm for the concrete plate is ensured. The following parameters are varied: • • • • • •

span 3,0 m ≤ L ≤ 8,0 m, ratio tr = bt /ht , $0,4 ≤ tr ≤ 0,9, horizontal spacing of beams 0,6 m ≤ be ≤ 1,2 m, initial stiffness of connection 0 ≤ c ≤ 750 N/mm2 , applied load capacity of joints 0 ≤ rv,k ≤ 350 N/mm, connection type expressed through the ratio kz of applied load capacity and stiffness of connection kz = rv,k /c, 0,4 mm ≤ kz ≤ 0,8 mm.

A linear stress analysis yields results, where the connection and the timber beam fail at the same load level. The limits were checked by using the non-linear FE-model. The goal of research is to ensure the safety level between ultimate state and design state according to the new design proposal. A huge number of FE-simulations results in the criterion

which ensures a sufficient safety level according to E DIN 1052, where X1 and X2 will determined with Figures 11 and 12. A linear interpolation of the obtained values is allowed. A distinction of different loading times is not necessary. The investigation of the creep behavior (see Schmidt et al. 2003b) shows an increase of the saturation of joints with an increase of time. At the time t = ∞, the connection is more loaded than at time t = 0. The assumed failure mode of composite beams ( joints must fail at first) is ensured. The influence of shrinkage is investigated in present research. 31

Figure 11. Factor X1 for the application of the new design proposal.

Figure 12. Factor X2 for the application of the new design proposal.

6 CONCLUSION The design of timber/concrete composite beams according to DIN 1052 and EC 5, respectively, is based on linearly calculated internal forces (approximated by so-called γ -procedure) and yields un-economical structures, because the safety level to ultimate state is much larger as usual in timber engineering. Above all, composite beams with gradation of joints have the highest safety factor. Although, the behavior is ductile and failure will be prefigured with large deflections. In opposite, 32

composite beams with a strong connection, which fail brittle, have the smallest safety factor. The design approach should be improved. As result of the investigations, a new design proposal is presented, which takes non-linearities into account and guarantees a constant safety-zone between failure load and working load. This proposal permits an economical design of timber/concrete composite beams. The applicability is checked and the limits of the new design proposal are presented. In practice, a lot of cases of timber/concrete composite beams and ceilings could be designed using the new proposal. REFERENCES Blass, H.-J. & Ehlbeck, J. & Schlager, M. 1995. Trag- und Verformungsverhalten von Holz-BetonVerbundkonstruktionen. Fraunhofer IRB-Verlag DIN EN 338:2003-09. Bauholz für tragende Zwecke – Festigkeitsklassen DIN V ENV 1994-1-1: 1994-02. Bemessung und Konstruktion von Verbundtragwerken aus Stahl und Beton DIN V ENV 1995-1-1. Entwurf, Berechnung und Bemessung von Holzbauwerken – Teil 1: Allgemeine Bemessungsregeln, Bemessungsregeln für den Hochbau DIN 18800-5: 1999-01: Stahlbauten – Teil 5. Verbundtragwerke aus Stahl und Beton; Bemessung und Konstruktion DIN 1045-1: 2002-07. Tragwerke aus Beton, Stahlbeton und Spannbeton – Teil 1: Bemessung und Konstruktion E DIN 1052: 2000-05. Entwurf, Berechnung und Bemessung von Holzbauwerken Kersken-Bradley, M. 1992. Unempfindliche Tragwerke – Entwurf und Konstruktion, Bauingenieur 67: 1–5 Schmidt, J. & Schneider, W. & Thiele, R. 2002a. Tragverhalten von Holzverbundbalken. Bautechnik 79: 727–736 Schmidt, J. & Schneider, W. & Thiele, R. 2002b. Ermittlung des Tragverhaltens von Holz/EstrichVerbindungen, Leipzig Annual Civil Engineering Report 7: 341–358 Schmidt, J. & Schneider, W. & Thiele, R. 2003a. Zur Bemessung von Holzverbundbalken. Bautechnik 80: 302–309 Schmidt, J. & Schneider, W. & Thiele, R. 2003b. Zum Kriechen von Holz/Beton-Verbundkonstruktionen. Beton- und Stahlbetonbau 98: 399–407 Schmidt, J. & Kaliske, M. & Schneider, W. & Thiele, R. 2004. Bemessungsvorschlag für Holz/BetonVerbundbalken unter Beachtung abgestufter Verbindungsmittelabstände. Bautechnik 81: 172–179

33

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Renovation and modernization of historical buildings – structural problems ´ e
czka A. Kozłowski & L. Sl Rzeszów University of Technology, Rzeszów, Poland

ABSTRACT: After the World War II many manor-houses and other historical buildings in Poland were being devastated, mainly due to lack of current maintenance and long term carelessness. Although they were a part of national heritage, the reasons of their devastation had often also political context. Nowadays, many of these buildings are being rebuilt to restore their natural beauty and function. The paper presents a few examples of the structural renovation and modernization of old buildings, devastated by long term carelessness. The problems of repairing and strengthening as well the adaptation to modern technical and architectural requirements are described. Methods of strengthening of foundations, walls, floors and roof structures have also been discussed. KEYWORDS: strengthening, underpinning, segmental barrel vaults, wooden floors, composite wooden–concrete floor. 1 INTRODUCTION It is well known that structures and buildings are subjected to deterioration, during their life, which comes from natural process of ageing, climatic actions and often due to lack of current maintenance. So, many old buildings need to be renovated and modernized due to their insufficient technical state and/or new predicted functions. Most of these buildings from constructional point of view are masonry structures with wooden or vault floors and wooden roof structure. Some of them after long time period of working are in such technical condition, that from economical point of view the cheapest way is to destroy them and build new one. But additional to social and economic value of existing buildings, it is their historical value. Even that they are not monumental and have no pure historical importance, they are reflection of history and part of our national heritage, so should be architecturally restored to new functional quality compatible with its historical period and structurally strengthened. Today knowledge of structural behavior, using modern materials and mixed building technology allow making structural renovation and modernization of old buildings, especially with using steelworks (Mazzolani, 2001). But every restoration of old building is very difficult engineering task and must be made with attention and care. Each case should be studied individually according to local condition, architectural and conservation protection restriction and also structural requirements. The paper presents examples of the structural renovation and modernization of buildings erected in nineteenth century, devastated by long term carelessness. The problems of repairing and strengthening as well the adaptation to modern technical and architectural requirements are described. Methods of strengthening of foundations, walls, floors and roof structures have also been discussed. ´ ˛SKI 2 MANOR HOUSE IN SIERAKÓW SLA 2.1 Technical condition of building before the restoration The dimensions of the manor house are 15.2 × 47.4 m in horizontal projection. It has basement on the whole area and it consists of three parts, which were made in different periods of time. The 35

´ ˛ski – front view. Figure 1. Manor house in Sieraków Sla

Figure 2. Basement floor plan.

oldest is central part of the building. Now it is one-storey with gallery at the same height as second levels in lateral parts of building (Fig. 1). The western part, two-storey was added probably in the end of nineteenth century and the eastern part, also two-storey, was built at the beginning of twentieth century, (Figs 2–3). The foundation of building was made of layer of stones with sands, which depth was equal about 0.25 m. Basement walls have no vertical nor horizontal damp insulation with the exception of walls in eastern part, which is the newest. All walls were made of common bricks. The thickness of walls is equal 0.3 ÷ 1.0 m, in accordance to level of building. Floor above the basement in western part was made as segmental barrel vaults with thickness 0.12 m, which are supported by steel beams made of railway rails 134/33 with distance between them equal to 1.5 ÷ 2.0 m. Floor above the basement in eastern part was made as reinforced-brick plate with steel beams I280 or I210, with distance 1.0 ÷ 1.3 m. Over the plate was the layer of backfill, timber and woodblock floor. The structural condition of floors in both, eastern and western, part was quite satisfactory. Floor above the basement in central part was made as barrel vaults with spans 2.5 ÷ 5.2 m, with brick thickness 0.15 m. Concrete plate with cracks also existed over barrel vaults, supported on sands and slag filling. The thickness of the filling in the place of keystones is equal to 0.1÷0.4 m. The barrel vaults were cracked, part of them was in emergency conditions and filling was moisturized. 36

Figure 3. First floor plan.

Figure 4. Underpinning and strengthening of foundations.

After the Second World War barrel vaults were partly rebuild into a beam-framed floor with steel beams and prefabricated reinforced concrete plates and additional 0.12 m thickness, concrete layer (Fig. 2). The other floors, above first and second storey were wooden beam floors, with 0.2 × 0.3 m sizes of beams, with 0.8 ÷ 1.0 m spacing. In lateral parts beams were deteriorated, in central part the gallery was completely destroyed (Fig. 3). Roof structure was wooden, mansard type in lateral sides and double slope in central part. The structural condition of central part of roof structure was emergency. 2.2 Methods of structural restoration It was assumed that rebuilding and restoration should not change the architecture of building ´ e
czka, 2000). (Kozłowski & Sl In connection with necessity of lowering the basement level and uncovering the existing foundations, there was proposed removing stones footing and underpinning. Footings should also be enlarged by reinforce concrete set-off, in order to provide for additional loadings, (Fig. 4). The concrete set-offs should be connected with brick wall by means of toothings and additional steel reinforcing rods. The underpinning works were proposed to be executed in short sectors; the wall over removed part of foundation acted then as arch (Masłowski & Spizewska, 2000). Reinforced concrete shell, 0.1 m thick, was designed over the barrel vaults to strengthen them, (Fig. 5), in central part of building. Before that, deteriorated areas were re-bricklaying to remove 37

existing cracks. Both the barrel vaults in emergency condition and a beam-framed floor within central part of the building should be demolished, so that the new barrel vaults could be rebuild, according to original shape of building. Existing construction elements in floors over basement in lateral parts of building were not changed, only filling and finishing layers were replaced, (Fig. 6). Strengthening of steel elements (beams) by welding was not applied because of chemical composition of old steel. It was decided that all wooden floors had to be changed, according to considerable degradation of wooden beams, and to necessity to assure of structural integrity under fire conditions. Wooden beams were replaced by steel beams (IPE 220) and reinforced-tile arch floor between them (Ackermann type), (Fig. 7). In order to provide lateral support of walls, each wooden beam was replaced by steel one by turns, and next a part of floor between them was concreting. In places where openings in existing walls were design by architects due to new functional requirements, horizontal steel profiles were inserted above them. In western part new steel columns were also designed into walls, to support large span horizontal elements. Openings in basement walls were strengthen by steel frames inserted into the openings, due to soil pressure.

Figure 5. Strengthening of barrel vaults.

Figure 6. Rehabilitation of segmental barrel vaults.

Figure 7. Reconstruction of wooden floors.

38

3 TWO-STOREY RESIDENTIAL BUILDING IN RZESZÓW 3.1 Description of the structure Building of the dimensions 21,5 × 14,2 was built in the end of the nineteenth century as a residential (Fig. 8). It is two storey, three bays building of traditional (masonry) structure. Floors above the basement were made as six brick segmental barrel vaults supported on the transverse walls. Floors above the ground storey were constructed as concrete slabs supported on the lower flanges of the steel beams. Very big problem was the identification of these beams profiles, because floors were built before the First World War using Austrian profiles – sectional properties were found in the archival tables. Floors above the second storey were made as timber structure which main structural element was wooden beams. The roof made of wooden trusses, wooden boards and steel sheeting covered the building. 3.2 Technical condition of the structure After one hundred years of exploitation this building was in the rather bad technical condition, mainly because of lack of current and general renovation. It was sold to computer company with the destination as the firm main head office and shops. To adapt to new function, many structural and architectural changes had to be introduced, like demolition of few internal walls, adding internal lift, adaptation of the attic to office rooms (Fig. 9). Investigation of the technical state (Woli´nski & Kozłowski, 1994) made before modernization design gave the results: – technical state of the main building structural walls and foundations were sufficient from the safety point of view. No crack and other significant destruction was observed, – wooden trusses of the roof structure had to be reinforced because of additional loading coming from adaptation of attic to office rooms (thermal insulation, gypsum plaster). It was done by increasing section of all trusses elements by connecting new wooden boards by nails, – allowable loading of the floors above cellar was not sufficient for the new predicted loading coming from changing function from residential to shops. This floor was strengthened by adding above existing vaults new concrete continuous concrete slab, supported on the transverse walls,

Figure 8. Façade of the building.

39

Figure 9. Example of ground floor plans before and after restoration.

– floors above the ground storey had sufficient safety reserve for new loading, after removing heavy layer of brick breakage placed on the concrete slab and applying light kermesite (Fig. 10), – wooden floor above second storey had not sufficient stiffness and should be replaced. It was prepared a few proposals of the new structure of this floor and they are discussed beneath. 3.3 Proposals for the reconstruction of wooden floor Coffered ceiling above the second storey was in the good technical state and it was decided to design new floor structure without replacing it. This restriction created special requirements for the design. Three following solutions were proposed. 3.3.1 Composite wooden–concrete structure Such solution can be applied only when the wooden beam and their support on the walls are in the good state. In the analyzed case this requirement was fulfilled. Structure of the floor is presented in Figure 11. 40

Figure 10. Changes in the floor above ground storey.

Figure 11. Composite wooden–concrete floor.

Figure 12. Shear connections in the composite wooden–concrete floor. 1 – steel nails, 2 – steel flat bars, 3 – undercuts.

The floor structure consists of: – concrete slab of the thickness 50 to 60 mm with transverse reinforcement made of 6 mm bars, – timber beams to which falswork of concrete slab is connected, – shear connectors. As shear connectors the following types can be used (Fig. 12): – steel nails of the diameter 4.5 to 5.5 mm assembled to upper face of the timber beam of amount 2 to 6 per each 100 mm beam length, – steel nails plus additional steel flat bars 40 × 2 of width equal to beam width, assembled in the 2 mm slots in the upper side of the beam, – steel nails and additional notches (cuts) on the upper surface of the wooden beam. These notches of the depth 20 mm and length 100 mm are spaced every 200 mm. This solution is especially effective, cheap and also increases fire resistance of the floor. 41

Figure 13. Independent steel floor.

Figure 14. New reinforced-brick arched floor.

3.3.2 Independent steel floor In this solution (Fig. 13) existing wooden beams and ceiling are left. Between timber beams, with the same spacing, steel I-beams are assembled. On the upper flange of the steel beams, precast concrete slabs are placed. 3.3.3 Reinforced-brick arched floor Popular ceramic floor, like Ackermann, can be used in this solution, (Fig. 14). Concreting of the new floor is predicted in two stages: first stage covers part between wooden beams, second, made after first stage concrete reached sufficient resistance, is made after removing wooden beams. Ceramic floor units are placed on the existing ceiling. 4 CONCLUSIONS After political and economic changes in Poland in 1990-s many activities of renovation and modernization of old building are in progress. The same activities are undertaken from years by all countries in Europe (Verhoef, 1999). Presented in the paper examples of strengthening of foundation, walls, vaults, floors and roof structures of old buildings are well known in design practice but are still big challenge for structural engineers. Most of old buildings during their life-time were rebuild and often essential structural changes were introduced. This is why, all such cases should be carefully studied and restoration works must be done with attention and care. One of the most effective way to strengthening of existing old steel and wooden floors is to change them to composite structures, by adding concrete slab connected by shear connectors. In this way also structural integrity under fire conditions is improved. 42

REFERENCES ´ e
czka, L. 2000. Investigation of the technical state of the manor house in Sieraków Sl ´ a
ski. Kozłowski, A., Sl Technical report (in Polish). Rzeszów. Masłowski, E., Spizewska, D. 2000. Strengthening of building structures (in Polish), Warsaw: Arkady. Mazzolani, F. M. 2001. Steelworks in restoration and consolidation. In Antonio Lamas et al (ed.), Enconto de Construcao Metalica e Mista 3. Proc. Symp. Universidade de Aveiro, Aveiro, 6–7 December 2001. Verhoef, L. G. W. (ed.) 1999. Proceedings of the International Congress on Urban Heritage and Building Maintenance “Problems and Possibilities”. Delft University of Technology, October 26th. Woli´nski, Sz., Kozłowski, A. 1994. Investigation of the technical state of the residential building in Rzeszów. Technical report (in Polish). Rzeszów.

43

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Behavior of masonry members confined with steel tying elements A. Mandara Second University of Naples – Italy

F.M. Mazzolani University of Naples Federico II – Italy

ABSTRACT: The paper summarizes some recent developments achieved by authors on the strengthening of masonry structures. In particular, the behavior of masonry elements confined with steel tied plates and subjected to axial compression is described herein. Relevant collapse mechanisms and application options are also shown and discussed. A purposely developed theoretical procedure, calibrated against the results of a non-linear F.E.M. numerical analysis, is proposed for the evaluation of the effect of confinement in horizontal directions, accounting for the inelastic behavior of both masonry and steel. Because of the reduced number of factors involved, the proposed model can be considered as an useful tool for practical calculations.

1 INTRODUCTION Confinement techniques for increasing the ultimate loading capacity or for recovering existing damage in masonry or stone elements are a millenary practice, widely applied still today. The effectiveness of confinement relies on the application of a compressive action in one or more directions transverse to that of the applied load, so to achieve conditions of multi-axial compressive stress. A quite rational structural system is obtained, in which materials are exploited in the most rational and effective way. In addition, when externally fastened plates are used, the intervention can be arranged in such a way to be easily controlled or removed, if necessary. The main result is a remarkable increase of the member compressive strength compared to the case of uniaxial stress, with a corresponding improvement of ductility properties at failure (Mandara 2002, Mazzolani 1996, 2000, 2002). Transverse confinement of masonry members can be applied in several ways, the most common practice being the use of metal elements working in tension, such as tie-bars or tie-beams, fastened to the masonry by means of contrasting end-plates (Fig. 1). Most adopted material in such interventions is steel, both mild, low carbon and high strength grade, but in the latest years innovative materials such as titanium alloys and Shape Memory Alloys have started to be applied with interesting results (Mazzolani & Mandara 2002). Confinement interventions can be also applied to a larger extent, by encircling the whole building with a suitable system of tensioned members (Mandara 2002, Mazzolani 2002). In spite of its effectiveness, confinement practice is rather surprisingly not yet adequately supported neither by a convenient theoretical assessment nor by practical procedures for the prediction of the load bearing capacity of confined masonry members. This paper illustrates the main steps of a comprehensive research activity which led to the formulation of a theoretical procedure for the prediction of the effect of confinement. A refined F.E.M. model running on the ABAQUS non linear code is used for calibration purposes. As a test benchmark, some experimental results carried out on masonry specimens coming from the collapsed Civic Tower of Pavia, Italy (Ballio & Calvi 1993) have been used for the set of the numerical model. This simulated analysis allowed to emphasize the main behavioral aspects of confinement as well as the relevant collapse types, 45

Figure 1. Confinement of masonry members: constructional options.

namely collapse by yielding of steel bars, collapse by crushing (punching) of the masonry in the confined area and collapse by shear-tension failure of masonry in the unconfined areas. 2 BEHAVIORAL ASPECTS AND THEORETICAL FRAMING 2.1 Collapse modes Possible failure types of confined masonry are by bar yielding (Fig. 2a), by crushing (punching) of the masonry in the confined area (Fig. 2b) and by shear-tension failure of masonry in unconfined areas (Fig. 2c). The main geometrical magnitudes influencing the collapse mechanism are the plateto-bar cross section ratio (Ap /As ) and the ratio between the unconfined wall length and the wall 46

a)

b)

c)

Figure 2. Collapse types of confined masonry: (a) bar yielding; (b) masonry crushing in the confined area; (c) shear-tension of masonry in unconfined areas.

Figure 3. The experimental set-up referred to in Ballio and Calvi (1993).

thickness (i/t). As a rule, the use of small plates gives place to failure by punching of the masonry, whereas a comparatively great distance between plates involves collapse by shear-tension in the unconfined area. Failure by bar yielding, which is the one providing the best performance in terms of ductility, occurs for relatively high values of Ap /As . Results of 10 confining tests on ancient masonry wall specimens coming from the collapsed Civic Tower of Pavia, Italy (Ballio & Calvi 1993) confirm the above indications on collapse types. These specimens were chosen for being very significant, in that they have been purposely conceived in such a way to achieve all possible collapse types (Fig. 2). The corresponding experimental set-up is shown in Figure 3, where 4-plate and 9-plate specimens are illustrated. The research pointed out the influence of confinement on both the ultimate strength and ductility of compressed masonry, showing all the effectiveness of this practice and also giving some hints to obtain a given failure mechanism. 2.2 The numerical F.E.M. model The above referred experimental tests have been considered for the calibration of a purposely conceived non-linear F.E.M. model. This model has then been used for the calibration of the theoretical procedure presented in Mandara and Scognamiglio (2003) and Mandara and Palumbo (2004) and shortly reviewed herein in the next section. The ABAQUS release 6.2.1 (2002) has been used. The F.E.M. model is shown in Figure 4, for both 4-plate and 9-plate specimens. Eight-node 47

Figure 4. The ABAQUS F.E.M. model of specimens tested in Ballio and Calvi (1993).

6

6

σ (Mpa)

5

σ (Mpa)

(i)

(c)

5

(a)

(d)

(e) (g)

4

4

(f) (l)

3

3 (h)

(b) 2

2 1

1

ε 0

ε

0 0

0.005

0.01

0.015

0

0.005

0.01

0.015

0.02

0.025

0.03

0.035

Figure 5. Calibration of numerical F.E.M. model (dotted line) against tests carried out in Ballio and Calvi (1993).

reduced integration C3D8R elements have been used. The standard material model *CONCRETE embedded in ABAQUS has been used for the representation of masonry behavior, because of its good accuracy in the reproduction of the progressive development of cracking and the consequent tension stiffening effect. In this case, the mechanical behavior of masonry is well interpreted, in spite of material inherent anisotropy. The behavior of steel bars has been assumed as elasticperfectly plastic, allowing for a small hardening in order to reduce numerical convergence problems. The contact problems between the steel confining plates and the underlying masonry have been properly taken into account, in order to consider possible slip phenomena. Due to the post-critical softening in the compressive response involved by any collapse mechanism, the *RIKS algorithm implemented in ABAQUS has been used. Also, allowance for second order effects (*NLGEOM) has been made. Material data have been inserted in ABAQUS according to the experimental measurements reported in Ballio and Calvi (1993). As specimens came from ancient, inhomogeneous masonry, a certain scattering of the results was observed. The reproduction of the load–displacement curve, however, is quite satisfying, with a faithful interpretation of the actual collapse mechanism in all cases considered. Comparison between experimental tests and corresponding simulation analyses is shown in Figure 5. Also, the wall deformed shape predicted by ABAQUS for the three collapse mechanisms discussed above fully confirms the experimental evidence (Fig. 6). 48

Collapse by bar yielding

Collapse by masonry punching

Collapse by shear-tension

Figure 6. Contour of horizontal displacement for observed collapse mechanisms.

2.3 The theoretical model A procedure initially set out in Mandara and Mazzolani (1998) has been further refined in Mandara and Scognamiglio (2003) and Mandara and Palumbo (2004), in order to take into account collapse mechanisms other than bar yielding. As the model was purposely conceived for design applications, it has been based on the simplified assumption of homogeneous continuum. For this reason, in case of masonry with complex texture, it requires the preliminary application of suitable homogenization criteria in order to get the equivalent masonry properties (Nemat-Nasser & Hori 1999). The main assumptions are: – isotropic behavior of masonry; – elastic-perfectly plastic behaviour of steel bars; – fully rigid steel confining plates, which results in the confining force to be evenly distributed across the wall side surface (uniform confinement); – pseudo-elastic relationship between the applied stress σm and the confining stresses σc,x and σc,y holding in both elastic and post-elastic range, which permits the use of Navier-like equations written in terms of secant modulus Em,s ; – masonry behavior in compression, both confined and unconfined, described by means of an appropriate nonlinear σ −ε law, with experimentally fitted parameters. Referring to Figure 7, considering the equilibrium equations along the wall transverse directions and expressing the global strain in the masonry in both load (εm ) and transverse direction (εc,x and εc,y ) by means of Navier-like equations written in terms of secant modulus Em,s , the following expression for σm can be obtained (Mandara & Palumbo 2004):

Assuming that both Em,s and the transverse expansion modulus ν (the Poisson’s modulus) are a function of the compressive strain εm , then Equation 1 may be considered as the σ −ε law of the confined masonry. The above equations hold until the stress in the tensioned bars does not exceed the steel yield stress fy . When this occurs, assuming that bars in both directions are yielded, it must 49

t

σm

Ap,x σc,x

σs,xAs,x

σc,x

σs,xAs,x

As,x

σm

Figure 7. The mechanical model of uniformly confined masonry.

result σc,x = − fy As,x Ap,x and σc,y = − fy As,y Ap,y . Equation 1 then becomes:

The above formulation can be easily applied to cases other than bi-directional confinement. For example, in case of walls confined along the transverse direction x, the position εc,y = 0 can be made. Correspondingly, the following equations are obtained (Mandara & Scognamiglio 2003, Mandara & Palumbo 2004):

for steel bars in elastic range, and:

for steel bars in plastic range. Similarly, for columns or males confined in the x direction only, one may assume σc,y = 0 and the corresponding equations for steel bars in elastic range are:

50

whereas for yielded bars they become:

For the application of the procedure, appropriate functions for Em,s and ν have to be assigned (Mandara & Mazzolani 1998). The secant modulus Em,s can be obtained starting from a suitable σ −ε relationship for plain masonry. In this study, a model derived from the Saenz’s model for concrete has been proposed (Sargin 1971). The corresponding expression for Em,s is:

where Em , σm,u and εm,u are the masonry initial elastic modulus, the ultimate compressive stress and the corresponding strain, respectively. Compared to the original formulation of the Saenz’s law, there are some differences in the model presented herein, namely the strength enhancement factor k due to confinement in order to take into account the increase of masonry resistance produced by the combined state of stress and the exponent z introduced instead of a numerical factor equal to 2, in order to obtain a more accurate reproduction of the softening branch of the σ −ε relationship. Both k and z have been found being basically dependent on both mechanical and geometrical properties of the masonry wall. An appropriate expression for k can be put into the form:

α being a numerical coefficient which depends on the masonry features and is to be fitted experimentally. The σc,x /σm and σc,y /σm ratios can be evaluated by suitable algebraic manipulation of equations given above as a function of ν, Em,s , Es , Ap,x Ap,y , As,x and As,y . Some hints for the practical application of the outlined model, requiring a trial-and-error procedure, are given in Mandara and Scognamiglio (2003) and Mandara and Palumbo (2004). Concerning ν, as well known, the meaning of the Poisson’s modulus in masonry is not exactly the same as in an elastic continuum, in particular when the collapse load is approached. This is mostly due to the onset of cracks along the load direction. A direct evaluation of the transverse expansion ratio ν (Faella et al. 1993) leads to a so-called “apparent” Poisson’s modulus whose mechanical meaning is far different from the one of an elastic, isotropic continuum, and whose values can be equal to or higher than 1.5÷2. Such values, clearly incompatible with the physical meaning of ν, cannot be assumed in case of confined masonry, as the development of cracks is counteracted by confining ties. In this case, cracks occur as well, but with a rather different aspect from the case of unconfined masonry. Also, the actual masonry texture, that is the block size and configuration as well as the mortar properties, should be considered in the assumption of the ν(εm ) function. In absence of reliable data on ν under combined stress conditions, assuming that in such a case a reduction of the void volume due to the local crushing of the masonry could take place, it seemed more appropriate to assign a law for ν reaching values not higher than 0.5 (no volume change) in the large displacement range. A mechanical meaning for ν consistent with the theoretical assumptions of the model was given in this way. Assuming that, as experimentally observed, for εm = 0 it should result ν = 0, a possible law for ν can be put into the form:

51

where coefficients a and b have to be fitted in order to adequately reproduce the results coming from either numerical simulation or direct experimentation.

3 CALIBRATION OF THE THEORETICAL MODEL A non-linear F.E.M. model running on the ABAQUS code release 6.2.1 (2002), fitted on the experimental data obtained by Ballio and Calvi (1993), has been used for the calibration of the theoretical model described before. A comprehensive parametric analysis was carried out, which pointed out the effectiveness of confinement on both ultimate strength and ductility of compressed masonry. The analysis has been concerned with the case of both uniformly and partially confined masonry walls, in order to investigate all possible failure modes (Fig. 8). Because of symmetry, the assumption εc,y = 0 was made. Three masonry types have been considered, whose main mechanical parameters are summarized in Table 1, together with relevant values of z, a, b and α, as found from a best fitting procedure of theoretical curves against F.E.M. results. Values of wall thickness equal to 300 and 480 mm have been considered. Different values of As,x , represented through the bar diameter , have been assumed, so as to emphasize the effect of confining steel area. A yielding stress fy = 600 N/mm2 has been considered in the analysis for steel bars. As a result of the analysis, together with the calibration of the theoretical model itself, some useful indications on how to get a given failure mechanism were also achieved, which can be considered as a general guidance for practical applications. A good agreement between the proposed method and the numerical simulation has been found in most cases, in particular in the estimation of ultimate load bearing capacity in case of uniform confinement (Fig. 9). Some minor discrepancies with respect to F.E.M. results only come out

Figure 8. ABAQUS F.E.M. models for uniform confinement (a) and partial confinement (b). Table 1. Global frame of masonry mechanical properties assumed in the parametric analysis and relevant calibration parameters for the material model (z, a, b and α). Masonry type

Em (MPa)

σm,u (MPa)

εm,u

t (mm)

 (mm)

z

a

b

α

1 2 3

3300 2300 660

3.5 2.5 2.5

0.0025 0.0025 0.007

480 300 300

8, 12, 16 8, 10, 12 4, 6, 8

1.5 1.8 2

1 3 3

2.1 2 1.8

0.2 0.5 0.5

52

around the knee point and in the softening branch of the σ −ε curve. This is mostly due to the σ −ε relationship assumed the unconfined material. It can be observed that, owing to the geometrical symmetry (εc,y = 0) of the masonry panel, a certain amount of confinement in the wall plane (y-direction) does exist even without plates in the x-direction. This results in the ultimate load and the corresponding strain of the unconfined wall being higher than those of the plain masonry. This aspect is well caught by the analytical model when the position As,x = 0 is made. In order to account for collapse mechanisms other than yielding of steel bars, partial confinement with uniformly spaced plates has been also considered (Fig. 8b). In such a case, depending on both plate spacing and As,x /Ap,x ratio, collapse by either local crushing or shear-tension may occur. For a given As,x /Ap,x value, the corresponding i/t value has been assumed as relevant parameter to establish whether the wall collapse occurs due to bar yielding, masonry punching or shear-tension failure of masonry between plates. As long as the i/t ratio increases, the corresponding wall collapse load decreases, according to curves shown in Figure 10. In particular diagrams at the right side show the ratio between confined and unconfined strength Rc /Rnc versus the i/t ratio, for As,x /Ap,x = 50, 100 and 400 and masonry type 2. As a rule, for i/t ≥ 1.5, the effect of confinement vanishes completely. Correspondingly, the failure mode moves from bar yielding to punching or shear/tension depending on the value of As,x /Ap,x . As a conclusion, a synopsis view of all possible failure conditions is given in Figure 11, where the relevant collapse mechanisms are also pointed out. An approximate border line between them has been traced, which can give useful indication from the design point of view. In practice, collapse conditions other than that involved by bar yielding should be avoided, as they cause confinement ineffectiveness and/or local crushing of masonry, and hence, a brittle behavior of the wall at collapse. For the purpose of practical calculations, an easy tool for the prediction of collapse load for a given value of i/t ratio is needed. The solution to this problem would require the definition of a very complex mechanical model, taking into account all relevant aspects of the collapse

2.5

2.0 Masonry type 1

σm/σmu

F = 16mm

2.0

F = 8mm F = 6mm F = 4mm

1.0

Unconfined

Unconfined

1.0

Plain masonry

Plain masonry

0.5

Masonry type 3

1.5

F = 12mm F = 8mm

1.5

σm/σmu

0.5 εm/εm,u

0.0 0.0

1.0

2.0

3.0

4.0

5.0

6.0

7.0

εm/εmu 0.0 0.0 8.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

Figure 9. Theoretical versus F.E.M. results (dotted line) for uniformly confined masonry walls. 12 11 10 9 8 7 6 5 4 3 2 1 0

5.0

σm (MPa)

4.5

Ap /As = 50

i = 100mm

Rc /Rnc

4.0 i = 150mm

Masonry type 2

Ap /As = 50

3.5 i = 200mm

3.0 100

i = 250mm i = 300mm i = 500mm

F.E.M. (dotted line) Equation 19 (full line)

2.5 2.0 400 1.5

i = 700mm

1.0 δ(mm) 0

1

2

3

4

5

6

7

8

9

10

0.5

i/t

0.0 0.0

0.5

1.0

1.5

2.0

2.5

Figure 10. Typical force–displacement relationship in case of partial confinement (left) and the corresponding influence of i/t ratio on the resistance of partially confined walls (right).

53

5 COLLAPSE BY BAR YIELDING

i/t

COLLAPSE MECHANISM BY SHEAR-TENSION

4

COLLAPSE MECHANISM BY PUNCHING

3

2

1 (Ap/As)*(σmu/fy) 0 0

1

2

3

4

Figure 11. Synopsis view of possible collapse modes.

mechanism. Such difficulty arises due to the fact that in case of bar yielding or local punching the wall collapse occurs due to masonry crushing between confining plates, whereas in case of shear-tension failure it is predominantly a matter of local instability of outer masonry leaves. Appropriate interaction models should be used to represent the actual collapse phenomenology of confined masonry. A simplified procedure has been proposed in Mandara and Palumbo (2004) for engineering purposes, based on the combined use of both F.E.M. results and theoretical prediction for uniformly confined masonry. Results referring to the theoretical model can be used for i/t = 0, whereas F.E.M. data can be exploited to reproduce the variation of the wall strength as long as the i/t ratio increases. Curves in Figure 10 can be used to this purpose, in order to fit a simple relationship relating Rc /Rnc ratio to i/t ratio. The following equation is proposed:

whose results are plotted in Figure 10 as well. Such equation can be used to predict the resistance of confined masonry for a given value of the i/t ratio when the resistance ratio (Rc /Rnc )0 of uniformly confined (i/t = 0) to unconfined masonry is known. Information on the relevant collapse mechanism can then be obtained from Figure 11. 4 CONCLUSIVE REMARKS This paper reports some of the results of a research activity mainly devoted to the use of steel elements in the strengthening of masonry structures. More in detail, the behavior of masonry members confined by steel ties has been analyzed herein. The problem has been faced starting from experimental data available in literature. In this framework, the possible collapse mechanisms have been highlighted, pointing out the most significant implications from the point of view of practical applications. In a second step, an ad-hoc theoretical procedure has been presented, which is able to predict the response of uniformly confined masonry members up to failure. The procedure has been calibrated on the basis of results coming from a non-linear F.E.M. simulation carried out by means of the ABAQUS code. The analysis has led to a thorough understanding of the global behavior of masonry in such loading conditions, highlighting all significant aspects relevant to each collapse mechanisms. With respect to existing models, mostly concerned with confined 54

concrete in compression, the theoretical procedure discussed herein is based on a reduced number of parameters, to be fitted on the basis of either experimentation or numerical simulation. With a suitable choice of these factors, the model exhibits a satisfying degree of accuracy, while remaining comparatively easy to apply in practical cases. In the end, collapse mechanisms other than that involved by bar yielding have been investigated by means of numerical simulation, with the purpose of defining the range of geometrical and mechanical properties to be adopted in practice for achieving collapse by bar yielding. To this purpose, F.E.M. results have been further exploited in order to set up a simplified procedure for the evaluation of the member load bearing capacity as a function of the main geometrical and mechanical parameters of the confined system. Results achieved, as well as ease of application, confirm the proposed procedure to be suitable to engineering practice.

ACKNOWLEDGEMENTS This research started in 1997 within the project Metal Systems for the Consolidation of Structures (resp. F.M. Mazzolani), in the framework of Progetto Finalizzato Beni Culturali issued by the Italian National Research Council (C.N.R.). The ongoing development is now framed within the project Innovative Metal Materials in the Seismic Strengthening of Masonry Structures (resp. A. Mandara), which is a part of the project Diagnostic and Safeguard of Architectonic Works, sponsored by Italian National Research Council (C.N.R.) with funds granted by Italian Ministry of University and Research (MIUR) (L. 449/97). REFERENCES ABAQUS, 2002. User Manual, Version 6.2, Hibbitt, Karlsson & Sorensen Inc. ABAQUS, 2002. Theory Manual, Version 6.2, Hibbitt, Karlsson & Sorensen Inc. Ballio G. & Calvi G.M. 1993. Strengthening of masonry structures by lateral confinement, Proc. of IABSE Symposium. Structural Preservation of the Architectural Heritage, Rome. Faella G., Manfredi G. & Realfonzo R. 1993. Stress-strain relationships for tuff masonry: experimental results and analytical formulations. Masonry International, Vol. 7, No. 2. Mandara A. 2002. Strengthening techniques for buildings, in Refurbishment of Buildings and Bridges (F.M. Mazzolani & M. Ivanyi Eds), Springer Verlag, Wien-New York. Mandara A. & Mazzolani F.M. 1998. Confining of masonry walls with steel elements, Proc. of Int. IABSE Conf. Save Buildings in Central and Eastern Europe, Berlin. Mandara A. & Palumbo G. 2004. Confined masonry members: a method for predicting compressive behaviour up to failure. Proc. of the International Seminar on Structural Analysis of Historical Constructions, Padova, Italy. Mandara A. & Scognamiglio D. 2003. Prediction of collapse behavior of confined masonry members with ABAQUS, Proc. of the ABAQUS Users’ Conference, Munich. Mazzolani F.M. 1996. Strengthening options in rehabilitation by means of steelwork, Proc. of SSRC International Colloquium on Structural Stability, Rio de Janeiro. Mazzolani F.M. 2000. Steel in structural rehabilitation (keynote lecture). Proc. Of ITEA Symposium, La Coruña, Spain. Mazzolani F.M. 2002. Principles and design criteria for consolidation and rehabilitation, in Refurbishment of Buildings and Bridges (F.M. Mazzolani & M. Ivanyi Eds), Springer Verlag, Wien-New York. Mazzolani F.M. & Mandara A. 2002. Modern trends in the use of special metals for the improvement of historical and monumental structures, Engineering Structures 24: 843–856, Elsevier. Nemat-Nasser S. & Hori M. 1999. Micromechanics: overall properties of heterogeneous materials, 2nd Ed., Elsevier, Amsterdam. Sargin M. 1971. Stress-strain relationship for concrete and analysis of structural concrete sections, Study n. 4, Solid Mechanics Division, University of Waterloo, Canada.

55

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Retrofitting of RC structures taking as an example multi apartment gallery-access block of flats building K. Wróbel & W. Kubiszyn Rzeszów University of Technology, Rzeszów, Poland

ABSTRACT: The technical state of object is product resultant of conditioning of all winding phases of investment process and way his exploitation. It is the object of article about twelve aboveground storeys building situated on one of Rzeszów housing estates. It is gallery-access block of flats (with communication from external terrace within the same floor). Functional solutions in which was adapted with Mediterranean climate. In article the causes of poor technical condition of object were analyzed and introduced. Proposed the ways on restoration his usability. These proposed solutions were realized during the complex repair of building. KEYWORDS:

Reinforced concrete structures, apartment building, durability, repair.

1 INTRODUCTION It the durability of building objects depends on: – the designer that is from practically the received solutions material – constructional and functional, the designer has to possess the skill of expectation of factors what on building will affect and their results, – the performer – from quality of applied materials, selection of suitable technology as well as care of realization building works, – the user of object which is duty specific exploitation – peaceable from destination the object, current control of technical state of object as well as the eliminating the nascent in time of current exploitation of construction damages peaceably with established level the maintenance. The influence of every factors of it units was showed for example one of building structures in Rzeszów. 2 DESCRIPTION OF STRUCTURES Analyzed building is multi flat building, about twelve above-ground storeys entirely. It part of basement create clearance height about 3 m. In this part building be support on columns. First storey technical plane be built-up entirely. It over technical plane is X habitable storeys. Staircase assures internal transport and two passenger lift. Building this is gallery-access block of flat it – possesses to unit of flats entries from external galleries directly. Building had been exploited since 1972 year without repair. 2.1 Foundation of building The building is situated on the reinforced concrete case foundation. The thickness of foundation internal walls and plate bottom – 50 cm, thickness of external walls carries out – 40 cm. Floor over cellars – RC two direction reinforced plate about thickness 15 cm. 57

2.2 Overhead part Above-ground part is skeleton construction consisting of: – in level of ground floor from monolithic spaced what 4.2 m in transverse direction columns and what ∼6.0 m in command the longitudinal building. The reinforced concrete monolithic beams on columns be base, and on them beam framed floor DZ-3. – there in level of technical plane from monolithically made RC 40 cm thickness longitudinal walls and transverse thicknesses from thickened pilasters about dimensions of cross section how the columns of ground floor. – on habitable storeys – from of the precast reinforced concrete the “H” type frames put in command transverse building what 6.0 m. The beams have cantilevers the hanks over about outreaches 1.25 m pose the external grain of columns. The floors are precast channel floor plates. Three the internal reinforced concrete walls assure the stiffness and stability of building skeleton. The remaining executed from different materials walls are the curtain walls, or retaining the bricklined after build of the reinforced concrete skeleton.

165

12

280

+ 33.60

280

280

+ 30.80

11

280

280

+ 28.00

10

280

280

+ 25.20

9

280

280

+ 22.40

8

280

280

+ 19.60

7

280

280

+ 16.80

6

280

280

+ 14.00

5

280

280

+ 11.20

4

280

280

+ 8.40

3

+ 5.60

310

702

277

2

1

280

390

+ 0.00

0

50

- 2.80 - 3.90

420

420

A

420

B

Figure 1. Structural schema of the building.

58

420

C

3 ESTIMATION OF TECHNICAL STATE OF STRUCTURES The technical state of the building was estimated considering all structural members and units of finish. It below the most important results of this opinion were have presented oneself was. 3.1 Galleries and west façade The general view of façade represents figure 2. – The arrises of load-bearing plate of gallery corroded (fig. 3). On bottom edge of gallery of the highest storey the making up the product of corrosion salt crystallized in the firm of stalactites (fig. 4).

Figure 2. Façades of the building before repair – west and south.

Figure 4. Stalactites on edge of gallery of the highest storey.

Figure 3. View of corroded element of loggia.

59

Figure 5. View of damaged fixing the balustrades.

Figure 6. View of corroded cantilevers of the reinforced concrete galleries.

Figure 7. East façade of the building before repair.

Figure 8. Corrosive damages of loggia plates.

– It the dangerous depending on total crumbling away in places the concrete of balustrades phenomenon was observed was their fixing to the steel posts and deformation of bars the balustrades, the screws for which be established the fixing balustrades (fig. 5). – It the extensive corrosion of reinforced concrete cantilevers was affirmed was supporting in direct neighbourhood galleries their connection with steel post (fig. 6). – It the different type of cracks on curtain wall of building along gallery were observed. 3.2 Loggias and east façade The general view of façade represents figure 7. – Delamination of all beams supporting of loggias plate from termoinsolation was affirmed. – It was observed the intensive and dangerous corrosion of steel and concrete of the reinforced concrete elements of precast balustrades. Figure 8 presents corrosive damages. 3.3 South façade and balconies The cracks of the walls, corrosive damages of balcony plates with different intensification and leakness flashing were observed. Figure 2 presents the façade, figure 9 presents corrosive damages. 3.4 RC columns The corrosion of concrete and steel of columns was affirmed and they lay the columns (fig. 10). As well as very weak or total the lack of adhesiveness of plaster to concrete of columns. 60

Figure 9. Corrosive damages of balcony plates.

Figure 10. View of the corroded reinforced RC columns of ground floor.

Figure 11. Corrosive damage of reinforced concrete cellars elements.

3.5 Cellars – They lay intensive corrosion steel and concrete all of reinforced concrete elements (fig. 11). – The perpendicular damp insulation of external cellars walls is ineffective, or does not it execute her at all. Walls these get soaked. 4 RESULTS OF INVESTIGATIONS It conducted inspections, detailed macroscopic investigations, executed strip mines, undamaged investigations of quality of concrete, stages of neutralization and over year-old observation of building technical state as well as analysis of building design were affirmed was: – The strength of analysed concrete elements on grip definite on the ground the audits answers the classes established in project or is higher. – On the ground the control of stages of neutralization degree of concrete cleading of the reinforced concrete elements affirm, that in places where she was damaged or about too small thickness it does not it make up the protection of reinforcing bars before corrosion. – Particularly disturbing situation stepped out in cellars, where the front of carbonatisation was more deeply than this with catfish of thickness of cleading and reinforcing bars results. This marks that the protective role of concrete in relation to reinforcement underwent fading. In places, where cleading did not it be damaged keeps she protective role in relation to reinforcing bars. 61

The relating the individual elements of building conclusions: 1. The lock the edge the plate of gallery – the balustrade and lack of effective download of storm sewages cause from saline blooms intensive corrosion of this edge. 2. Using two materials about different thermal expansibility co-operating with himself directly ( the screw joint of the reinforced concrete balustrades plates and steel posts, the connection of steel posts from brows of cantilevers of the reinforced concrete frames) the crumbling away in direct neighbourhood of steel units the concrete causes. Damaged places are susceptible on destructive working atmospheric factors which shortens cycle of failure-free running of these elements, the labour intensity of repairing runnings enlarges and the cost of exploitation of object raises the price with the same. 3. Crack on external walls of building have different morphology: – thermal cracks nascent on point of contact of different materials used to bricklaying of walls – elimination of their will happen after termorenovation of object. – cracks from mechanical damages – it is possible to minimize regulating or founding (there where there have not no them) the mechanisms shutting the doors himself. – the cracks on point of contact the precast element – the brick wall these outline – was eliminated was it will not it give. They are then the characteristic outlines for this object type. – the cracks as a result of bad quality of building works outlines – it is possible was across to build of sections of walls as a new or the realization on them the strong cement plaster or shotcrete on steel mesh (after survey of plaster existing). 4. Corrosion of balcony plates be due many years’, unfavorable the influence the atmospheric factors and the lack of proper protection. 5. RC columns of ground floor – well-made, they from cleading of rods about proper thickness do not it wake the restrictions, focus of corrosion formed in places of every executive inaccuracies they – require in frames of repair the liquidation. 6. The cellar – the technical state of structural member is bad. The microclimate of cellar space (the lack of ventilation and moisture penetrating inwards across leak of isolation) acting by many years caused the failure the protective proprieties of concrete in relation to steel and corrosion of both these materials. 5 FINAL RECOMMENDATIONS In aim restorations the usable efficiency of building were have resolved to do on his complex repair. The range of comprehensive repairing was qualified below. 5.1 Galleries and loggias 1. Disassembly of the reinforced concrete balustrades. 2. Modification and repair of steel posts of galleries and their fixing. 3. Repair of the reinforced concrete elements of gallery and loggia with replacement of the old flashings and the exchange of finishing layers. 4. Making of new light steel balustrades from full screens and clearance at pit. 5.2 Balconies Repair of the reinforced concrete plates of balconies with replacement of the old flashings and the exchange of finishing layers. 5.3 Cellars 1. Realization of the perpendicular external isolation of foundation walls. 2. Realization of effective gravitational ventilation. 62

Figure 12. Façades of the building after repair.

3. Repair of corroded reinforced concrete elements (the columns, wall of foundation case, openings lintel). 5.4 Columns of ground floor 1. 2. 3. 4.

Erase of coming off plaster. Repair of corroded of the reinforced concrete columns. Realization of new plasters or surfacing. Superficial protection.

5.5 Curtain walls of building 1. In places, where walls be cracked: – erase plaster, – the current opinion of technical state, – repair according to current defined range. 2. Termorenovation of the external walls.

6 CONCLUSIONS 1. Gallery-access block of flats buildings they are buildings adapted with climate type Mediterranean, completely it checking oneself in weather Polish conditions and creating the numerous exploitational problems. 2. Poor technical condition of object was result of overlapping mistakes and negligences of all winding phases of investment process and lack any the repairs throughout exploitation period. 63

3. Proper selection of way of repair depends from correct recognition of causes which they caused the degradation of object, the lodged level of maintenance and foreseen period of future exploitation. REFERENCES Czarnecki L., Emmon P.H. Repairing and protection of concrete structures, Polski Cement, Kraków, Poland 2002 Wrobel K., Kubiszyn W. Estimation of technical state of multi-flat building near Da˛browki street 21 in Rzeszow, Rzeszów, Poland 2001–2002 Wrobel K., Kubiszyn W. Design of building repair near Da˛browki street 21 in Rzeszow, Rzeszów, Poland 2001–2002

64

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Full refurbishment of an office building in Innsbruck G. Huber Goetzens/Austria

C. Aste aste konstruktion, Innsbruck/Austria

ABSTRACT: For the full refurbishment of the AK office building in the center of Innsbruck some new technologies of MBT have been applied very successfully such as slim floors with semi-continuous beam-to-column moment connections and local load introduction from the slabs into the steel-concrete hollow columns by powder actuated nails. During construction work the offices in the upper stories were normally used. The final result with considerable gain of space and improved architectural appearance without soil settling or cracks worth mentioning was highly appreciated by the client. KEYWORDS: semi-continuous beam-to-column moment connections, local shear introduction into the hollow column by powder-actuated nails and bolts, slim floors.

1 INTRODUCTION The five storey office building of the “Arbeiterkammer” (labour chamber) in Innsbruck dates back to the 19th century. The main construction material was brick with some concrete and steel elements due to previous reconstructions. The unsatisfying situation in view of the entrance location, room arrangements, multiple different floor levels, dark corridors and overall architectural appearance led to the demand of a generous refurbishment of the lower levels – ground floor and cellar – whilst the upper four floors should be continuously used as offices with a lot of client contacts. By lowering the ground level of the cellar for 1,5 m the gained overall room height allowed for the creation of an additional mezzanine. The extension of the cellar walls downwards into the

Figure 1. AK office building before and after refurbishment.

65

Figure 2. Overview of the refurbishment – replacement of brick walls by a column and beam grid.

ground was realized with a compaction grouting system. Very stiff, low slump mortar is injected into the soil with very high pressure – displacing and compacting the existing soil in place. This also ensured the foundation capacity in view of the new and increased building loads without considerable additional soil settling. Beginning in April 2002 the massive brick walls of up to 80 cm thickness where replaced step by step by very slender composite columns and a load transferring beam grid at their top. The load introduction from the concentrated column bases into the existing cellar walls was taken over and spread by a massive concrete basement girder. To fulfill the actual normative requirements especially in view of accidential loads also some remaining floor and wall elements were identified to need strengthening or even had to be replaced. 2 COMPOSITE COLUMNS The existing walls between the level of 0,0 and 6,9 m where replaced by steel-concrete hollow columns with a diameter of 22 or 25 cm. The load introduction at the bottom and at the top was realized with simple top/bottom plates. The introduction of the floor loads docking to these columns takes place in the following way. The slim floor steel beams are supported by steel brackets welded to the column tubes. The further load transfer from the steel tube to the interior chamber concrete is ensured by nails which are placed powder actuated in the shop or on site. As already well-tried at the floor connections of the Millennium Tower in Vienna the Hilti fasteners X-HVN32P10 have been applied. These are nails which are also used for conventional fastening purposes. According to Eurocode 4 shear connectors have to provide sufficient resistance against uplift; in the actual application case of a chambered concrete such uplift is automatically prevented and though the nails can be applied without further measures. Despite of high strength steel material of the nails they proved to behave very ductile due to the chamber effect within the hollow steel section and therefore could be classified as ductile connectors. 66

Figure 3. Column grid replacing the old brick walls.

Ø 4,5

nails Ø 10

32

Figure 4. Powder-actuated fasteners (nails) for shear transfer (foto Millennium Tower).

Figure 5. Replacement of external and internal brick walls.

67

3 COMPOSITE SLIM FLOORS The demand for an additional mezzanine floor led to a very limited construction height for the ground and mezzanine floor of at maximum 17,5 cm in combination with a very localized vertical load transfer from the slabs into the slender column tubes. Additionally the main span between the column axes amounts considerable 7,2 m and furthermore the concreting of these floors should

ground floor

mezzanine

T-shaped steel beam concrete C30/37 Hilti connectors shear reinforcement additional reinforcement end deformation of sheeting re-entrant steel sheeting fire protection

Figure 6. Ground and mezzanine floor with slim floor construction.

Figure 7. Composite slim floors at the ground and mezzanine level.

68

be enabled without temporary supports leading down to the basement, where already screed work should start at this time. From these demands a slim floor construction withT-shaped, cambered steel beams in a transverse distance of 3,4 m proved to be most suitable. Trapezoidal steel sheeting was placed on the lower flanges of these beams and served for an immediate working platform. Hilti shear connectors were fixed in every knuckle of the steel sheeting with powder actuated nails going through the steel sheeting into the beam flange. This provided both – end anchorage of the composite slabs and partial shear transfer in longitudinal direction of the slim floor T-beams resulting in a bi-directional composite action of these slabs. 4 BEAM-TO-COLUMN MOMENT CONNECTIONS For improvement of the ultimate and serviceability limit state (deflections and vibrations) these composite slim floor beams spanning between the composite tube columns should not only be used CONCRETE SLIM FLOOR

MR

COMPOSITE

COMPOSITE SLIM FLOOR

CONCRETE

VR- PUNCHING PROBLEM
R


R

VR by STEELWORK CONNECTION
R

MR

DR - LIMITED COMPRESSIVE RESISTANCE

DR -

AND BRITTLE BEHAVIOR

MR

HIGH COMPRESSIVE RESISTANCE AND DUCTILE BEHAVIOR

DR

Figure 8. Comparison between conventional concrete and innovative composite slim floors.

Figure 9. Semi-continuous moment connection between the slim floors and the column tubes.

69

1

3

2 z 7 8 9

5

4

6

Figure 10. Connection characterization and modeling.

Figure 11. AK office building during and after refurbishment.

single span. Moment resisting and semi-rigid beam-to-column connections at both ends transferred the single span beams into semi-continuous ones. This semi-continuity at the beam ends was already applied very successfully at the Millennium Tower in Vienna with more than 50 storeys. The beam end restraint was realized by a horizontal force couple. The lower compression forces are introduced via the lower beam flange into the column bracket. There the gap due to construction tolerances is closed with shim plates. The upper tension forces are activated by reinforcement loops going around the columns. 70

The resulting additional bending moments in the columns proved to be not so significant than the high normal compression forces. This multiple frame effect between the external columns, the slim floors and the internal columns was also used as a contribution to the overall horizontal building stabilization in view of wind and earthquake loads. After concreting the T-beams – except the lower flanges – are fully integrated into the slab depth. Summarising the advantages were slim floors with a high serviceability quality, a fast construction progress, considerable pre-fabrication standard and a reduced noise disturbance. 5 CONCLUSIONS The complete refurbishment of a 19th century brick-concrete office building in Innsbruck was very successfully realized with new MBT technologies. Compaction grouting to extend the existing foundation walls into the ground, local shear transfer from the steel column tubes into the chamber concrete with powder actuated nails and slim floors with partial shear connection and semi-continuous end moment connections to the columns. REFERENCES Angerer, T., Rubin, D., Taus, M., 1999, Verbundstützen und Querkraftanschlüsse der Verbundflachdecken beim Millennium Tower (Composite columns and vertical support connection of the slim floor beams at the Millennium Tower), Stahlbau 68, Ernst & Sohn, Berlin, p. 641–646. Anderson, D., 1999, Design of Composite Joints for Buildings, ECCS Publication No. 109, ISBN 92-9147000-52, Brussels. Beck, H., 1999, Nailed shear connection in composite tube columns, Conference Report Eurosteel ’99 in Praha, ISBN 80-01-01963-2. Hanswille, G., Beck, H., Neubauer, T., 2001, Design Concept of nailed shear connections in composite tube columns, RILEM Conference Proceedings “Connections between steel and concrete” Stuttgart 2001, ISBN 2-912143-27-6. Huber, G., 1999, Non-linear calculations of composite sections and semi-continuous joints, Doctoral Thesis, Ernst & Sohn, Berlin, ISBN 3-433-01250-4. Huber, G., 2004. Steel-concrete moment connection. COST-C12 – Output of the Co-operative Activities, Balkema. Huber, G., Beck, H., 2004. Shear transfer between steel and concrete within composite tubes. COST-C12 – Output of the Co-operative Activities, Balkema. Huber, G., Michl, T., 1999, Beispiele zur Bemessung von Riegel-Stützen-Verbindungen (Example calculation for a beam-to-column joint), Fachseminar und WorkshopVerbundbau 3, Fachhochschule München/ Munich, Germany. Huber, G., Rubin, D., 1999, Verbundrahmen mit momententragfähigen Knoten beim Millennium Tower (Composite frames with semi-continuous joints at the Millennium Tower), Stahlbau 68, Ernst & Sohn, Berlin, p. 612–622. Müller, G., 1998, Das Momentenrotationsverhalten von Verbundknoten mit Verbundstützen aus Rohrprofilen (The moment-rotation response of composite joints with composite tubular hollow columns), Doctoral Thesis, IStHM, University of Innsbruck, Austria. Taus, M., 1999, Neue Entwicklungen im Stahlverbundbau am Beispiel Millennium Tower Wien und Citibank Duisburg (New developments in composite construction at the Millennium Tower in Vienna and the Citibank in Duisburg), Commemorative publication Prof. Dr. Ferdinand Tschemmernegg, IStHM, University of Innsbruck, Austria, ISBN 3-9501069-0-1. Taus, M., 1999, Verbundkonstruktion beim Millennium Tower – Fertigung, Montage, neue Verbundmittel (Composite construction at the Millennium Tower – production, erection, new shear elements), Stahlbau 68, Ernst & Sohn, Berlin, p. 647–651. Tschemmernegg, F., Beck, H., 1998, Nailed shear connection in composite tube columns, ACI-Paper, Houston Convention. Tschemmernegg, F., 1999, Innsbrucker Mischbautechnologie im Wiener Millennium Tower (Mixed building technology of Innsbruck at the Millennium Tower in Vienna), Stahlbau 68, Ernst & Sohn, Berlin, p. 606–611.

71

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Flexibility of building structures R. Blok & F. van Herwijnen TU/e University of Technology, Eindhoven, The Netherlands

ABSTRACT: Flexibility of Building structures and/or Structural Flexibility is a property of buildings which greatly influences the Service Life of existing buildings as well as the expected Service Lives of new buildings. To specify and quantify Structural Flexibility in more detail and to optimize decisions in Integrated Life Cycle Design (ILCD), a new definition of Structural Flexibility and Structural Adaptability is proposed. From this a method to evaluate Structural Flexibility is derived. Key factors and indicators which influence the Structural Flexibility are given.

1 INTRODUCTION 1.1 Redevelopment of urban areas and buildings Looking back on urban developments at the end of the last century it shows that many cities have redeveloped areas that were degenerating for various reasons. Former industrial areas with old warehouses have been transformed to housing and/or office locations. Many buildings have been demolished, other buildings have been refurbished and given a second functional life. Apparently some buildings were more suitable for redevelopment and refurbishment than others. In many cases of these refurbishment of buildings in the past, the conditions that made these buildings suitable for a changed functional use, were not foreseen in the design process but were favourable more or less by chance. These buildings were not designed to serve functional changes. It is interesting however to look at their inherent qualities because some qualities of the building structure, for example a large floor to ceiling height of the building, make it possible to accommodate new functions with new appropriate services. 1.2 Favourable aspects for elongation of a buildings service life Apparently favourable aspects for a second functional working life are mentioned below: • • • •

The quality of the location of the building The architectural quality and/or historical value of the building The technical quality of the building and its structure The economic aspects: costs of refurbishment against costs of demolishing and rebuilding, and in the end: the resulting net returns • The local urban planning rules and building regulations • The vision of the owner(s) and/or authorities (politics) • The environmental aspects (air pollution, soil condition, etc.) 1.3 Integrated Life Cycle Design (ILCD) In a more and more dynamic society organisations tend to change quite fast. At the same time, because of the rising level of prosperity the demands on the built environment changes. The required area in m2 as well as the required quality level of buildings change. Due to this it seems that more and more relatively young buildings, for example buildings of only twenty to twenty-five years old, become evaded, empty and in fact obsolete (sometimes because of their functional qualities, 73

sometimes because of their location). In terms of investments of money but also in terms of energy, materials, waste production, or rather: Sustainability, this is not a desirable situation. Elongation of the service life of buildings, especially buildings with high investment costs such as high rise buildings, could be achieved if the buildings could easily be adapted to new user-requirements. Therefore it is obvious we need to look at buildings from a broader perspective than just the first user-requirements. Integrated Life Cycle Design aims to optimize the building design by looking at the whole of the service life of the building from the extraction and production of building materials up to the demolition and possible re-use of materials and components. 1.4 The meaning of Flexibility in Integrated Life Cycle Design For a large degree, the lifetime of a building structure no longer depends on its Technical Service Life (TSL). The structure is usually the longest lasting part of a building (its service life can easily stretch beyond fifty or seventy-five years). Fast changing user-demands during this period strongly influence the Functional Working life an therefore greatly influence the resulting life-time of a building and its structure. The Functional Working Life (FWL) therefore is of major importance. This means that an important question to answer in Integrated Life Cycle Design of buildings and in particular of building structures is: “How can we design buildings and building structures, or adapt existing buildings which will continue to meet the (quite likely) fast changing (future) user-demands during the life-time of these buildings?” We can distinguish a number of different approaches to this problem, but one approach is to design buildings which are more Flexible, buildings which can accommodate future changes in use. 2 DEFINING FLEXIBILITY 2.1 A building model with different building layers To improve on Performance Based Design and to improve on Integrated Life Cycle Design (compare, balance and optimise the decisions at the design stage) it is necessary to define and quantify “Flexibility” more accurate. Another reason to define “Flexibility” more accurate is that it gives us a tool to evaluate the building structures of our existing building stock in more detail and look at the possibilities of future refurbishment of these existing buildings. To get a grip on how buildings can be changed or adapted, we need a building model. We can look at what can be changed but also we can try to distinguish what is fixed. Steward Brand [2] distinguished seven building layers: Furniture, Space, Services, Access, Structure, Façade and Location. Bernard Leupen [3] looked at dwellings in a similar way: he defined a house as a space, defined by five different, more or less integrated layers, each with their own level of flexibility: Scenery, Access, Servant elements, Structure and Skin. From this a slightly adapted list of building layers is proposed: • • • • • • • •

Scenery (furniture, interior finishes, ceilings) Space plan (partition walls) Access (stairs, corridors, lifts) Servant elements (building services, pipes, cables and involved spaces) Envelope (façades, base, roof) Compartments (firewalls) Structure (floors, columns, beams, load-bearing walls) Location (building environment)

2.2 A Flexible Building In general, a Flexible Building can be defined as a building with the capacity to accommodate, in a relatively easy way, (future) changes in use. This can be achieved by allowing for “relatively 74

Figure 1. Definition of building layers.

easy” changes to one or more of the following building layers: Scenery/Servant elements/Envelope (Skin)/Access/Structure/(Location). This definition poses the problem of what to regard as, or how to define “relatively easy”. Relatively easy could be defined by the extend of the work necessary for a certain change, for example “two men with a screwdriver and … can do the job”. Another way of defining could be: A change to a certain building layer is “relatively easy” if it can be achieved without the necessity to affect or change other building layers as well. For example: A building with a load-bearing elevation wall, combines the layers of Structure and Envelope. It is not possible to change the Envelope layer without also changing the Structure. Regarding this aspect the building is not flexible. It is possible however, that the same building is Flexible with regard to another building layer, for example the Servant elements or the Scenery (partition walls). In case only one (or may be a very limited number) of building layers is involved in the change, a large part of the building’s functions and activities can stay in place, while the changes are being carried out. Therefore it is not always necessary to close the building while these changes take place. Because Flexibility can involve different building layers it means that the term “Flexible Building” should be looked at in more detail and specified further: For a “Flexible Building” flexibility is probably needed for many building layers. The question becomes: “Which of the building layers can be changed without affecting the others. Which aspects of the building are Flexible, how and where and to which degree are the different layers separated?” 2.3 Definition of Structural Flexibility Structural Flexibility and Structural Adaptability are each defined as qualities, or properties of the building structure. (Note: In IFD building, flexibility is sometimes used as referring to the design stage in which building components can be arranged in different possible compositions.) Looking at the building structure with regard to other building layers, Structural Adaptability can be defined as: The capacity of the building structure to accommodate changes to the structure itself without or with minor consequence to other building layers. (This implies that other building layers can obstruct the structures Adaptability.) From this Structural Flexibility can be defined likewise: Structural Flexibility means: The property of the building structure to accommodate changes in use by providing sufficient space and load-bearing capacity and allowing for changes in one or more other building layers (for example scenery, space plan, servant elements) without the necessity to change the structure itself. This way Adaptability is defined in terms of accommodating changes to the specific layer (structure) itself. Flexibility is defined in terms of accommodating changes to other layers. This implies that in case of sufficient Structural Flexibility the building structure need not to be changed in case of a required change in use. Flexibility should be specified in more detail: With regard to which aspects is the structure flexible? Does the Flexible Structure allow for easy changes to the Space plan (partition walls), the Services, to the Envelope (Façade) …, to one or all of these layers? 75

3 INDICATORS FOR STRUCTURAL FLEXIBILITY 3.1 General approach From the definition of Structural Flexibility it is now possible to look for the main qualities and indicators which influence Structural Flexibility. We can look at the building structure and evaluate its relation with each of the other building layers, basically in three steps: • First the aspect of the relation of the structure with other building layers is examined. Is the structure entwined with other building layers and how are the connections with the other building layers (Step A). • Secondly the structure’s quality of “Providing Sufficient Space” to each of the other building layers is looked at (Step B). • Thirdly the aspect of “Providing Sufficient Load-bearing Capacity” is examined (Step C). A large provided space and bearing capacity together with a high degree of separation from other building layers, will result in a high score on Structural Flexibility. In order to achieve Structural Flexibility the other building layers should not physically be entwined with the building layer Structure. At the same time the structure should accommodate the other building layers as much as possible. A very clear example of this separation (independence) and integration at the same time can be seen with the design of new IFD floor systems [4]. 3.2 Elaboration of the proposed method The proposed steps in paragraph 3.1, General approach are carried out in the research undertaken at the University of Eindhoven. For each of the building layers the relation with the Building Layer Structure is be evaluated and indicators representing the qualities are being sought. The following paragraphs gives an example of these examinations. Further research will result in fine-tuning the various indicators and Flexibility factors. 3.2.1 Example approach Indicators Structural Flexibility with regard to building layers Scenery and Space plan Applying Step A: We need to answer the question “Is the building layer Structure entwined or interwoven with Scenery and Space plan or is the Structure independent of these building layers, and how are the layers connected?” Indicator A1: The degree of contribution of Scenery and Space plan elements to the load bearing function. (Rem. For example in case of sloped floors in theatres to accommodate the spectators, the building layers structure and scenery can be seen as entwined, en therefore less flexible.) Step B. Evaluating whether the structure provides sufficient space to allow for changes to Scenery and Space plan: Indicator B1: The functional free floor to ceiling height (The functional free floor to ceiling height is the part of the structural storey height which is allocated to functional use. The part allocated to services is the free service height, the part allocated to the structure: structural height. (In case of integrated solutions these dimensions can overlap.) Indicator B2: The floor span: Minimum functional column-free widths and areas. (Rem: The Dutch building regulations provide minimum widths, minimum areas and minimum floor to ceiling heights depending on the function of the building. Generally it is clear that the larger the column free space and floor to ceiling heights the larger the Flexibility will be. It is more easy to accommodate changes in use in an oversized building.) Step C. Providing sufficient bearing capacity to allow for changes to Scenery and Space plan: Indicator C1: Allowable life floor loads. (Rem. Building regulations give minimum values for life floor loads depending on the buildings function.) 76

Indicator C2: The amount of allowance in the permanent floor loads allocated to the support of (freely arrangeable) partition walls. (The total bearing capacity of existing structures can be partly allocated to life loads and partly to permanent loads for partitioning and services.) 3.3 Indicators Structural Flexibility with regard to Access, Servant elements and Envelope The process similar to paragraph 3.2 (step A, B and C) has been repeated for the other building layers. This has resulted in indicators which describe the relation of the building structure with each of the other building layers. 4 EVALUATING AND SCORING STRUCTURAL FLEXIBILITY 4.1 Evaluation of different building structures With the identified indicators involved in the Flexibility of building structures it becomes possible to compare and evaluate different building structures. Both existing building structures as well as new designs for building structures can be subject to an investigation. For new buildings the design stage is crucial in incorporating possibilities and preventing obstructions for future changes. The design stage will play the most important role in the determination of the Functional Working Life of the building. 4.2 Flexibility classes The Indicators that have been identified in Structural Flexibility have been looked at in more detail. Different classes of Structural Flexibility from “not flexible” to “extreme flexible”, have been defined. By scoring the indicators and attributing weighing factors the level of Structural Flexibility of a building becomes more comparable and may even be expressed in a single score Flexibility Index. multi criteria charts can visualise the extend of Flexibility of a given building structure. For a general idea some examples of indicators with preliminary values are given (see Table 1). An example of a resulting multi criteria chart is given in Figure 2. The chart shows that the evaluated structure shows Limited Flexibility on Envelope, Not flexible on Servant elements, Average Flexibility on Access and Very Flexible with regard to Scenery/Space plan. Rather than a single score Flexibility Index the multi criteria charts visualizes the structures strong and weak points very clear. Furthermore a single score Flexibility Index will require a thorough search for the implementation of values and weighing factors (perhaps based on expert opinions).

Figure 2. Example of Structural Flexibility chart for four building layers (from centre outward: 0, Not flexible, Limited flexibility, Average flexibility, Very flexible, Extreme flexible).

77

4.3 Example of Flexibility Indicators Table 1. Indicator B1 functional free height.

Flexibility class Free height

Functional free floor to ceiling height Hf (m)

I II III IV V

Hf < 2,4 2,4 <= Hf < 2,6 2,6 <= Hf < 2,8 2,8 <= Hf < 3,5 Hf >= 3,5

Not flexible Limited flexibility Average flexibility Very flexible Extreme flexible

(The functional free height is influenced by the other assumed heights for example height for servant elements) Values are base on Dutch building regulations Offices Housing Schools, classrooms Sport facilities: Hf = 5,0 m Hf For Hf >= 5,2 m Mezzanine floors become possible

Flexibility factor FB1 0,2 0,4 0,8 1,0 1,0–2,0 (Depending on value of Hf )

5 CONCLUSIONS With the proposed definition of Structural Flexibility (and Structural Adaptability) together with the proposed evaluation method it becomes possible to evaluate and compare both existing as well as newly designed building structures with regard to their Flexibility. The proposed multi criteria charts give a clear visualisation of the Structural Flexibility of a given structure showing its strong and weak points. A Structural Flexibility Index will need a closer examination and further search for weighing factors and Flexibility factors. Evaluation of Structural Flexibility is important because a high Structural Flexibility will increase the structures functional qualities. Future adaptations to changing user requirements will be easier. This will result in a higher probability of a longer Functional Working Life of the structure, a better match between Technical Service Life and Functional Working Life. It will increase the possibilities for future adaptations and refurbishment, thus resulting in a lower waste production and a lower environmental impact of the building structure in general. It will increase the possibility of balancing investments at the design stage against returns over the buildings service life. The proposed evaluation method of Structural Flexibility can be developed into a useful tool in Integrated Life Cycle Design and Engineering. Further research to specify and quantify the necessary indicators is needed.

REFERENCES [1] Herwijnen, F. van, Blok R. LCA comparison of two different building structures etc. ILCDES 2003. Integrated Life Cycle Design and Engineering of Structures, conference proceedings, Kuopio Finland. [2] Brand, S. How buildings learn: what happens after they’re built. 1994, New York, Viking. [3] Leupen, B. Frame and generic space, A research on adaptable housing. (In Dutch, summary in English) Rotterdam 2002, 010 Publishers. [4] Herwijnen, F. van. Integrated floor design, based in the IFD design approach. Proceedings of the conference on Advances in Structural Engineering and Mechanics, ASEM 2004, Seoul.

78

Session 2: Structural integrity 1

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Seismic upgrading of RC structures by means of composite materials: a state-of-the-art review G. Della Corte, E. Barecchia & F.M. Mazzolani Department of Structural Analysis and Design, University “Federico II” of Naples, Naples, Italy

ABSTRACT: FRP-material systems are gaining increasing popularity as seismic rehabilitation tool, because of several well-known advantages with respect to more traditional materials, such as lightness, easy of application, resistance to aggressive environments. Main fields of application are strengthening of shear-deficient structural components, confinement of plastic hinges, clamping of lap-splices. This paper presents a brief review/overview of the research on the use of FRP-material systems for the seismic repairing/upgrading of reinforced concrete structures. The main focus is on the most up-to-date experimental research. Some results from a recent experimental activity on a full-scale RC existing structure are also presented as study case.

1 INTRODUCTION Generally speaking, FRP-material systems can be used for rehabilitating civil engineering structures with the following purposes: – – – –

increasing flexural strength and stiffness; increasing axial load capacity; increasing shear (and torsion) strength; increasing ductility and displacement capacity.

The first three objectives are not earthquake-engineering specific. Whereas the last one is very typical of seismic up-grading activity. In earthquake engineering applications, also the first three design goals, in addition to the strength increasing, are often finalized to the structure ductility improvement, by eliminating brittle collapse mechanisms. For example: the shear strength increase can shift the failure mode towards a flexural-dominated one; the bending and compression strength increase of columns can allow satisfaction of modern hierarchy criteria in frame member strength distribution. However, FRP-material systems can also be addressed to directly increase the local ductility, e.g. by improving the compressive strain capacity of concrete by confinement of plastic hinge regions. Then, within a specific earthquake-engineering perspective, the design goals, to be satisfied using FRP materials, can be grouped into two main approaches: a) Increasing local and/or global ductility and deformation capacity by favouring the most ductile failure mechanisms. This could involve changing of the type of plastic hinge (e.g. from shear to flexure) and/or the location of plastic hinges within the structure (e.g. from story to global mechanisms). b) Increasing local ductility and deformation capacity of existing potential plastic hinges. The difference is that in the second case the designer does not aim to change the type either the location of plastic hinges, as he does in the first case. 81

2 OVERVIEW OF THE TECHNICAL LITERATURE 2.1 General This section provides a summary of several technical papers dealing with the use of FRP materials in the seismic strengthening of RC structures. The overview is mainly focused on experimental activities and provides only information about the most recent research. Hence, it is not considered to be exhaustive of a subject with a steady worldwide increase in the number of studies. Papers collected from existing technical literature have been grouped in the following main categories: 1. 2. 3. 4. 5.

Changing the failure mode from shear to flexure Avoiding lap-splice failure Strengthening partition walls Full-scale tests Special topics.

Categories 1 and 2 are the most typical, with papers dating back to 1999. Category 1 includes papers addressing the topic of favouring the most ductile failure mechanisms, by eliminating brittle shear failures and forcing the formation of ductile flexural plastic hinges, according to the above approach a). Contrary, category 2, includes papers dealing with the goal of improving ductility of existing potential plastic hinges, therefore, following the approach b). Category 3 is relatively new, being all papers published in 2004. The idea of strengthening existing partition walls, thus making them effectively participating in the structural response up to collapse, opens the door to a perspective of application of FRP that only few years ago could be thought to be impracticable using FRP. It gives rise to the possibility of improving the global structural response by correcting irregularities (torsional response) in the building behaviour. Category 4 is includes papers dealing with both a) and b) approaches, in the sense that the tested FRP-strengthening systems could be addressed to satisfy of one or a combination of the generic design goals previously stated. Category 5 includes papers which show different interventions for solving very specific problems of given structural types. 2.2 Changing the failure mode from shear to flexure Mosallam (2000) carried out physical tests on half-scale laboratory models of interior beam-tocolumn joints of RC frame structures. Both un-reinforced and fibre composite reinforced joints were tested, showing appreciable increase in strength (up to 53%) and ductility (up to 42%) of joints. Ghobarah and Said (2001) tested a seismic rehabilitation system for shear-reinforcement deficient joints in RC framed structures. The proposed system, in case of exterior beam-column joints, consists on wrapping the joint area with a U-shaped G-FRP laminate, with the free ends of the U tied together by threaded steel rods, driven through the joint section, and a steel plate. In this way, they bypass the problem of passing through the existing beam with the fibre sheet, but allow the development of the full strength of the laminate preventing premature failure by delamination of the fibre wrap and forcing the formation of a flexural plastic hinge at the beam end. El-Amoury and Ghobarah (2002) presents experimental results of tests carried out on plain RC and G-FRP reinforced RC beam-column joints, showing once again the possibility to avoid the shear failure of the joint through the externally bonded composite reinforcement and forcing the plastic hinge to form by flexure at the beam end. In addition, they strengthened the bottom part of the beam end in such a way to integrate the existing non-adequately anchored steel reinforcement. The latter G-FRP sheets reduced fixed-end rotation effects improving the sub-assemblage stiffness. In all the retrofitting schemes, steel plates and angles were used for avoiding the composite sheets debonding. The FRP-debonding problem was, instead, not encountered by Li et al. (1999), who tested prototype-scale beam-column joints, both plain and reinforced, using an innovative hybrid FRP 82

sheet. The hybrid FRP sheet was made of E-glass woven roving, plain carbon cloth and chopped strand mat and glass fibre tape with a vinyl-ester resin. This hybrid composite resulted in a low elastic modulus that helped, in the views of theAuthors, eliminating the debonding of the sheets. The glass woven roving and carbon cloth were disposed with a bi-axial plain weaving which provided equivalent strength in both axial and hoop directions. Both the beam and the column were wrapped with the bi-axial composite, having the care of increasing the radius of curvature in the corner deviations for reducing peeling stresses. One of the (few) analytical studies of FRP-strengthened beam-to-column joints is reported by Parvin and Granata (2000), who developed numerical finite element models of exterior beamto-column joints reinforced by using FRP-materials and compared them with the response of an un-reinforced control specimen. The reinforcement was supposed to be both in the longitudinal and transverse directions of both beam and column, with a fibre wrap placed at the corner deviations in order to absorb peeling stresses. Results showed an increase in the moment capacity of up to 37%. Prota et al. (2003, 2004) carried out physical tests on joints reinforced using near surface mounted (NSM) FRP round bars passing through the joint and, thus, integrating the shear strength of the joint. Both monotonic and cyclic tests were performed, again showing promising results in terms of both strength and ductility capacity improvement. The possibility to control the local failure mode of RC structural members is testified also by the experimental and numerical results of Lee et al. (2004). These tests show that C-FRP wrapping can produce an increase of the member shear strength large enough to allow plastic hinging in bending. Ghobarah and Khalil (2004) investigated the shear strengthening and ductility improvement of RC shear walls by using C-FRP sheets and C-FRP or steel anchors. Experimental test results illustrate the change of the collapse mechanism from shear to flexure thanks to the bidirectional (±45◦ ) C-FRP reinforcement and the improvement of ductility thanks to the C-FRP wrapping of the two ends of the shear wall where high compressive strain demand can develop. Test results also emphasize the importance of an effective anchoring of the C-FRP sheets, in order to avoid premature debonding and consequent loss of strength, with a better response of the steel anchors with respect to the C-FRP anchors. Tsonos (2004) carried out experimental tests on beam-to-column RC joints, explicitly including the presence of the horizontal slab. Both original and strengthened joints were tested, comparing jacketing by C-FRP sheets with the more classic RC jacketing. Both pre-earthquake strengthening and seismic repairing/upgrading of joints were experimented. Results show that the original specimens failed by shear in the joint area, whilst the strengthened specimens, both with FRP and RC jackets, exhibited a flexural plastic hinge in the beam. Actually, a better response of the post-earthquake repairing with RC jacket with respect to the analogous FRP system was measured. 2.3 Avoiding lap-splice failure Extensive experimental results on the effects of FRP wrapping of RC rectangular columns with lap-splices of existing longitudinal bars are reported by Bousias et al. (2004). Their study includes variation of parameters such as the type of bar (smooth with hooked ends or straight with ribs), the length of splices, the number of FRP wrapping layers, the longitudinal length of the FRP wrapping, in addition to the bond properties of the bars. The Authors indicate that there was no appreciable improvement of the response in case of smoothed bars with hooked ends, independent of the examined parameter values. In case of straight ribbed bars, the increase of the number of C-FRP layers (from 2 to 5 layers) slightly improved the effectiveness of the wrapping, but the improvement effectiveness was not commensurate to the number of C-FRP plies and the effects were also strongly dependent on the length of the existing steel reinforcement lap-splices. In particular, the Authors indicate that the adverse effects of short lap-splices cannot be fully removed by the FRP-wrapping technique if the lap splicing is as short as 15 bar-diameters. Experimental results on FRP-wrapping of RC rectangular columns are also presented by Ilki et al. (2004). They tested specimens made by low strength concrete, reinforced by straight ribbed bars and with inadequate transverse reinforcement. Both specimens with lap-splices of longitudinal 83

bars in the plastic hinge region and with continuous reinforcement were tested. The Authors indicate that when short lap-splices are present FRP wrapping does not improve the lateral inelastic response as much as they can, when ductility is limited by concrete crushing and longitudinal bar buckling (continuous reinforcement). Analogous results were obtained byYalçin and Kaya (2004), who conducted tests on RC columns with a rectangular cross section wrapped in the plastic hinge zone with C-FRP sheets. The Authors suggest that wet-lay-up C-FRP sheets do not provide the required confinement stress improving the bond-slip response in case of lap splices of longitudinal straight bars. Contrary, this technique was effective in case of continuous longitudinal bars. Results of static cyclic tests on hollow square-section bridge piers (1:4 scaled), strengthened with both FRP wrapping and additional longitudinal FRP reinforcement are given in Pavese et al. (2004). The Authors notice that, in case of usual lap splices of existing longitudinal steel reinforcement at the base of the pier, FRP wrapping does not provide a large enough increase of confinement able to guarantee the transfer of the tensile forces in the cross section through the lapped steel bars. In this case, additional longitudinal FRP reinforcement is required. However, the basic problem of the foundation-anchoring of this newly added reinforcement must still be solved, in such a that it proves to be effective under large tensile forces, but keeping the simplicity of the plain FRP system. Schlick and Breña (2004) presented an experimental study on the use of FRP for wrapping the plastic hinge region of bridge columns with a circular cross section. The Authors indicate that FRP jackets, fabricated with a wet-lay-up procedure, changed the failure mode of the tested specimens from a non-ductile lap-splice failure at the base to a ductile flexural plastic hinge failure mode. Besides, the confining pressure of the FRP jackets increased the lateral bending strength between 19% and 40%, meanwhile maintaining the integrity of the column by avoiding the longitudinal bar buckling at large lateral displacements. The possibility of using FRP wrapping for improving the inelastic response of plastic hinges of circular-section RC columns, with lap-splices of longitudinal bars, is also indicated by the experimental results obtained by Chung et al. (2004), who tested bridge piers in a 1:2.5 scale. 2.4 Strengthening partition walls Erdem et al. (2004) tested a RC frame upgraded by using the shear strength of hollow clay tile walls reinforced by means of diagonally placed C-FRP sheets, which were epoxy-bonded on the wall surface, extended on the frame members and connected to them by C-FRP anchor dowels. The test results were compared with those relevant to the bare frame and to the frame strengthened by using RC shear walls instead of the hollow clay tile walls. The results indicate that the lateral strength and stiffness of both the upgraded frames were about 5 times and 10 times those of the bare frame. However, the hollow clay tile walls failure mode was relatively more brittle, owing to the loss of strength of the C-FRP anchor dowels. Experimental test results on the contribution of C-FRP laminates to stiffness, strength and deformation capacity of brittle walls made of hollow bricks are also given in Erol et al. (2004). Garevski et al. (2004) presents experimental dynamic shaking table tests on 1/3-scale specimens of RC frame structures with infill walls strengthened with C-FRP strips epoxy-bonded on the inner and outer faces of the walls and also mechanically connected with anchor dowels. Their results indicate a remarkable reduction (−48%) of the lateral displacement demand to the strengthened specimen with respect to the un-reinforced one. 2.5 Full-scale tests on existing real structures Pantelides et al. (2004) carried out pushover cyclic tests on 5 real RC bridge bents. Three of them were tested as control specimens and the remaining two bents were tested after retrofitting with externally bonded FRP sheets. In addition, one of the control specimens was repaired and strengthened using FRP materials and then re-tested. The retrofitting system involving C-FRP 84

materials consisted in both an additional flexural reinforcement in the longitudinal direction of columns and a C-FRP wrapping in the transverse direction of the plastic hinge regions, in addition to a shear reinforcement of the joint area. The bents ultimately failed always owing to lap-splices failure of existing steel reinforcement, but the strengthened structure exhibited larger strength and displacement capacity than the original bents, meeting the seismic performance objectives as set at the design stage. One of the first extensive testing of an FRP-based system for seismic repairing/strengthening of RC structures is presented by Fyfe and Milligan (1998). The Authors also give several examples of practical applications to bridge and parking garage structures. According to the Authors, two of these structures withstood the 1994 Northridge earthquake, performing as designed, what could be considered a full-scale test of an existing real structure. 2.6 Special topics Johnson and Robertson (2004) presented the results of experimental tests on “gravity-only” slabcolumn connections failing in punching shear and retrofitted using C-FRP shear studs. Their results indicate promising effectiveness of the proposed technique in increasing the displacement capacity corresponding to the punching-type shear failure, which is important for buildings located in seismic areas, where also the “gravity-only” slab-column connections must maintain their vertical load bearing capacity up to the lateral displacements required by earthquakes.

3 CRITICAL ANALYSIS OF EXISTING EXPERIMENTAL RESULTS Until now, the majority of both numerical and physical tests of FRP-strengthened RC joints has been carried out considering externally-bonded FRP-materials applied in such a way to wrap up the joints with fibres often disposed both along the member axis and in the transverse direction, but never, passing through the beam-to-column node. This obviously reflects the actual difficulties of passing fibres through an existing monolithic joint. Unfortunately, this difficulty strongly limits the potentials of FRP materials, because of debonding problems, in the form of peeling-off failure at corners for the longitudinal FRP flexural reinforcement (where the reinforcing member intersects another surface) or delamination at the ends of shear-strengthening plies. In fact, the plain FRP system has often been proposed in association with steel anchoring devices (see, e.g. Ghobarah and Said, 2001, El-Amoury and Ghobarah, 2002, Pavese et al., 2004), especially in the case of shear strengthening of RC beam-to-column joints. The problem of anchoring FRP flexural reinforcement at corner deviations is strictly connected to the problem of eliminating or, at least, reducing bond-slip effects and improving the ductility and the structural integrity of plastic hinges by FRP-wrapping. In fact, the critical regions are quite always located at the ends of the member axis, where the intersection with another member or, very often, with a slab occurs. Besides, in case of existing gravity-load designed old structures, these critical regions are characterised by the lap-splicing of longitudinal steel reinforcement, with inadequate lap lengths. In all these cases, the impossibility to add a continuous FRP reinforcement in the critical section, which would reduce the fixed-end rotation effects, seems to have addressed the research towards the use of wrapping for clamping lap-splices and reducing bond-slip effects. Existing experimental results on the effectiveness of FRP-wrapping are promising in case of columns with a circular cross section, such as it could be the case of bridge piers (Schlick and Breña, 2004, Chung et al., 2004). Unfortunately, results for rectangular or square (either full or hollow) sections (more often encountered in building structures) are much less encouraging, indicating that the FRPwrapping cannot fully solve the problem, especially in case of short lap-splices (Bousias et al., 2004, Ilki et al., 2004, Yalçin and Kaya, 2004, Pavese et al., 2004). Perhaps, this is the reasons why several researchers turned to the use of the wrapping technique in conjunction with the addition of flexural reinforcement (El-Amoury and Ghobarah, 2002, Pavese et al., 2004), which reduces the tensile forces to be transmitted by the existing steel-reinforcement splices. 85

Two more observations can be made looking at existing literature papers: – experimental investigations are much more numerous than theoretic studies; – there are no experimental studies, at the authors’knowledge, dealing with the problem of avoiding the formation of flexural plastic hinges in columns of building structures. As far as the second aspect is concerned, it must be observed that, in case of existing gravity load designed RC building structures, the increase in flexural strength of columns required for moving plastic hinges to beams, could be relatively large, because of the small initial column over beam ratio of flexural strengths. Besides, the bending strength increase that can be achieved by means of externally bonded FRP reinforcement is limited by two problems: 1) peeling-off at corner deviations usually does not allow the development of the full composite action; 2) the FRP contribution to flexural strength increase reduces as far as the column axial force increases, because of the usually small compressive strength of externally bonded fibre composites. The first problem could be bypassed, in case of RC framed structures, in several ways: a) by using special anchoring devices, such as steel plates and rods as proposed by El-Amoury and Ghobarah (2002) and Pavese et al. (2004); b) by using near surface mounted bars, such as proposed by Prota et al. (2004), where round bars are to be inserted in small holes made with appropriate machines passing through the joint, with the care of avoiding the cut of existing steel reinforcement. Anyway, it must be emphasized that the anchoring devices make the plain FRP-system loose its simplicity. All these problems seem to strongly limit what can be done with FRP-material systems, when the aim is to change the type of collapse mechanism by moving plastic hinges from columns to beams. However, a large part of existing RC old buildings exhibits a slab-column connection at the least in one direction. In this case, some vertical holes can be locally made around the columns, without requiring temporary supports, in such a way to allow a continuous fibre application in the longitudinal column direction. Besides, in this case, the ideal beam for connecting column can be identified in slab, with a depth equal to the slab thickness and the width appropriately chosen by considering the effective contributing portion of slab. The latter observation implies that the column over beam bending strength ratio of the initial structure is relatively higher, due to the smaller plastic strength of beams, thus reducing the required flexural strength increase in columns.

4 STUDY CASE: A FULL SCALE TEST This section presents experimental tests of the lateral load-displacement response of an existing gravity-load designed 40 years old RC structure. The study is part of a wider research, named the ILVA-IDEM project (Mazzolani et al., 2004a), where ILVA-IDEM is the acronym of ILVAIntelligent DEMolition. In fact, the tested structure has been obtained starting from an existing building located in the Bagnoli district of Naples, in the area of the previous steel mill named ILVA. After the political decision to dismiss the steel mill and convert the industrial plant into a cultural and leisure center, the building was destined to be demolished. Then, the idea was to carry out an “intelligent demolition”, by using some of the existing structures as full-scale specimens for a testing activity. The general purpose of the whole research is the experimental/theoretical study of several advanced technologies for the seismic retrofitting/upgrading of existing RC structures. The building under study can be considered as representative of a large number of existing RC buildings in the South of Italy, built after WW2 during the 50 s, 60 s and 70 s before the inclusion of Naples in a seismic prone area. Figure 1a shows the building at the beginning of the investigation. In order to increase the potential number of specimens for testing different upgrading solutions, slabs were cut at the first and second floor, in such a way to divide the whole building into six separate structures to be analysed. Before cutting the slabs, external and partition walls, as well as non-structural 86

Figure 1. The building under investigation. 160

Base Shear (kN)

140 120

Positive Envelope

100

Negative Envelope

80 60 FRP-Strengthened

40

Un-strengthened

20 0 0

0,05

0,1

0,15

0,2

0,25

0,3

Top Displacement (m)

Figure 2. Summary of results of pushover tests.

elements, were removed. Figure 1b shows the building after these preliminary operations, also highlighting the different seismic upgrading systems under investigation. The study presented in this paper deals with the response of structure number 3, that is the third structure shown in Figure 1b starting from the left. The following main steps summarises the research articulation: 1. Lateral loading up to collapse of the initial structure, in order to provide the control response of the un-strengthened structure. 2. Design and application of a seismic repairing/upgrading system based on the use of carbon fibre reinforced polymers (C-FRP). 3. Lateral loading up to collapse of the strengthened structure. Figure 2 is a synthesis of tests results in quantitative terms. In particular, Figure 2a summarises results of both the test on the original un-strengthened structure and the test on the FRP-strengthened structure. It can be seen that the original structure exhibited a top-story sway collapse mechanism, whilst after the FRP rehabilitation the collapse mode was characterised by the formation of plastic hinges at the column bases and in the horizontal floor beams. Figure 2b illustrates the comparison between the response of the original and upgraded structure in quantitative terms, reporting the base shear vs. top displacement relationships. The response curves for the FRP-strengthened structure are the envelopes of the cyclic response in the positive and negative fields of loading. From Figure 2, the strength of the FRP-strengthened structure is seen to be increased by 86% of the initial value if the positive envelope is considered and 100% of the initial value if the negative envelope curve is contemplated. Analogously, the lateral top-displacement capacity is increased of about 100% of the initial value irrespective of the sign of the imposed displacement. 5 CONCLUDING REMARKS FRP-material systems are gaining more and more attention both in the research and application fields, as an advanced seismic retrofitting tool. Advantages of these materials, such as lightness, easy of application, possibility of tailoring for specific needs and resistance to aggressive environments, are well-known features and apply also in the seismic field of implementation. 87

In the relatively brief history of applications in the earthquake engineering field, shear strengthening, confinement of plastic hinges and clamping of lap-splices are the earliest uses which have been proposed. Currently, these applications are still to be considered as the most interesting and diffused. However, more recently different uses are being proposed and developed, such as strengthening of brittle partitioning walls or improving the bending strength of columns in order to move plastic hinges to beams; special applications have also been proposed, such as retrofitting of “gravity-only” slab-column connections. However, two main aspects need to be carefully evaluated, for a full exploitation of FRP. Namely: a) peeling-off failure at corner deviations, where the reinforcing member intersects another surface; b) small contribution of FRP in compression, if not adequately restrained against local buckling. Research is expected in the near future for deeper understandings of these two problems, in order to further improve the potentials of FRP as seismic upgrading system.

REFERENCES Bousias, S., Spathis, A.-L., Fardis, M.N. (2004). “Seismic retrofitting of columns with lap-splices through CFRP jackets.” Proc. of the 13th World Conference on Earthquake Engineering, (CD-Rom, paper n. 765), Vancouver, B.C., Canada, August 1–6. Chung, Y.S., Lee, D.H., Park, C.K., Song, H.W. (2004). “Curvature variation of earthquake experienced RC bridge pier in the plastic hinge region.” Proc. of the 13th World Conference on Earthquake Engineering, (CD-Rom, paper n. 2097), Vancouver, B.C., Canada, August 1–6. El-Amoury, T., Ghobarah, A. (2002). “Seismic rehabilitation of beam-column joint using GFRP sheets.” Engineering Structures, (24), 1397–1407. Erdem, I., Akyuz, U., Ersoy, U., Ozcebe, G. (2004). “Experimental and analytical studies on the strengthening of RC frames.” Proc. of the 13th World Conference on Earthquake Engineering, (CD-Rom, paper n. 673), Vancouver, B.C., Canada, August 1–6. Erol, G., Yuksel, E., Saruhan, H., Sagbas, G., Tuga, P.T., Karadogan, H.F. (2004). “A complementary experimental work on brittle partitioning walls and strengthening by carbon fibers.” Proc. of the 13th World Conference on Earthquake Engineering, (CD-Rom, paper n. 979), Vancouver, B.C., Canada, August 1–6. Fyfe, E.R., Duane, J., Milligan, G.P. (1998). “Composite materials for rehabilitation of civil structures and seismic applications.” Proc. of the Second International Conference on Composites in Infrastructure, Eds.: Saadatmanesh and M.R. Ehsni, Tucson, AZ. Garevski, M., Hristovski, V., Talaganov, K., Stojmanovska, M. (2004). “Experimental investigations of 1/3scale RC frame with infill walls building structures.” Proc. of the 13th World Conference on Earthquake Engineering, (CD-Rom, paper n. 772), Vancouver, B.C., Canada, August 1–6. Ghobarah, A., Said, A. (2001). “Seismic rehabilitation of beam-column joints using FRP laminates.” Journal of Earthquake Engineering, 5(1), 113–129. Ghobarah, A., Khalil, A.A. (2004). “Seismic rehabilitation of reinforced concrete walls using fibre composites.” Proc. of the 13th World Conference on Earthquake Engineering, (CD-Rom, paper n. 3316), Vancouver, B.C., Canada, August 1–6. Ilki, A., Tezcan, A., Koc, V., Kumbasar, N. (2004). “Seismic retrofit of non-ductile rectangular reinforced concrete columns by CFRP jacketing.” Proc. of the 13th World Conference on Earthquake Engineering, (CD-Rom, paper n. 2236), Vancouver, B.C., Canada, August 1–6. Yalçin, C., Kaya, O. (2004). “An experimental study on the behaviour of reinforced concrete columns using FRP material.” Proc. of the 13th World Conference on Earthquake Engineering, (CD-Rom, paper n. 919), Vancouver, B.C., Canada, August 1–6. Johnson, G.P., Robertson, I.N. (2004). “Retrofit of slab-column connections using CFRP.” Proc. of the 13th World Conference on Earthquake Engineering, (CD-Rom, paper n. 142), Vancouver, B.C., Canada, August 1–6. Lee, Y.-T., Kim, S.-H., Hwang, H.-S., Lee, L.-H. (2004). “Evaluation on the shear strengthening effect of RC columns with carbon fibre sheets.” Proc. of the 13th World Conference on Earthquake Engineering, (CD-Rom, paper n. 1369), Vancouver, B.C., Canada, August 1–6. Li, J., Bakoss, S.L., Samali, B.,Ye, L. (1999). “Reinforcement of concrete beam-column connections with hybrid FRP sheet.” Composite Structures, (47), 805–812.

88

Mosallam, A.S. (2000). “Strength and ductility of reinforced concrete moment frame connections strengthened with quasi-isotropic laminates.” Composites: Part B, (31), 481–497. Pantelides, C.P., Duffin, J.B., Reaveley, L.D. (2004). “Design of FRP jackets for seismic strengthening of bridge T-joints.” Proc. of the 13th World Conference on Earthquake Engineering, (CD-Rom, paper n. 3127), Vancouver, B.C., Canada, August 1–6. Parvin, A., Granata, P. (2000). “Investigation on the effects of fiber composites at concrete joints.” Composites: Part B, (31), 499–509. Pavese, A., Bolognini, D., Peloso, S. (2004). “Seismic behaviour of RC hollow section bridge piers retrofitted with FRP.” Proceedings of the 13th World Conference on Earthquake Engineering, (CD-Rom, paper n. 2831), Vancouver, B.C., Canada, August 1–6. Prota, A., Nanni, A., Manfredi, G., Cosenza, E. (2003). Capacity assessment of RC sub-assemblages upgraded with CFRP. Journal of Reinforced Plastics and Composites, 22(14), 1287–1304. Prota, A., Nanni, A., Manfredi, G., Cosenza, E. (2004). “Selective upgrade of under-designed RC beam-column joints using CFRP.” ACI Structural Journal, September–October, vol. 101, n.5. Schlick, B.M., Breña, S.F. (2004). “Seismic rehabilitation of reinforced concrete bridge columns in moderate earthquake regions using FRP composites.” Proc. of the 13th World Conference on Earthquake Engineering, (CD-Rom, paper n. 508), Vancouver, B.C., Canada, August 1–6. Tsonos, A.G. (2004). “Effectiveness of CFRP-jackets and DR-jackets in post-earthquake and pre-earthquake retrofitting of beam-column sub assemblages.” Proc. of the 13th World Conference on Earthquake Engineering, (CD-Rom, paper n. 2558), Vancouver, B.C., Canada, August 1–6.

89

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Structural integrity of buildings with precast load bearing walls under gas explosion G. De Matteis, I. Langone & F.M. Mazzolani Department of Structural Analysis and Design, University of Naples “Federico II”, Italy

ABSTRACT: This paper is addressed to the analysis of the main strategies to avoid the progressive collapse of buildings with precast RC load bearing walls under gas explosion. The possibility of a building to experience a progressive collapse depends on several factors which are related to both the gas explosion hazard and the structural robustness. Two different protection strategies are analyzed in this paper aiming at reducing the risk of progressive collapse: (1) the key element strategy and (2) the alternative load paths strategy. The former is based on the design of appropriate precast load bearing walls, whose structural response can be estimated by means of the Baker’s method. The obtained related numerical results are represented in a Pressure–impulse diagram considering different features of the connecting system. The latter is based on the acceptance of extensive damages, namely the collapse of at least one story. To show the applicability of the above strategies, a typical building based on precast load bearing walls is analyzed in the paper, computing comparatively the cost related to the examined design solutions to avoid the progressive collapse of the structure.

1 INTRODUCTION Gas explosion is defined as a “fast chemical reaction of gas in air, happening at high temperatures and high pressure and having as a result the propagation of a pressure wave”. The main difference between gas explosion and detonator explosion is in the speed of propagation of the flame, it being is in the range of 10–100 m/s in the case of gas explosions. To burn all the gas in the compartment, the flame should have a higher speed, since a large part of the gas gets expelled outside (Smith and Rose, 2002). In Figure 1, the typical time-dependence of the pressure measured in a compartment with windows and without any windows is shown. The progressive collapse of a building can be defined as the collapse of a relatively small portion of the structure due to an abnormal action, which results in the failure of a major portion of the structure. So the progressive collapse is an extensive structural failure initiated by a local structural

without windows

with windows

Figure 1. Typical time variation of pressure wave due to the gas-explosion.

91

damage and a consequent local failure. The risk of progressive collapse depends on the gas explosion hazard, the compartment geometry, the building exposition and the structural robustness. In particular, the building exposition depends on the developed activity (private habitation, hospital, cinema, theatre . . . etc.). The compartment geometry determines the value of peak pressure (irregular geometry produces turbulences that increase the flame front and the amount of gas burnt, while windows and sacrificial external walls could be used to expel some amount of unburned gas). Gas explosion hazard can be strongly reduced if sensors or automatic techniques avoiding loss of flammable gas are applied. The structural robustness is related to the applied structural design since it is provided by member and connection ductility as well as by structural over-strength and redundancy (Mainstone, 1972). The problem of progressive collapse due to gas explosions is particularly felt for buildings whose structural scheme is based on the use of precast bearing walls (Armer, 1977a; Kumar, 1977). In fact, in such a case, when an economical design is carried out, the robustness of the structure is usually limited, leading to a high vulnerability of the building against gas explosions. Different strategies can be adopted to avoid the progressive collapse of a structure: – Event control, which aims at avoiding and/or protecting the building against an accident that might lead to the progressive collapse; – Indirect design, which aims at providing adequate resistance against the progressive collapse through a minimum level of strength and ductility of the applied structural components. – Direct design, which considers explicitly the strength of the structure for progressive collapse (specific local resistance method) and its ability to absorb damages (alternative load path method). Usually, when the direct design method is applied, two main strategies are considered: – Key element strategy (specific local resistance method), which is based on the correct detailing of the structural components more vulnerable against the explosion (in case of precast RC bearing walls, slab-to-wall connections and the walls themselves have to be designed to absorb the input energy by means of elastic-plastic deformations, see Figure 2). – Alternative load paths strategy, which is based on the acceptance of the failure of some elements, but with the preservation of the main structural components (for instance, when the bearing wall collapse is considered, the floor slab has to be designed with adequate robustness, allowing the collapse prevention for the consequent exceptional loading condition).

Figure 2. Precast load bearing walls with steel connections.

92

2 HISTORICAL REVIEW AND REGULATION ASPECTS Buildings with precast load bearing walls are used since many years. In particular, the concept of predefined building systems made a great progress in Europe just after the World War II, when it was necessary to provide mass housing for thousands of persons made homeless by the war destruction. Prefabricated load-bearing units become an integral part of the structure, taking the vertical and horizontal storey forces. Precast concrete walls are very useful and convenient for this purposes. In fact, for low or mid-rise structures, the amount of reinforcement required handling and erect precast concrete walls is often more than necessary for carrying imposed loads. For taller building, limited additional reinforcement may be necessary for the lower level panels (Waddell, 1974; Eliot, 2002). The Ronan Point building collapse (London, 1968) shown that precast bearing wall buildings were susceptible to progressive collapse. An explosion occurred on the 18th storey with a pressure peak of about 80 kN/m2 resulted in a collapse of an entire corner of the building. It should be observed that in Great Britain during the two years from 1968 to 1970, 122 cases of progressive collapse due to the gas-explosion happened. Also, Italian statistics show numerous cases of gas explosion in the period from 1959 to 1999, where the deaths due to progressive collapse of buildings have been about 65. After the Ronan Point accident, in 1972 a British Building Regulations document was introduced for buildings with five or more storeys and a guidance was also given for reducing the sensitivity of buildings to disproportionate collapse in the event of an accident (Armer, 1977b). These rules made no reference to dynamic loads, debris impact or ductility requirements. Besides, the provisions were developed at a time when computational analysis methods were limited. The above document has also been criticized for being too narrowly focused on the avoidance of a future Ronan Point type of accident without any consideration to the risk associated to other hazard or exceptional actions (Mainstone, 1974, 1976). Eurocode 1 – Part 1–7 (2004) and advanced national regulations (Moore, 2003) provide possible strategies to be adopted for safeguarding buildings against foreseeable accidental actions and unforeseeable actions provoking local damages. Foreseeable actions includes also internal explosions for which specific rules are given, while unforeseeable actions are related to accidents that could affect the structure and cannot be anticipated by the designer. Unforeseeable actions determining local damage foreseen by the designer are related as accidental actions, while exceptional actions are those which are unforeseen in the design process. General rules for building protection against unforeseeable actions are also given by strategies intended to ensure that the structure can have sufficient robustness (over strength, ductility and redundancy). 3 APPLICATION OF THE KEY ELEMENT STRATEGY 3.1 General The structural response of a building compartment with precast load bearing walls is based on the response of the single wall. In particular, the compartment model is composed of two floor slabs connected to a load-bearing wall (Figure 2). Usually, the wall is fixed to the floor slab to avoid relative horizontal displacements, while it is not connected to the contiguous walls. Therefore each wall can be schematized as simply supported on two sides. The structural response may be determined by means of the Baker’s method (Baker et al., 1992). Therefore, the mechanical response of the wall can be defined as: – Impulsive, if the duration of blast pressure is significantly shorter than the natural vibration period of the precast wall [t d /T < 0.4]; – Quasi-static, if the duration of blast pressure is significantly longer than the natural period of the precast wall [t d /T > 2.0]; – Dynamic, if the duration of blast pressure is similar to the natural period of the precast wall [0.4 < td /T < 2.0]. 93

Figure 3. Idealized pressure–time diagram for gas explosions.

The impulsive realm response does not depend on the peak pressure rise time (tr ), while it is safe to assume a pressure rise time equal to zero (tr = 0) in case of quasi-static realm (see Figure 3). Only when the structural response is in the dynamic realm, the actual pressure rise time value is important and it should be estimated with particular care (Corley et al., 2000). 3.2 Idealized pressure–time diagram The history of pressure P versus time t is assumed as triangular with zero value of peak pressure rise time (tr = 0) to determine either impulsive asymptote or quasi-static asymptote. The assumption of a triangular diagram having a defined time rise (tr = 0) provides an exact estimation of quasi-static asymptote, but more refined calculations are necessary. The simplification due to assuming tr = 0 when the pressure blast is applied also for a long duration (compared to the system natural period) is that the external work may be computed for the pressure maximum value P* corresponding to the time when the maximum strain energy of the system is produced. 3.3 Modeling of connections The adopted connecting system for precast RC walls is very important to provide stability and robustness to the whole structure. In fact, several cases of progressive collapse are due to connections, since they have not been designed for abnormal loads due to fire, impact, explosion, and so on. Indeed, the strength of a structure should be governed by the strength of the structural members rather than by the strength of the connections, unless the connecting elements are designed as key elements whose failure is controlled in the design phases (Owens and Moore, 1992). In general, steel connections used in precast RC walls may be schematized through an elasticplastic behavior characterized by an elastic stiffness kϕ , a moment capacity Mpl (activated at the elastic limit rotation ϕel ) and a limit plastic rotation ϕpl , so that global ultimate rotation may be defined as ϕul = ϕel + ϕpl . 3.4 Modeling of precast walls Usually, a vertical wall panel can be schematized like a slender beam, whose response under transversal loads is characterized by the bending stiffness (EI). The base material is not ductile so the bending response of the system has to be limited to the attainment of the material elastic limit deformation. The beam-deformed shape is assumed as a static deformed shape, which gives the exact response of the system only in the quasi-static loading realm. Indeed, when an impulsive loading realm is of concern, the assumption of a parabolic beam deformed shape gives a better response compared with the static deformed shape. Anyway, the influence of the assumed deformed shape on the energy solution is very small when impulsive loads are applied, so, as a simplification, in the following a static deformed shape is assumed even for an impulsive loading realm. 94

3.5 Evaluation of system response 3.5.1 General The impulsive and the quasi-static response of the system are determined by applying the Energy Balance Method. Therefore, strain energy, kinetic energy and external work must be determined. The strain energy (SE) may be defined as:

where the beam strain energy (SE)b and the elastic-plastic spring strain energy due to each connecting element (SE)s are defined as following:

where w(x) is the deflection of precast wall, M (x) is its bending moment, L is the height of precast wall. The kinetic energy (KE) may be computed as:

where b = dc is the section width of precast wall and also the connection distance, h is the depth of precast wall section, dm is the mass unit, ρ is the concrete density, i is the impulse normalized to the area unit, it being computed as the integral of the assumed P–t diagram (see Figure 3). Finally, the external work (EW ) is defined as:

where p∗ = P ∗ · b, P ∗ being the peak pressure of the gas explosion. 3.5.2 Quasi-static realm When the loading is applied for a long duration, the work done by external loads is that corresponds to the maximum value of pressure p* exhibited during the duration time td in the time range p(t), which produces the maximum strain energy in the system. Therefore:

After laborious calculations, one can find:

95

where β = kϕ /EI is the elastic stiffness ratio and:

For instance, the deflection value for a simple supported beams results to be:

3.5.3 Impulsive realm When the applied pressure is impulsive, the initial kinetic energy is imparted to the system and removed before the system itself develops significant strain deformation. Therefore, the applied kinetic energy is absorbed later on by strain energy:

Appropriate calculations lead to compute the impulse i as:

3.5.4 Dynamic realm A transition for dynamic loading realm can be found by using a hyperbolic tangent squared relationship, which, based on practical experience, fits quite well the actual complicated behaviour of the system (Smith and Hetherington, 1984):

For small values of the argument, the hyperbolic tangent equals the argument and we obtain the impulsive asymptote. For large values of the argument, the hyperbolic tangent equals the unity and we obtain the quasi-static asymptote. For different conditions we find that the applied loads (P ∗ , i) satisfying the above relationship. 3.5.5 Pressure–impulse diagram The application of the above relationships leads to the following diagram (Figure 4), which has been obtained considering the following values of the parameters: – for connections: kϕ = 45000 kNm/rad; Mpl = 150 kNm; dc = 1.5 m; ϕpl = 0.005 rad; – for precast concrete walls: h = 200 mm; L = 3 m. In the impulsive realm only the impulsive value is of concern, while in the quasi-static realm the response of the system depends only on the pressure value. When the response is dynamic the “P–i” combination is important. When the response is quasi-static the maximum value of the pressure corresponds to the quasistatic asymptote, in the case being P* = 69 kN/m2 . When the response is impulsive the maximum value of the impulse corresponds to the impulsive asymptote, in the case being i = 0.68 kNs/m2 . Pressure–impulse diagram allows checking whether a “p–i” combination is compatible or not with the relevant precast wall system. In fact, “p–i” combinations falling to the left and below of the curve will not produce the failure of the system, while those falling to right and above the curve are concerned with excessive damage of the system. 96

Figure 4. Pressure–impulse diagram for a precast wall with ductile connections (µ = ϕul /ϕel = 1.5). Plate of connection t a front weld

a3 a1

bmin bmax

lateral weld

Wall-to-wall connection Plate welded in situ front weld

Fc

Ft

Wave press.

Figure 5. Proposed wall-to-wall connection.

3.6 Connection design Connections commonly used for precast concrete walls do not possess the mechanical features required to absorb a blast wave due to the gas-explosion. Therefore, an innovative type of connection system has been studied and proposed (Figure 5) to ensure a high response in terms of plastic moment, stiffness and ductility (µ = ϕul /ϕel ). Wall-to-wall connections are made of steel plates welded on both the internal and the external sides of the precast panel. Steel plates are in-situ 97

welded by both lateral and front welds. The bearing capacity of the connection system depends essentially on the tensile strength of the minimum section of the applied steel plate. The shear force produces a significant effect only close to the wall-to-wall gap, where the shear is absorbed by the connecting plates. Lateral welds ensure that the bending deformations are not important compared to the axial deformation. The connection ductility capacity depends on the length of the plate ongoing in plastic range, which, as a first approximation, could be assumed equals to three times the plate thickness t. Besides, it is worth noticing that the maximum elongation εpl of this plastic zone has been limited to 0.1. The precast load bearing wall response is a quasi-static response. Therefore, to absorb a peak pressure equal to P ∗ = 69 kN/m2 (see Figure 4) the connection has to be designed in such a way to ensure the following characteristics: Mpl = 150 kNm; kϕ = 45000 kNm/rad; dc = 1.5 m. A connection having the following characteristics meet the above requirements: material strength fy = 430 N/mm2 ; weld thickness aw = 7 mm; plate thickness t = 7 mm; plate width bmax = 440 mm and bmin = 260 mm; plate length a1 = 30 mm, a2 = 50 mm, a3 = 50 mm, a = a1 + 2a2 + 2a3 = 230 mm.

4 APPLICATION OF THE ALTERNATIVE LOAD PATH STRATEGY The alternative load path strategy is based on using floor slabs having sufficient robustness, i.e. structural capacity in terms of strength and ductility to avoid the development of a more extensive collapse as a consequence of the failure of some walls or at the most of one storey. Progressive collapse could be developed from the compartment that contains gas stored, for instance the kitchen of a flat. Several accidental situations may be taken into consideration. In the following three typical situations are considered. – Accidental situation 1, when the collapse is related to an external load-bearing wall, due to the loss of the external support. The possibility to avoid the collapse of the storey depends on the robustness of the upper floor slab, which has to work with a different structural scheme respect to the design one. It is interesting to find the minimum robustness of the floor slab to avoid the entire storey collapse, when an external load-bearing wall fails. To this purpose, a floor slab supported on three load-bearing walls is considered (Figure 6a). – Accidental situation 2, when the central load-bearing wall collapses. Then, a large bending moment develops in the mid span with an opposite sign compared to the design condition. Also in such a case the possibility to avoid the storey collapse is due to slab robustness. – Accidental situation 3, when the slab has not enough robustness and the local failure of the single wall due to gas-explosion produces the collapse of the entire storey. In such a case, the relevant storey impacts with the lower storey with a large amount of kinetic energy. Even accepting one storey collapse, the progressive collapse initiated by the failure of the lower storey should be avoided, such additional failure meaning the collapse of the whole building. To avoid floor-byfloor collapse the impacted floor slab has to have sufficient robustness to absorb the kinetic energy transmitted by the upper failing part. In particular, when the accidental situation 3 is of concern, to find the minimum robustness of the floor slab we consider that the impacting dynamic load is applied as an uniform load at the rise a) External wall collapse

b) Internal wall collapse

Static gravity load

Static gravity load

c) Collapse of the upper storey dynamic load static load

Figure 6. Examined accidental situations.

98

time tr , corresponding to the impact time of upper storey debris. Generally, this loading condition can be considered as quasi-static. Then, it is realistic to assume that the reinforced concrete slab has a rigid-plastic behaviour. Therefore, the collapse mechanism is characterized by a plastic hinge located on the central support and two plastic hinges located in the mid spans of lateral bays. The floor slab should absorb the kinetic energy of the debris by means of internal strain energy. To find the kinetic energy we suppose that the interstorey depth h is equal to 3 m. So the kinetic energy is given by the following equation:

where M is the debris mass, v is the debris velocity and G is the debris weight. The vertical load G impacting on the lower storey may be evaluated as a function of the characteristic permanent loads (gk ) and the characteristic value of variable actions (qk ), the latter reduced by a proper combination factor due to the exceptional event, as in the following:

The interaction force resulting from the impacting mass (M ) may be determined according to Eurocode 1 – Part 1–7. When assuming as elastic body the impacted structure and as rigid body the impacting object, the interaction force F is given by:

where, in the case being, k represents the stiffness of floor slab, which can be determined considering the full inertia of the slab and the elastic modulus of the concrete. Examined common cases show that concrete slabs are not able to absorb the corresponding input energy. In particular, the ratio between the external work and the maximum internal strain energy is estimated to be about 25% (Langone, 2003). The minimum safety coefficient s1,min that should be adopted in the design of the lower slab for a normal loading condition, avoiding the collapse of that slab when the gas explosion occurs causing the collapse of the upper storey, can be evaluated solving the following equation for s1 , which provides the allowable safety coefficient s2 in the exceptional situation:

where M1,Rd and M2,Rd are the moment capacity for normal loading (γr = 1.15; γc = 1.6) and exceptional loading (γr = 1.0; γc = 1.0), respectively; M1,Sd and M2,Sd are the maximum value of bending moment for normal loading (γG = 1.4; γQ = 1.5) and exceptional loading (γG = 1.0; γQ = 0.2), respectively; χmech is the structural mechanism coefficient. For the floor slab typology supported by three walls here analysed χmech = 1.47, while:

So the minimum safety coefficient to be adopted in the normal loading case, corresponding to s2 = 1, is given by:

99

On the other hand, if a reduced safety coefficient in the normal condition is considered (s1 = 1), the above relationship shows that the work of external actions is not absorbed by the internal strain energy, the resulting safety coefficient being:

5 THE STUDY CASE A typical precast concrete bearing walls building is considered in the following (see Figure 7) to estimate the additional costs due to ensure a correct design under gas explosion load for avoiding storey progressive collapse. A compartment with higher risk of gas explosion, which could be either the kitchen or a room close to a gas tube, is selected (ignition point). When an explosion occurs, it is supposed that the closing walls (not load bearing walls) are expelled with the advantage to reduce the maximum pressure due to the explosion. The load bearing walls hit directly from the explosion should be designed to resist to this accidental action or an alternative load path have to be identified. The following protection strategies are considered: – All the walls (from W1 to W4 ) are designed to absorb the gas explosion energy, leading to a constructional additional cost C1 ; – The wall from W1 to W3 are designed to absorb gas explosion energy and floor-slab2 is designed with a sufficient robustness, leading to a constructional additional cost C2 ; – Floor-slab1 and floor-slab2 are designed considering the failure of the wall W1 , W2 , W3 and W4 , with upper storey collapse, but the lower floor slab has a sufficient robustness to avoid progressive collapse, leading to a constructional cost C3 . When the key element connection is designed to absorb a pressure of equal to 100 kN/m2 , the cost of connection for a meter of wall may be estimated equal to 135 euro/meter, assuming a connection weight equal to 27 kg/m and a cost for worked steel of 5 euro/kg. Assuming a price 2a W1

W2

W3 Ignition point I1

I3+

b fl. slab 1

+ I2

fl. slab 2

W4

Figure 7. The assumed building plan.

100

for normal connections equal to 40% of cw , the additional cost for “explosion design” may be estimated equal to ckey = 81 euro/m. The panels with these connections have an additional steel bar reinforcement that is not important compared to the basic reinforcement necessary to absorb handling and erection loads. When the slab length is equals to L = 5 m, the additional steel bar reinforcement ((FeB44k) to avoid the floor collapse, when the external wall is expelled (accidental situation 1), is equal to Af = 3.11 cm2 /m, which means 2.4 kg for slab square meter. Assuming for steel bars a cost of 1 euro/kg, the additional cost for accidental situation 1 may be estimated equal to cacc,1 = 2.4 euro/m2 . The additional steel bar reinforcement to avoid the storey progressive collapse due to upper slab failure (accidental situation 3) is related to the slab length L (see Table 1). In Table 2, the total additional costs for the three different design solutions are given. In Figure 8, different cost curves are provided for different slab length and peak pressure values due to gas explosion. The relevant cost analysis is related to the single compartment and the connection cost (C1 ) is evaluated considering a linear variation due to the decrease of the pressure. In such a Figure, different cost curves Ci,40 , Ci,60 , … Ci,100 refer to a specific value of the assumed peak pressure. It appears that the solution C2 is more favorable than the solution C1 both under the economical point of view and because, due to the failure of wall W4 , it allows a significant reduction of the design peak pressure on the bearing walls. The solution C3 does not depend on the peak pressure. From a technical point of view, it can be adopted up to slab span equal to 6 m. In particular, for reduced length of floor slab (less than 5.50 m) it appears the most economical one, the other ones providing a more satisfactory performance only for applied peak pressure P* ≤ 60 kN/m2 , which represent lower values than those that should be considered in practical cases. Possible realistic examples of such an application are two-storey industrial buildings, where there is dangerous or flammable material stored in the upper storey. Table 1. Accidental situation 3 – additional cost for superior and inferior steel bar reinforcement. Slab length (L)

A∗f1 [mm2 ]

A∗∗ f2 [mm2 ]

Cost1 [euro/m2 ]

Cost2 [euro/m2 ]

7.00 6.50 6.00 5.50 5.00 4.50 4.00

11789 9387 7077 5161 3637 2106 1368

544 599 649 696 739 778 812

92 74 56 40 29 17 11

4.3 4.7 5.1 5.5 5.8 6.1 6.4

∗ Additional

superior reinforcement is considered on a length equal to 0.35 L. inferior reinforcement is considered on a length equal to L.

∗∗ Additional

Table 2. Cost of different strategies (P* = 100 kN/m2 ). Plan length (a = b = L)

C1 [euro]

C2 [euro]

C3 [euro]

7.00 6.50 6.00 5.50 5.00 4.50 4.00

2268 2106 1944 1782 1620 1458 1296

1916 1740 1566 1409 1270 1132 992

– – 1693 1110 716 413 263

101

Figure 8. Additional cost curves.

6 CONCLUSIONS Precast concrete load bearing walls can be designed by using appropriate connecting systems performing as key elements under a gas explosion. On the other hand, if a high damage level is accepted (one storey collapse), a suitable design of the floor slab with sufficient robustness can avoid the progressive collapse of the structure. The analyses carried out in this paper, which have been referred to a three supported floor slab under the impact of the upper storey, shows that the minimum robustness to be assigned to avoid the progressive collapse corresponds to a design safety coefficient for normal design conditions nearly equal to 4. The proposed Pressure–impulse diagram represents a practical design tool for precast panels under gas explosion. It allows the selection of the mechanical characteristic of the wall-to-wall connection for a given peak pressure and impulse load due to gas-explosion. It is important to observe that all the results obtained in this study are related to the application of simplified calculation methods. The refined structural analysis under gas explosion loading could be also carried out by means of sophisticated finite element models, but the correct set up of such models would require the implementation of experimental analyses based on both dynamic and static loading conditions. Therefore, further developments in these directions are strongly advisable.

REFERENCES Armer, G.S.T. 1977a. Abnormal dynamic loading of a multi-storey panel structure, NR B 484/77, BRE, Garston, p. 15. Armer, G.S.T. 1977b. A general review of recent BRE research into behaviour of panel structures, NR B 485/77, BRE, Garston, p. 12. Baker, W. E., Spivey, K. H. and Baker, Q. A. 1992. Energy-Balance Methods for Structural Response and Damage Analysis Under Transient Loads, Structural Design for Hazardous Loads 229–237, E and FN Spon, London.

102

Corley, W. G. and Oesterle, R. G. 2000. Dynamic Analysis to Determine Source of Blast Damage, Abnormal Loading on Structures, E and FN Spon, London. Elliot, K. S. 2002. Precast Concrete Structures, Butterworth-Heinemann, Oxford. Esper, P. and Hadden, D. 2000. Dynamic Response of Buildings and Floor Slabs to Blast Loading, Abnormal Loading on Structures, E and FN Spon, London. Eurocode 1, Part 1–7, 2002. General Actions – Accidental actions due to impact and explosions, Stage 34 Draft. Kumar, S. 1977. Lateral loading tests on an 18-storey large panel model structure, NR B 486/77, BRE Garston, p. 12. Langone, I. 2003. Structural Integrity of Buildings with precast load bearing walls in case of explosion, Graduation Thesis, Tutors: F. Mazzolani, I.A. Louca, G. De Matteis, University Federico II of Naples, Naples. Mainstone, R. J. 1972. The Hazard of Internal Blast in Buildings, Building Research Establishment, Garston, Watford. Mainstone, R. J. 1974. The Hazard of Explosion, Impact, and Other Random Loadings on Tall Buildings, Building Research Establishment, Garston, Watford. Mainstone, R. J. 1976. The Response if Buildings to Accidental Explosions, Building Research Establishment, Garston, Watford. Moore, D.B. 2003. The UK and European Regulations for Accidental Actions, BRE, Garston. Owens, G.W., Moore D.B. Feb. 1992. The robustness of simple connections, The Structural Engineer, Vol. 70., No.3. Smith, P. D. and Hetherington, J. G. 1994. Blast and Ballistic Loading of Structures, Butterworth-Heinemann, Oxford. Smith, P. D. and Rose, T.A. 2002. Blast Loading and Building Robustness, Progr. Struct. Engng Mater. 4: 231–223. Waddell, J. J. 1974. Precast Concrete: Handling and Erection, American Concrete Institute, Detroit.

103

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Seismic response of light-gauge steel stick-built constructions: a research project L. Fiorino & G. Della Corte Department of Structural Analysis and Design, University of Naples “Federico II”, Naples, Italy

R. Landolfo Department of Constructions and Mathematical Methods in Architecture, University of Naples “Federico II”, Naples, Italy

ABSTRACT: This paper presents some results obtained from both experimental and theoretical studies carried out at the University of Naples on the seismic behaviour of light-gauge steel stickbuilt residential buildings. Experimental tests have been carried out on two full-scale prototypes designed in such a way to contain all the structural elements and components used for this type of structures. Experimental results allow considerations on typical seismic behaviour issues (capacity design, in plane strength and stiffness of horizontal diaphragms, performance criteria) to be drawn. The theoretical part of the study has been dedicated to both judge the seismic performance of the structure under increasing seismic intensity levels and develop an ad-hoc cyclic loading protocol for physical tests. Finally, the planning of a new testing program aiming to characterize the behaviour of sheathing-to-member connections is then illustrated.

1 INTRODUCTION In recent years, the use of light-gauge steel (LGS) members in low-rise residential buildings has increased significantly in industrialised countries. In fact, LGS-framed housing is growing in popularity in the USA and Australia. In particular, in Hawaii 25–30% of all new houses have been built with this constructional typology. The next highest percentage is Australia with about 12%. California comes in third with just under 5%. Although the market for LGS-framed housing is increasing in Europe, it represents still less than 1%. Even if buildings with LGS frames have advantages over more conventional structures, at this time they have not made much impact on the European market. The reasons for this occurrence is mainly the result of low familiarity with these innovative kind of structures in comparison to the well known masonry, wood and concrete. Also in the USA and Australia designers and contractors were rather diffident in using LGS products for residential buildings. During the last years, this trend has changed through an intensive diffusion of information regarding materials, construction technologies and details, building process, structural design for LGS housing. Recently, also in Europe different initiatives have been undertaken for promoting the development of LGS constructions (Ogden et al. 1998, Burstrand et al. 1998). The basic construction systems for low-rise LGS houses are: (1) “stick-built” constructions, (2) “panelised” constructions and (3) “modular” constructions. Undoubtedly the “stick-built” system is the most used solution, therefore, this paper is particularly focused on this construction system. The stick-built constructions are directly derived from the wood construction systems. They are obtained by the assembling on the job-site of LGS members with C or Z section. Namely, the walls (vertical-load bearing system) is obtained by means of single or coupled back-to-back C-section (studs), interconnected at each end by tracks and sheathed by gypsum-based or wood-based board, plywood or, in some cases, steel sheet or corrugated sheathings. The floors (horizontal-load bearing system) are usually made of joists interconnected on the upper side by a floor deck made of wood, 105

steel or composite materials. Steel trusses, whose members have C and/or L sections, often realize the roof structure. This paper presents some results obtained from both experimental and theoretical studies carried out at the University of Naples on the seismic behaviour of LGS stick-built low-rise residential buildings. Moreover, the planning of a new testing program aiming to characterize the behaviour of sheathing-to-member connections is illustrated. 2 EXPERIMENTAL STUDY Physical tests have been designed starting from a typical one-family one-story dwelling. The plan dimensions of the house were about 7 × 11 m, while its total height about the ground level was about 6 m. The structure was a stick-built construction in which both horizontal (roof and floors) and vertical (walls) diaphragms were cold-formed frames sheathed with structural panels. The experimental program was based on two nominally identical wall sub-assemblages (Della Corte et al. 2003, Fiorino 2003, Landolfo et al. 2004). One sub-assemblage was tested under monotonic loading, the other was instead subjected to a purposely developed cyclic loading history. The generic stud shear wall sub-assemblage is shown in Figure 1. The generic wall framing, which was 2400 mm long and 2500 mm height, consisted of single top and bottom tracks, single intermediate studs and double back-to-back end studs, spaced 600 mm on centre. The floor framing consisted of joists spaced 600 mm on centre, with single span of 2000 mm. The foundation was simulated by two 280 × 380 mm (depth × width) rectangular concrete beams. The walls were connected to the foundation by intermediate shear anchors and purposely-designed steel hold-down connectors placed in correspondence of the end studs. All the specimen components (members, panels and connections) were designed according to capacity design principles, in such a way to promote the development of the full shear strength of sheathing-to-wall framing connections. Two types of load were applied: gravity and racking loads. A gravity load of 45 kN was applied on the floor of the prototype. Racking loads were applied to the floor panels by means of two programmable servo-hydraulic actuators. Figure 2 shows a global view and some details of specimen and testing apparatus. Two load regimes were applied: monotonic and cyclic. In the monotonic regime, the specimen was loaded up to a displacement of 150 mm. In the cyclic test, the specimen was subjected to a

Figure 1. Global 3D view of the tested prototype.

106

specific loading sequence based on the results of a numerical study on the probable deformation histories the structure would be subjected to, as better illustrated in the next Section. The following symbols will be used in the following: V1 and V2 forces measured by actuators a1 and a2, respectively; d1 and d2 displacements measured by actuators a1 and a2, respectively; Lt = 4800 mm total length of walls. The global behaviour of the sub-assemblage may be synthesized by means of the relationship between the unit shear resistance v = (V1 + V2 )/Lt and the mean lateral displacement d = (d1 + d2 )/2. The v–d response curve is shown in Figure 3. During the monotonic test different behaviours were identified: – For a lateral displacement equal to 10 mm, tilting of the screws in the oriented strand board (OSB) sheathing-to-frame connections started, while bearing of the gypsum wallboard (GWB) panels begun. – For a lateral displacement equal to 36 mm (maximum shear resistance), tilting of screws in the OSB connections, as well as bearing in the GWB panels, were evident (see Figures 4 a and c). – For a lateral displacement equal to 80 mm, in both the OSB and GWB-to-frame connections screw heads initiated to pull through the sheathings. – When the lateral displacement was equal to 130 mm the screw heads completely pulled through the sheathings (see Figs 4 b and d).

(b) (b) Rotational-hinge

(c) Sliding-hinge

(c) (a) Global 3D view

(d) Close up view on the load actuator

Figure 2. Global view and some details of specimen and testing apparatus.

Figure 3. v vs. d curve for monotonic test.

107

(a)

(b) Deformation of the OSB connections

(c) (d) Deformation of the GWB connections

(e) Deformation of the wall

Figure 4. Specimen condition during the cyclic test.

Figure 5. v vs. d curve for cyclic test.

For all displacement levels, the wall framing deformed into a parallelogram and the sheathings had rigid body rotation (Fig. 4e). The global cyclic response in terms of unit shear resistance (v) vs. mean displacement (d) curve is shown in Figure 5. In this Figure, vMAX +1 represents the maximum (positive) unit shear measured during the whole loading history; vMAX +3 represents the positive unit shear measured at the third cycle of displacement corresponding to vMAX +1 ; vMAX −1 and vMAX −3 are the analogous quantities measured in the opposite direction of loading. During the cyclic test, for lateral displacements less than the ones corresponding to the maximum shear resistance, the behaviour of OSB sheathing-to-frame connections resulted from a combination of the tilting of the screws and the screw heads pulling through the OSB sheathings. The response of the GWB sheathing-to-frame connections was characterized by a combination of the bearing of the GWB panels and the screws’ heads pulling through the GWB panels. For lateral displacements larger than the one corresponding to the maximum shear resistance, heads of the end screws completely pulled through the sheathings in the upper half of the walls or, in some cases, the screws caused the rupture of the sheathing edges. For these displacement levels the deformation of the wall framing still had the shape of a parallelogram, while due to the rupture of sheathing-to-frame connections, the rotation of the sheathings was limited. Strength degradation after the achievement of the peak strength was more pronounced in the cyclic loading test, with respect to the monotonic case. This is well evidenced by the comparison between the unstable part of response in the monotonic and cyclic regimes of loading (Fig. 5). More details about these experimental results can be found in Landolfo et al. (2004). 108

7 Sa,e/g 6 5 4 3 2 1 (Sa,e,d/g)max 0 0.000

0.005

(Sa,e,d/g)min 0.010

d/h

du/h

0.015

Figure 6. IDA curves.

3 THEORETICAL STUDY An experiment can provide information on capacities, but, because of the strong interrelation between capacity and demand, due consideration must be given to seismic demand issues (Krawinkler 1996). A fully nonlinear with pinching mathematical model of the hysteretic response has been adopted in this numerical study (Della Corte et al. 2000). The model has been calibrated using both the experimental results obtained in the current study and existing experimental cyclic tests (Serrette et al. 1996a, b; Serrette et al. 1997b; COLA-UCI 2001). Only the stable part of the response has been simulated. Analyses were carried out using 26 far field records from Central Italy and adopting the incremental dynamic analysis (IDA) procedure (FEMA 350 2001). Figure 6 shows the obtained IDA curves. In this Figure the elastic (5% damped) spectral acceleration (Sa,e ) is plotted versus the maximum required inter-story drift angle (d/h). The ultimate value of the inter-story drift angle (du /h) and the design value of the elastic spectral acceleration (Sa,e,d ) are also reported in Figure 6. For the discussion, the following parameters are defined: – Sa,e,d : design value (10% probability of being exceeded in 50 years) of the elastic (5% damped) spectral acceleration; – Sa,e,u : the elastic (5% damped) spectral acceleration corresponding to the ultimate value of the inter-story drift angle. Besides, the following displacement-controlled limit states are defined: – Damage-limiting limit state: it is the attainment of a limiting value of the inter-story drift angle beyond which plastic deformations are so large to produce appreciable damage to the structure. This limiting value of the inter-story drift angle is set equal to 0.0035, on an empirical basis. – Collapse limit state: it is the attainment of a limiting value of the inter-story drift angle beyond which the residual safety of the structure against collapse is assumed negligible. This limiting inter-story drift angle is set equal to 0.013, which corresponds to the attainment of the maximum lateral strength on the static pushover curve. On the basis of the obtained numerical data, the following comments can be made: – Under the design earthquake intensity (Sa,e,d ), damage is negligible, the maximum inter-story drift angle demand being 0.33% < 0.35%. 109

µmax

average (µmax) = 6.1

60 Np 55 50 45 40 35 30 25 20 15 10 5 0

average

6 5 4 3 2 1

10 9 8 7 6 5 4 3 2 1 0

(c)

µ

average ( µ ) = 7.2

(b) 0.50 0.45 0.40 0.35 0.30 0.25 0.20 0.15 0.10 0.05 0.00

average

LA-A-01 LA-A-02 LA-A-03 LA-B-01 LA-B-02 LA-B-03 LA-C-01 LA-C-02 LA-C-03 LA-D-01 LA-D-02 LA-D-03 LA-E-01 LA-E-02 UM-A-01 UM-A-02 UM-A-03 UM-B-01 UM-B-02 UM-B-03 UM-C-01 UM-C-02 UM-C-03 UM-E-01 UM-E-02 UM-E-03

(a)

LA-A-01 LA-A-02 LA-A-03 LA-B-01 LA-B-02 LA-B-03 LA-C-01 LA-C-02 LA-C-03 LA-D-01 LA-D-02 LA-D-03 LA-E-01 LA-E-02 UM-A-01 UM-A-02 UM-A-03 UM-B-01 UM-B-02 UM-B-03 UM-C-01 UM-C-02 UM-C-03 UM-E-01 UM-E-02 UM-E-03

0

(d)

average (Np) = 25

average

LA-A-01 LA-A-02 LA-A-03 LA-B-01 LA-B-02 LA-B-03 LA-C-01 LA-C-02 LA-C-03 LA-D-01 LA-D-02 LA-D-03 LA-E-01 LA-E-02 UM-A-01 UM-A-02 UM-A-03 UM-B-01 UM-B-02 UM-B-03 UM-C-01 UM-C-02 UM-C-03 UM-E-01 UM-E-02 UM-E-03

7

ρp

average (ρp) = 0.23

average

LA-A-01 LA-A-02 LA-A-03 LA-B-01 LA-B-02 LA-B-03 LA-C-01 LA-C-02 LA-C-03 LA-D-01 LA-D-02 LA-D-03 LA-E-01 LA-E-02 UM-A-01 UM-A-02 UM-A-03 UM-B-01 UM-B-02 UM-B-03 UM-C-01 UM-C-02 UM-C-03 UM-E-01 UM-E-02 UM-E-03

8

Figure 7. Some characteristics of the inelastic deformation demand.

– The average Sa,e,u /Sa,e,d ratio is relatively large ((Sa,e,u /Sa,e,d )av = 5.4), but dispersion of data is also large. The minimum value of the ratio is 1.7, which results acceptable according to modern code suggestions for very rare earthquakes (prEN 1998-1 2003, ATC-40 1996). The obtained numerical results have also been used for selecting an appropriate loading history for cyclic testing. It has been based on the following seismic demand parameters: – – – –

Maximum normalised displacement (ductility): µmax = (d/dy )max Number of plastic deformation excursions: Np Sum of normalised plastic deformation ranges: µ ¯ = dp,i /dy Average over maximum plastic deformation range ratio: ρp = | dp |av /| dp |max

For a given value of µmax , the parameters Np and µ ¯ give a measure of the cumulative damage effects produced by the earthquake. The value of ρp gives, instead, information about the distribution of the plastic deformation ranges. Starting from the monotonic pushover physical test carried out in this research, the maximum normalised displacement capacity has been fixed equal to 6 (µmax,c = 6). Then, peak ground accelerations of considered natural records have been artificially scaled up to values corresponding to the attainment of a ductility demand equal to 6. Results of these analyses are summarised in Figures 7a through 7d. Figures 7b and 7c show the required number of inelastic excursions (Np ) and the required sum of normalised plastic deformation ranges (µ). ¯ Figure 7d illustrates instead the computed values of the ratio (ρp ) between the average and the maximum plastic deformation ranges. More details on the seismic demand numerical study can be found in Della Corte et al. (2004). ¯ and ρp have been adopted as the basis for deriving the cyclic loading Average values of Np , µ ¯ and ρp characterising the first part of history to be applied in the physical test. Values of Np , µ the loading history (µ < µmax,c ) have been derived on a trial-and-error basis, by searching the best possible matching of the average values derived from the numerical analysis of demand, 110

40

40 displacement d (mm)

displacement d (mm)

30

30

20

20

10

10

0 -10

0 0

5

10

15

20

25

-10

-20 -30

0

5

10

15

20

25

-20 -30

number of cycles

number of cycles

-40

-40

(a) “Non conventional” deformation history

(a) “Conventional” deformation history

Figure 8. Comparison between “non-conventional” and “conventional” deformation histories.

under the constraint to have a loading protocol similar to that suggested by ATC-24 (1992). The remaining part of the loading history (µ > µmax,c ) has been defined strictly following the ATC-24 (1992) suggestion. This subdivision between the ranges µ < µmax,c and µ > µmax,c derives from the limitations of the numerical model, which is able to simulate only the stable part of the physical response. The comparison between the cyclic testing protocol obtained through the numerical analysis of demand (“non-conventional”) and the loading history proposed by ATC-24 (1992) (“conventional”) is illustrated in Figure 8. From this Figure, it can be noted that in the “nonconventional” protocol there are more cycles with a smaller peak deformation in comparison with the “conventional” deformation history.

4 CONCLUSIONS Results of experimental and theoretical study illustrated in the previous Sections allow the following conclusions to be drawn: – All the components of this structural system can be designed according to capacity design principles, imposing collapse in the shear walls’ sheathing-to-frame connections (most ductile collapse mechanism), without significant increase of the cost. – In the monotonic test, the collapse mechanism was invariant during the increasing lateral displacement, whilst in the cyclic test some modifications (more brittle collapse mechanism) occurred after that the peak lateral load was achieved. These modifications produced strength degradation, after attainment of the peak load, in the cyclic test stronger than in the monotonic test. – The horizontal diaphragm can adequately transfer the horizontal loads to the vertical shear walls, without any appreciable damage. – The maximum inter-story drift angle demand, under the whole set of considered acceleration records and for the design value of the spectral acceleration, was equal to 0.33%, which is smaller than the damage-limiting value (0.35%) coming from the experimental tests carried out. – The minimum value of the ratio between the ultimate elastic spectral acceleration and its design value was equal to 1.7. This value satisfies the minimum requirement of several different seismic codes (e.g. Eurocode 8, ATC-40) for very rare earthquakes.

5 FUTURE DEVELOPMENTS: PLANNING OF TESTS ON SHEATHING-TO-MEMBER CONNECTIONS The experimental tests previously illustrated will be completed with tests on sheathing-to-member connections, in order to characterize the shear response of screw connections. 111

loaded edge distance load

single cold-formed C-section

600x600mm sheathings back to back coupled cold-formed C-sections

load

self-drillingscrew

Figure 9. Global view of the sheathing-to-member connection specimens.

The objectives of this testing program are: – to investigate the behaviour of different sheathing materials; – to examine the effect of the a/φ ratio, in which a is the edge distance from the centre of the screw to the adjacent edge of the connected part in the direction parallel to the load transfer (loaded edge distance) and φ is the nominal diameter of the screw; – to study the effect of different loading protocols. Specimens are designed using an assembly similar to that described in Serrette et al. (1997a). The sheathing-to-member connections’ test specimens consist of two single 600 × 600 mm sheathings attached to the opposite flanges of cold-formed steel members, as shown in Figure 9. Steel members are made of a single C-section on the top side whereas two back-to-back coupled C-sections are used for the bottom side. For attaching the specimens to the universal testing machine, top and bottom profiles are bolted to steel T-sections. Moreover, in order to avoid significant web deformation, a steel plate is placed at the internal side of the web of both top and bottom cold-formed profiles. The different sheathing types will be: 9.0 mm thick OSB, 12.5 mm thick GWB, and 12.5 mm thick cement board. These sheathings will be connected to the profiles with three screws (150 mm on centre spacing) to the top member and with two rows of six screws (100 mm on centre spacing) to the bottom members. For each sheathing type, two different values of a/φ ratio will be adopted. Each specimen typology will be tested with three loading histories: monotonic, “conventional” cyclic and “non-conventional” cyclic. In particular, the suggestion reported in ATC-24 (1992) will be adopted for the “conventional” cyclic loading protocol, whereas, the “non-conventional” loading history will be derived on the basis of statistic characterization of deformation demand carried-out in the theoretical study illustrated in Section 3. REFERENCES ATC-24. 1992. Guidelines for cyclic seismic testing of components of steel structures (ATC-24). Applied Technology Council. Redwood City, USA.

112

ATC-40. 1996. Seismic evaluation and retrofit of concrete buildings (ATC-40). Applied Technology Council. Redwood City, USA. Burstrand, H., Grubb, P.J. 1998. Light-gauge Steel Framing Leads the Way to an Increased Productivity for Residential Housing. Elsevier, J. Construct. Steel Res. Vol. 46/1-3. COLA-UCI. 2001. Report of a testing program of light-framed walls with wood-sheathed shear panels. Final report to the City of Los Angeles, Department of Building and Safety, Structural Engineers Association of Southern California. Irvine, USA. Della Corte, G., De Matteis, G., Landolfo, R. 2000. Influence of connection modelling on seismic response of moment resisting steel frames. In F.M. Mazzolani (ed.), Moment resistant connections of steel building frames in seismic areas. E & FN SPOON. Della Corte, G., Fiorino, L., Landolfo, R., Di Lorenzo, G. 2003. Seismic performance of steel stud shear walls: planning of a testing program. In proceedings of the 4th International Conference on Behavior of Steel Structures in Seismic Areas (STESSA 2003). Naples. Italy. Della Corte, G., Landolfo, R., Fiorino, L. 2004. Seismic behaviour of sheathed cold-formed structures: numerical tests. ASCE, Journal of Structural Engineering (submitted paper). FEMA 350. 2001. Seismic Design Criteria for New Moment-Resisting Steel Frame Construction. Federal Emergency Management Agency, Report n. 350. Washington, DC, USA. Fiorino, L. 2003. Seismic Behavior of Sheathed Cold-Formed Steel Stud Shear Walls: An Experimental Investigation. Ph.D. Dissertation, Department of Structural Analysis and Design, University of Naples “Federico II”. Naples, Italy Krawinkler, H. 1996. Cycling loading histories for seismic experimentation on structural components. Earthquake Spectra, Vol. 12, No. 1. Landolfo, R., Della Corte, G, Fiorino, L. 2004. Testing of sheathed cold-formed steel stud shear walls for seismic performance evaluation. In 13th World Conference on Earthquake Engineering (WCEE 2004). Vancouver. Landolfo, R., Fiorino, L., Della Corte, G. 2004. Seismic behaviour of sheathed cold-formed structures: physical tests. ASCE, Journal of Structural Engineering (submitted paper). Ogden, R.G., Lawson, R.M., Grubb, P.J. 1998 Steel in Housing as Part of ECSC Mega Project. Elsevier, J. Construct. Steel Res. Vol. 46/1-3. prEN 1998-1 2003. Eurocode 8: Design of structures for earthquake resistance – Part 1: General rules – Seismic actions and rules for buildings. European Committee for Standardization. Bruxelles, Belgium. (final draft December 2003). Serrette, R., Nguyen, H., Hall, G. 1996a. Shear wall values for light weight steel framing. Report No.LGSRG3-96, Light Gauge Steel Research Group, Department of Civil Engineering, Santa Clara University. Santa Clara, USA. Serrette, R., Hall, G., Nguyen, H. 1996b. Dynamic performance of light gauge steel framed shear walls. In 13th International Specialty Conference on Cold-formed Steel Structures. St. Louis, USA. Serrette, R.L., Encalada, J., Juadines, M., Nguyen, H. 1997a. Static racking behaviour of plywood, OSB, gypsum, and fiberboard walls with metal framing. ASCE, Journal of Structural Engineering, Vol. 123, No. 8: 1079–1086. Serrette, R., Encalada, J., Matchen, B., Nguyen, H., Williams, A. 1997b. Additional shear wall values for light weight steel framing. Report No.LGSRG-1-97, Light Gauge Steel Research Group, Department of Civil Engineering, Santa Clara University. Santa Clara, USA.

113

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Shaking table test of efficiency of ALSC base-isolation system Lj. Tashkov IZIIS, University St. Cyril and Methodius, Skopje, Republic of Macedonia

A. Antimovski Skopje, Republic of Macedonia

M. Kokalevski Ministry of Transportation and Communication, Skopje, Republic of Macedonia

ABSTRACT: Almost Lifted Structure Concept (“ALSC”) is a specific solution which belongs to the well known “sliding concept”, where the main structure is split from the supporting structure, thus allowing the main structure to slide, when forced. The sliding force is controlled by means of system for lifting of the main structure. The efficiency of the “ALSC” system has been demonstrated on 1/3 scale model of liquid storage tank, tested on two-component seismic shaking table at IZIIS dynamic testing laboratory. It was proved that the “ALSC” system produce significant reduction of energy transmission from supporting to the main structure within the broad frequency range, for different intensity of ground shaking. Because of its high efficiency, relatively simple installation, low cost and simple maintenance, the ALSC system is highly recommended by authors to be applied on reservoirs, special industrial structures (turbo-generator foundations, chemical laboratories), public structures (schools, hospitals, libraries, administrative buildings), residential houses, buildings etc. KEYWORDS: system.

Base isolation, sliding, Coulomb friction, lifting force, transmissibility, ALSC

1 INTRODUCTION The experience from the recent catastrophic earthquakes shows human losses and heavy damage of the structures. Many of them have been designed in accordance to the actual official codes, but still not sufficient to resist earthquake because of higher intensity of ground motion than expected or unfavorable predominant frequency. That is a very clear message for earthquake engineering scientists, that a new approach in aseismic design and construction of the structures, such as control of structural behavior should be considered. The “ALSC” system belongs to above mentioned new approach, based on sliding base isolation concept, offering an opportunity for remarkable reduction of earthquake effects, using the new original concept of “almost lifted structure” – ALSC. 2 DESCRIPTION OF ALSC SEISMIC BASE-ISOLATION SLIDING SYSTEM The friction force at the contact surface of base isolation sliding systems is defined by the following simple equation:

115

Figure 1. “ALSC” system applied on steel liquid storage tank model.

where Fr = friction force transmitted to the main structure; µ = friction coefficient in the contact surface, N = compression force at the contact surface, L = lifting force, G = gravity force (weight of the main structure). The classical sliding base isolation concept is well known in earthquake engineering practice. The main efforts in this case are oriented to reduction of friction coefficient µ by using special sliding surface (teflon pads etc) or sliding bearings. Sometimes it is expensive solution, because of high cost of the low friction materials and sliding devices. The ALSC concept improves the said sliding concept, offering a possibility to use standard sliding surface materials, such as epoxy resin or cement mortar, but still having low friction force as well. The reduction of the friction force is achieved by reduction of compression force N, on the manner that the main structure is subjected to vertical lifting force L, up to the state near lifting (before loosing the contact). Namely, the structure becomes “lighter”, and according to above expressions: (1) and (3), the friction force could be significantly reduced. This lifting system is applicable on the buildings as well as on many industrial and special facilities. In this research study, the results of experimental investigations of a base isolated liquid storage tank by “ALSC” system is presented. The main parts of the system applied on the tested model are presented in Fig. 1. The main structure is supported over flat supporting base without any interface connection. The base should be smooth in order to reduce the friction in the contact and to prevent the leakage of the liquid, which is under pressure. The bottom plate of the main structure and supporting structure are designed to resist pressure produced by liquid. The soft rubber tube fixed peripherally by metal ring, prevents the leakage of the liquid under the structure. The “working pressure” is obtained by infilling of the empty space under supporting structure by liquid, producing pressure:

which acts at the bottom and tends to uplift the structure. Low friction epoxy resin placed on the flat supporting structure reduces friction force. Finally, the centering devices such as springs, located around the tank and connected as shown in Fig. 1, keep the structure in centered position always when sliding starts. The springs are designed as much flexible as possible, but stiff enough to resist the wind effects. 3 EXPERIMENTAL TESTING OF 1/3 SCALE MODEL ON SEISMIC SHAKING TABLE 3.1 Description of the model The model was designed and built in scale 1/3, consisting of three main parts as shown on Fig. 1 and Fig. 2: 1 – steel tank infilled by water; 2 – sliding RC plate serving as a base for fixing the tank by 116

Figure 2. Instrumentation set-up.

steel bolts simulating upper foundation structure; 3 – RC supporting base fixed to the shaking table, simulating the foundation supported to the ground. The model mass in total was 17.0 tons. The contact surface between two RC plates was specially smoothed by epoxy mortar. The hole under sliding plate was designed to accept a liquid layer to produce negative pressure and to “lightening the model” and to keep it in almost lifted position. Eight steel springs symmetrically distributed around circular sliding plate were placed in order to keep the sliding plate in centered position during vibration. The piezo-metric plastic tube has been installed in order to control visually the level of the hydraulic pressure. 3.2 Instrumentation set-up The instrumentation set-up is shown on Fig. 2. Seventeen channels have been instrumented by 9 accelerometers, 3 LVDT transducers, 3 linear potentiometers and 2 strain gauges. The absolute displacement of the RC plates as well as steel tank, relative displacement between two RC foundation plates, horizontal and vertical acceleration of the RC plates as well as steel tank, stress in two point at the bottom of steel tank have been monitored for any performed test. Two resonant indicators at each RC plate have been installed in order to show transmissibility of the sliding plate i.e. effectiveness of ALSC base isolation system. 3.3 Testing program The objective of the testing program was to verify the efficiency of a base isolation system “ALSC”. First step of the program was to investigate the sliding-friction force at the contact between RC plates under static and dynamic conditions, considering different springs and liquid pressure conditions. The sliding-friction force in static conditions without springs was measured first, than the slidingfriction force with participation of the springs. The efficiency of the liquid pressure in reduction of sliding-friction force was also investigated under static and dynamic conditions. Three series of tests have been conducted for different type of dynamic excitations produced by shaking table. First of all, the resonant frequency of the model has been defined by random excitation test. In the next step, a series of harmonic excitations has been applied within the frequency range of 0.5–15.0 Hz, producing an acceleration of the shaking table within the range of 0.1–1.25 g in horizontal and 0.1–0.5 g in vertical direction. A series of earthquake excitations has been applied in the last phase of testing program in horizontal as well as by-axial, h-v direction. 117

3.4 Test results The resonant frequency of the system was obtained by performing of random excitation tests within the frequency range of 0.1–30 Hz. The frequency f = 3.4 Hz was considered as a first sliding-friction frequency. It should be pointed out that in practice, the resonant frequency of the structure base isolated by ALSC system, should have a resonant frequency lower than predominant frequency of designed earthquake which should be controlled by adjusting of centering springs stiffness. If the spring stiffness is zero, the resonance frequency should be also zero. This property is very important, because in case of earthquake, the structure practically cannot be excited in resonance, i.e. it will not move, while the base move. Figures 3 and 4 show the performance of the model base isolated by ALSC system under harmonic and earthquake excitation respectively. For harmonic excitation, series of tests within the frequency range of 0.5–15.0 Hz has been performed in order to check effectiveness of the ALSC

Figure 3. Performance of the model isolated by ALSC system under harmonic excitation.

Figure 4. Performance of the model isolated by ALSC system under earthquake excitation.

118

system. Significant reduction of the input acceleration (2–6 times) is obtained as shown in Fig. 3. Comparing to the classically designed (fixed base) structures, the maximum structural response is reduced 5–15 times depending of resonant frequency of estimated structure. Fig. 4 shows the same comparison for the earthquake excitation. As can be seen, the same reduction factors are valuable within the lower frequency range (0.5–3.0 Hz). Ones again should be pointed out that this reduction factor could be increased if stiffness of the centering springs is reduced. In this particular case, the sliding-friction force is limited to 0.2 g and it was not exceeded even the input acceleration was increased up to 1.2 g. The tabular presentation of response parameters are given in Table 1. Besides response acceleration of the sliding plate, the absolute displacement, relative displacement to the fixed plate, vertical displacement of the sliding plate, top displacement of the model, top acceleration of the model, acceleration response of resonant indicators are given as peak numerical values obtained from time history response plots. Comparative presentation of fixed plate displacement and sliding plate displacement relatively to fixed base are shown in Fig. 5. The plots show nonlinear behavior of sliding ALSC plate during earthquake excitation because of the centering springs keeping the model out of the resonance during table shaking. After stopping the shaking, the position of the model was checked. The central position was always kept. It proves the effectiveness of the springs. The tests with vertical earthquake component show very stable behavior of the model without loosing the contact and changing the working pressure under the sliding plate. The same results are obtained in by-axial tests by horizontal and vertical components of Montenegro earthquake. In this case the model reacts with dominating sliding effect without up-lifting of the ends of the sliding plate.

4 ADVANTAGES OF THE ALSC SYSTEM The “ALSC” system has many advantages comparing to the classical structural systems as well as the other base isolation systems, particularly because: • It is reliable for any earthquake intensity, as long as the supporting structure remains intact • The technical solution is very simple and doesn’t require any mechanical or electrical source • The system offers maximum reduction of energy transmission from supporting to the main structure, regardless the intensity of ground shaking • The bending and shear forces in the main structure, and consequently, relative story drifts, are almost eliminated • The sliding range of the ALSC system during earthquake is not limited as in case of rubber base isolation system • The system is not sensitive to vertical excitation component • In case of by-axial excitation (vertical and horizontal) the system react by horizontal sliding only, which is not case of rubber base isolation system • The system offers a possibility for easy displacement and/or rotation of the main structure, if required.

5 CONCLUSIONS • Model testing of a liquid storage tank, base isolated by the ALSC system, showed very effective reduction of input energy transmission by sliding of the reservoir over smooth surface, lightened by pressurized liquid layer. This simple and original solution, keep the structure out of the resonant within broad frequency range of the excitation force. • The ALSC system is controlled by two parameters: sliding-friction force and up-lifting force. The first parameter depends of smoothness of the contact surface and springs stiffness, while the second one depends of hydraulic lifting pressure. The control of above mentioned parameters was passive in this case, which doesn’t require any external electric source. Next research activities 119

120

Record

Harmonic excitation 1 RAND-H 2 RAND-V 3 HARM-H 4 HARM-H 5 HARM-H 6 HARM-H 7 HARM-H 8 HARM-H 9 HARM-H 10 HARM-H 11 HARM-H 12 HARM-H 13 HARM-H 14 HARM-V 15 HARM-V 16 HARM-V 17 HARM-V

N

3.4 15.3 3.5 3.5 3.5 0.5 1.0 2.0 3.5 5.0 7.0 10.0 15.0 1.0 5.0 10.0 15.0

FR. (HZ)

1.6 0.6 4.3 6.4 8.5 46.7 34.8 22.1 11.7 6.7 4.3 2.3 1.0 0.1 2.7 0.8 0.2

1.80 0.26 0.27 0.41 0.61 0.18 2.70 0.55 0.86 1.04 1.21 1.23 1.18 0.08 0.31 0.35 0.26

1.50 0.27 0.21 0.34 0.50 0.14 0.21 0.45 0.71 0.86 1.02 1.01 0.99 0.07 0.31 0.36 0.24

0.22 0.22 0.12 0.12 0.13 0.07 0.16 0.23 0.17 0.16 0.16 0.99 0.04 0.07 0.28 0.37 0.39

CH 20,18

AC 2,7

AC 1

0.66 0.38 0.11 0.12 0.19 0.11 0.19 0.89 0.23 0.17 0.23 0.19 0.12 0.03 0.10 0.26 0.56

CH 19

AC 4

1.26 0.17 0.40 0.57 0.75 0.38 0.41 0.70 1.11 2.72 1.67 0.69 0.37

CH 22

AC 5

1.03 0.66 0.31 0.25 0.30 0.25 0.21 0.58 0.32 1.09 0.28 0.08 0.23

CH 23

LP 3

1.5 3.9 6.0 7.9 45.7 33.8 21.0 11.4 6.4 3.8 1.7 0.6

6.2 11.2 11.9 2.4 15.9 36.8 15.2 8.1 0.2 0.1 0.1

CH 25

2.2

CH 24

2.2 5.8 4.6 47.6 43.1 2.5 4.5 0.8 0.2 0.2 0.1

1.9

CH 26

LP 2

LV DT

CH 2,4

CH 21,17

AC 3,6

CH 1,3

Displacement response (MM)

Acceleration response (G)

Displ of tabl (MM)

ACC-OF tabl (G)

Maximum response values

Input parameters

Table 1. Response parameters of the model for harmonic and earthquake excitation.

LP 1

continued

1.7 4.0 2.8 45.9 43.3 9.9 2.6 0.7 0.2 0.2 0.1

1.7

CH 27

121

Earthquake excitation 18 IZMIT-H 19 IZMIT-H 20 IZMIT-H 21 IZMIT-H 22 KOBE-H 23 KOBE-H 24 ELCENTRO-H 25 ELCENTRO-H 26 PETROVAC-H 27 NORTHRIDGE 28 MEXICO-H 29 ALMA ATA-H 30 ALMA ATA-H 31 ALMA ATA-H 32 PETROVAC-V 33 PETROVAC-H,V

1.71 1.71 1.71 1.71 0.83 0.83 1.46 1.46 1.46 1.07 0.48 2.39 2.39 2.39 25.0 2.17 25.0

53.0 51.6 76.7 109.7 51.3 102.3 68.7 97.1 71.7 73.4 77.9 10.0 24.5 29.6 26.7 72.1 31.7

0.29 0.11 0.17 0.22 0.12 0.28 0.25 0.35 0.45 0.34 0.22 0.17 0.42 0.52 0.12 0.46 0.15

0.22 0.09 0.14 0.18 0.10 0.23 0.20 0.28 0.38 0.29 0.19 0.14 0.37 0.44 0.13 0.40 1.45

0.17 0.13 0.18 0.23 0.14 0.21 0.23 0.25 0.48 0.20 0.11 0.17 0.24 0.26 0.11 0.62 1.66

0.78 0.54 0.88 1.20 0.13 0.39 0.78 0.88 0.84 0.85 0.43 0.69 1.45 1.66 0.76

0.24 0.20 0.27 0.32 0.23 0.27 0.35 0.37 0.79 0.37 0.17 0.32 0.44 0.43 0.81

1.61

0.65 0.30 0.54 0.69 0.23 0.62 0.81 1.21 1.33 10.7 0.54 0.97 1.48 1.5 77.7

2.74 6.64 16.3 35.3 4.6 39 23.9 31.5 76.5 64.2 4.5 16.4 53.4 67.6

7.8

52.3 50.2 75.3 108 50.1 100 65.6 95.5 71.2 48.3 76.1 9.1 23.5 25.7

71.3

54.3 53.5 81.4 121 54.2 115 78.9 113 97.9 103 77.9 19.1 54.9 66.9

99.0

52.6 52.5 79.9 119 52.1 110 78.4 107 95.5 101 74.4 18.5 52.8 65.7

Figure 5. Sliding effect of ALSC system for different earthquakes (ch13- displacement of shaking table, ch12- relative displacement between fixed and sliding foundation plate).

122

will be oriented to automatic control of the friction force and automatic regulation of hydraulic pressure. • Because of its high efficiency, relatively simple installation, low cost and simple maintenance, the ALSC system is highly recommended by authors to be applied on reservoirs, special industrial structures (turbo-generator foundations, chemical laboratories), public structures (schools, hospitals, libraries, administrative buildings), residential houses, buildings etc.

ACKNOWLEDGMENT The construction and testing of this large scale model required lot of efforts and material costs. The authors of this paper are very grateful to many institutions and companies participating by financing, consulting or technical assistance for successful realization of this innovative project. Particular acknowledgment should be expressed to IZIIS staff – for given technical and financial support; Ministry of education and science, Trading center – Skopje – for financial support; the companies: MZT-Hepos, Sofi-inovator, Expert, Fakom, Polat & Nikex, Tehnoguma – for technical support, and many colleagues participating in some of the phases of preparation and/or realization of the tests. REFERENCES Arya A.S. (1995) “Sliding Concept for Mitigation of Earthquake Disaster to Masonry Buildings”, University of Roorkee, India Chopra A. “Dynamics of Structures”, Prentice Hall, Inc., USA Duarte R., Emilio F., Carvalhal F.J., Pires F. (1998) “A new high-performance system of base isolation”, 11 ECEE, Paris, France Kelly J.M. (1998) “Seismic isolation as an innovative approach for the protection of engineered structures”, 11 ECEE, Paris, France Kitazawa K., Ikeda A., Kavamura S. (1991) “Study on a Base Isolation System”, Taisei Corporation, Japan Tashkov, Lj., AntimovskiA., Kokalevski M. (2002) “Seismic isolation based on almost lifted structure concept”, EURODYN2002, Munich, Germany Tashkov, Lj., Antimovski A., Kokalevski M. (2002) “Almost lifted structure concept for seismic base isolation of the structures”, 3rd WCSC, Como, Italy Tashkov, Lj., Antimovski A., Kokalevski M. (2002) “Experimental and analytical investigation of aseismic almost lifted structure concept”, International Conference VSU2002, Sofija, Bulgarija Tashkov, Lj., Antimovski A., Kokalevski M. (2002) “Seismic isolation applying almost lifted structure conceptmodel testing”, 8-th Symposium on theoretical and applied mechanics, Skopje, Republic of Macedonia Tashkov, Lj., Antimovski A., Kokalevski M. (2001) “Seismic base isolation based on almost lifted structure concept”, 9-th International Symposium of MASE, Ohrid, Republic of Macedonia Xiyuan Zhou, Miao Han (1996) “Optimum design of resilience – friction-slide base isolation system for low cost buildings”, 11 WCEE, Acapulco, Mexico Zamorano R., Sarazin M., Toro G. (1996) “Development and testing of teflon sliding bearings”, 11 WCEE, Acapulco, Mexico

123

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Behaviour of wood skeleton – OSB cladding shear-walls under monotonic and cyclic loading L.A. Fülöp, S. Bálint-Major & D. Dubina Politehnica University of Timisoara, Timisoara, Romania

ABSTRACT: House constructions in Europe are going through important transformations. Light framed constructions, both using steel and wood, are gaining more and more market share. As these construction techniques are new for most European countries there is a generalized need for codes of practice to be available for the design and detailing of such structures. Furthermore, some of the regions in Europe are also seismically active, and in these regions there is need for earthquake design specifications, which are non existent at present. The present paper describes a part of an experimental investigation undertaken to determine the lateral load bearing capacity of wood-framed OSB-sheeted wall-panels utilized in housing applications. In the first phase components of the wood-frame panels were tested and a FE modelling attempt was made to determine the load bearing capacity of entire wall-panels. The preparation phase of the wall experiments is also briefly presented. 1 INTRODUCTION 1.1 State-of-art of the light-weight housing solutions Light framed constructions, both using steel and wood skeleton, are gaining more and more market share in all European countries. In most of these countries light-framed constructions are new, having no tradition in the industry, and consequently there is lack of design codification too. The lack of design codes is usually substituted by engineering judgement and design is made according to the knowledge and expertise of the structural engineer in charge with the project. Unfortunately, the structural scheme of a light-framed house is very complex and certain aspects of the structural behaviour cannot be easily identified and are often overlooked by designers. Even more so, when earthquake design of light-framed structures is concerned. In the light of this complexity, an important way of understanding certain aspects of the behaviour of framed house structures is experimental investigation of the wall-panels, which are the main load bearing elements against lateral loads. Several researchers performed racking tests on wood and steel framed wall-panels, especially in the United States, identifying particularities of the panel behaviour and suggesting simplified design procedures. Recently, as framed structures become more used, there is a preoccupation in Europe to also develop design procedures. The experimental program on wood-framed panels at the Politehnica University of Timisoara comes as a response to the interest expressed by the local wood-framed housing industry to investigate the practical solutions presently employed in Romania. 1.2 Overview of existing wall-panel research A relatively big number of experimental researches on the racking performance of wall-panels is available in the technical literature. Until 1980 most of the tests were carried out using monotonic testing procedures and provided information on the behaviour of wall panels if the load acts preponderantly in one direction. Later, as testing procedures improved, quasi-static cyclic tests have been employed for the case of earthquake loading where the load acts in both directions. Later on, pseudo-dynamic and shake table test were also used. 125

Summaries of the experimental research on wood-framed wall panels have been published by Tissel (1993), Wolf (1991) and Foliente (1994). The main findings of these experiments, in relation with the effect of different components on the wall-panel behaviour are: (1) There are no qualitative differences in the behaviour of the wall panels sheeted with different materials (OSB, Plywood, Gypsum), the hysteretic loops having essentially the same shape. Gypsum board alone is not recommended as load bearing sheeting, but can be taken into account in combination with other materials. In this case the supplementary gypsum board increases the stiffness and the strength of the wall-panels, but the ductility is reduced. (2) If the framing members are of usual dimensions (i.e. not extremely thin or slender), the framing has little influence on the behaviour of the wall-panel. (3) The main parameters governing the behaviour of the panels are the connections. The connections between the sheeting and the frame, but also the connections between the elements forming the frame are important. From this point of view, there are more problems with the wood framing because the connection between the studs and the lower and upper track rely on the withdrawal resistance of a group of nails. As a result, there is a possibility (due to uplift forces), of stud-to-track connection failure (see experiments by Salenikovich 2000). (4) Other critical component of the wall panel assemblages are the overturning restraints (i.e. anchorages, connections to other building parts). Usually, tests are made on isolated shear walls, and the effect of vertical forces and interconnections with other walls or floors are neglected in the test. As a consequence uplift forces are transmitted to the anchoring system of the wall-panel and all researchers highlight the importance of the anchoring detail. Therefore, the components of the wall-panel that are to be investigated are the sheeting-toframing connections and the anchoring details. These components are important because they have the strongest influence on the behaviour of the panel, and also because the behaviour of the framing elements and that of the sheeting material is less ambiguous. The aim of such component research is to find the relationship between the behaviour of these connections and the overall wall-panel in order to be able to predict the behaviour of the wall based on the characteristics of the components. This could reduce the necessity for full scale testing of walls. 2 EXPERIMENTAL PROGRAM 2.1 Description, phases, expected results The prepared experimental program has to investigate the lateral load bearing characteristics of different wood-framed OSB-sheeted wall panels currently in use, under monotonic and cyclic loads (Figure 1). The investigation comprises: (1) experimental investigation of components judged to have a crucial role in determining the overall behaviour of wall-panels; (2) based on the experiment on components attempt to build a possibly simple Finite Element (FE) model to predict the response of the entire wall-panel; (3) monotonic and cyclic experiments on wall-panels; (4) comparison of the predicted FEM results with the experimentally obtained ones and (5) integration of the experimental results on wall-panels in the design of a house under earthquake loading. In this paper partial results concerning phases 1 and 2, are reported. 2.2 Experimental investigations on wall-panel components In order to evaluate the load bearing characteristics of the wall-panels by means of FE modelling, the characteristics of the wall’s subcomponent have to be known. These subcomponent-specimens have to have the same physical properties as their corresponding parts in the wall, have to be loaded similarly as in the panel and have to have a representative influence on the overall performance of the panel. Three sections were chosen, which were considered the weakest and most representative components of the wall-panels (Figure 2). 126

1200 3600

1200

1200

1200 3600

1200

160

2440

2440

2440

1940 1200

160

160

2440

1170

1200

1200 3600

1200

1200

1200 3600

1200

160

Figure 1. Wall panels to be investigated. 100 445

100 650 445

Stud

105

L profile 105

100 220

Specimen P-4

100

Stud

210

120

100 220

Nail

210

105 205

205

Anchor

OSB

100

345

755 450

L profile

Specimen P-2

750 120

100

100

Stud

220

100

Specimen P-3

310

Stud

100

100

20

220

Nail

310

OSB

750 20

100

Specimen P-1

Anchor

Figure 2. Tested wall-panel components.

The first sets of specimens (P-1, P-2) represent the nailed connections between the OSB sheeting and the studs. P-1 are nailed connections on the perimeter of the OSB plate, at 20 mm distance from the margin of the OSB, while P-2 represent nailed specimens from the central part of the OSB plate (where the margin of the OSB is far and does not influence the behaviour). The second set (P-3, P-4) represents the typical hold-down detail in case of the panels. The anchoring system is composed of a special L profile (BMF-105), which is fixed in the base concrete by means of anchors and is laterally connected to the wood panel with nails and/or wood-screws. For the anchoring system both the case of symmetrical (P-3) and asymmetrical (P-4) anchoring was taken into account. With specimens P-1 and P-2 the rigidity of the sheeting-to-framing joint was determined, which consists of a nail (Galvanized Spiral Nail; d = 3.5 mm net diameter; L = 60 mm), connecting the OSB to the stud or track. In the test the same nail and stud which is usually used on site has been considered. The stud was made of dried fir wood without any special treatment. The specimens were tested monotonically under increasing displacement (i.e. displacement control). During the tests the load bearing capacity of the specimens (F – Figure 3), and the relative slip in the tested component (d1 , d2 – Figure 3) has been measured. Five specimens of each typology have been tested. The specimens were tested with a TESTWELL universal testing machine in the laboratory of the “Politehnica” University of Timisoara. The equipment was connected to a computer controlling the experiment and making the readings during the test (Figure 4). 2.3 Results of the wall component tests 2.3.1 Specimens in group P-1 and P-2 Specimens in these two groups behaved essentially in the same way. Because the failure of the connection was due to bending of the nail and cutting of the nail into the softer fir-wood (Figure 5), 127

Specimen 1

Specimen 2

F

F

d1

d2

d1

Specimen 3

Specimen 4 F

F

d2

d1

d2 d2

d1

F F

F

F

Figure 3. Measured characteristics during the tests.

Figure 4. Testing setup in the TESTWELL machine.

Figure 5. Typical failure mode of nailed connections (P-1, P-2).

the edge distance in the OSB-panel did not have any influence on the load bearing capacity or rigidity of the nailed connections. Exceptions from the described failure mode were observed in the case of specimens P-1/2 and P-1/4 when cutting of the nail trough the edge of the OSB has been observed, but only in the final phase, after extensive bending deformation of the nails. Therefore, results in terms of rigidity and load bearing capacity from group P-1 and P-2 were essentially identical. For the experiments the force-displacement diagrams are illustrated in Figure 6, where the force is given by measurements and the displacement is calculated as (d1 + d2 )/2. From the characteristic curves it seems that this type of connection (i.e. with nails) is very ductile, a great advantage for seismic regions. The ductility of the connection is mostly due to the nail’s bending deformation under the local forces in the connection (Figure 7). Based on the characteristic curves the rigidity and bearing capacity values of the connections have been calculated, the elastic load (Fel ) being considered to be 90% of the maximum load reached during the loading history (Fmax ). The rigidity of the connection (K) has been considered as the secant stiffness at the force level of 0.6 × Fmax . The ductility of the connection (Duct ) was determined as the ratio between the displacements at Fel in the unloading branch of the experimental curve, and the displacement corresponding to the intersection of the rigidity line (K) and the horizontal line corresponding to the elastic force level (Fel ). Characteristic values, for the two groups of specimens, are summarized in Table 1 and Table 2. 128

Characteristic curves. Series P-2

3000

3000

2500

2500

2000

2000

Force (N)

Force (N)

Characteristic curves. Series P-1

1500 1000

1500 1000 500

500

0

0 0

10 20 30 Displacement (d1+d2)/2 (mm) P1/1 P1/4

P1/2 P1/5

0

10 20 30 Displacement (d1+d2)/2 (mm)

P1/3

P2/1 P2/4

P2/2 P2/5

P2/3

Figure 6. Characteristic curves for connections in group P-1 and P-2.

Figure 7. Typical nail deformation (P-1). Table 1. Properties of connection group P-1.

Table 2. Properties of connection group P-2.

Series

K (N/mm)

Fel (N)

Duct

Fmax (N)

Series

K (N/mm)

Fel (N)

Duct

Fmax (N)

P-1/1 P-1/2 P-1/3 P-1/4 P-1/5

843 940 694 805 804

787 968 968 1220 1020

20 18 16 15 11

875 1076 1075 1355 1134

P-2/1 P-2/2 P-2/3 P-2/4 P-2/5

511 976 951 682 831

886 1190 915 959 963

11 14 15 11 14

985 1322 1016 1066 1070

Mean

817

993

16

1103

Mean

817

988

14

1098

2.3.2 Specimens in group P-3 and P-4 Specimens in the two groups are meant to test if the asymmetry of the fixing detail has an influence on the load bearing capacity of the anchoring. The use of the L profiles to anchor the walls is very practical, especially when wall panels are prefabricated and sheeted on both sides with OSB from the workshop (Figure 8). Placing the anchoring L profile to both sides of the wall-panel is only possible in case of internal walls, therefore, to get a clear picture, both single sided and double sided fixings had to be tested. As a first attempt (P-3/1) the L profile was anchored with a single anchor bolt in the closest hole to the margin of the L profile (Figure 9). This eccentricity between the anchor and the fixing to 129

Figure 8. L-profile used for the anchoring of the wall.

Figure 9. Failure of P-3/1 profile due to eccentricity.

Figure 10. Improved behaviour of specimens starting from P-3/2.

the stud (i.e. force transmission plane) allowed for bending moment to develop in the profile and resulted in reduced rigidity and capacity of the fixing. In order to improve the behaviour, in the following specimens, two anchor bolts were used placed close to the stud fixing in order to reduce the lever arm of the forces (Figure 10). For these specimens better performance has been obtained. Between the L profile and the stud six spiral nails and a wood screw has been used as standard fixing. To test the possibility of reducing this fixity, tests were made with the six nails, or the screw only. In case of P-3/2, P-3/3 and P-3/4 full fixity, six nails and screw has been used, while in case of P-3/5 only the six nails have been applied. In the case of the asymmetric anchoring (group P-4) the L profile was always anchored with two bolts to the base plate, while the fixing to the stud was made with: screw only in case of specimen P-4/1, both screw and six nails in case of specimens P-4/2, P-4/2 and P-4/3 and with six nails only in case of P-4/5. Characteristic curves of all specimens are presented in Figure 11. The procedure for calculating rigidity, load bearing capacity and ductility was identical as in case of specimens from group P-1 and P-2, the results for the wall fixing connections (P-3, P-4) being presented in Table 3 and Table 4. 130

Characteristic curves. Series P-3

Characteristic curves. Series P-4 10000

15000

Force (N)

Force (N)

20000

10000

5000 2500

5000 0

7500

0

0

10 20 30 Displacement (d1+d2)/2 (mm) P3/1 P3/4

P3/2 P3/5

P3/3

0

10 20 30 Displacement (d1+d2)/2 (mm) P1/1 P1/4

P1/2 P1/5

P1/3

Figure 11. Characteristic curves (P-3, P-4). Table 3. Properties of connection group P-3.

Table 4. Properties of connection group P-4.

Series

K (N/mm)

Fel (N)

Duct

Fmax (N)

Series

K (N/mm)

Fel (N)

Duct

Fmax (N)

P-3/1* P-3/2 P-3/3 P-3/4 P-3/5*

1288 2808 2283 2760 1675

4876 7156 6653 9056 4935

5 11 11 8 7

5418 7951 7393 10062 5483

P-4/1* P-4/2 P-4/3 P-4/4 P-4/5*

514 1074 1939 1939 1182

3503 7598 8488 8225 5137

4 4 6 6 4

3892 8443 9431 9139 5708

Mean◦

2617

7622

10

8469

Mean◦

1651

8104

5

9004

* Unusual fixing of the L profile (see description in text). ◦ In the calculation of the average values the cases of unusual fixing have been ignored.

* Unusual fixing of the L profile (see description in text). ◦ In the calculation of the average values the cases of unusual fixing have been ignored.

3 FE MODELLING OF THE WALL BEHAVIOUR 3.1 Aim of the FE modelling Analytical modelling can be used as a complementary tool in understanding performance and developing design methods of structural systems. Many analytical shear wall models have been developed to predict response of shear walls to lateral loads. Assume a shear wall with tie-down anchors attached at the end studs. A simple mechanical model representing the shear wall deformation under lateral load is shown in Figure 12. The lateral load applied at the top causes the frame to distort in the shape of parallelogram. Racking of the frame is prevented by the OSB sheeting through fasteners connecting it to the frame. Due to the high rigidity, the sheeting panels deform less then the framing. The difference between the sheeting and framing deformations is resisted by the fasteners (nails and screws). Along with shear deformation of the sheeting, the shear strain of the wall is a function of the fasteners’ slip. The behaviour of the panel can be predicted using FE modelling also. A specialized FE program, WALSIZE, has been developed by Dolan, based on earlier work by Foschi (1977), but a general FE modelling tool can also be used. The benefit of the FEM is that it provides detailed information on performance of various components of the structure. Discretisation should be balanced between accuracy and computational efficiency. In this study, a FE model to predict the behaviour of an entire wall, as presented in Figure 1, was prepared in Axis VM (2000). 131

v y

H

v vH

vH L

Figure 12. Typical deformation mechanism of panel.

3.2 Preparation of the FE model The properties of the component materials were chosen as follows: (1) pin wood E = 11300 N/mm2 , δ = 480 kg/m3 ; OSB E = 9000 N/mm2 , δ = 950 kg/m3 . In the model springs were used in place of fasteners. The model was made of a pine wood framing, with cross-section of 80 × 130 mm. In the study a wall with aspect ratio of 2:3 was selected, meaning a wall segment of 2.44 × 3.6 m, with the distance between the vertical studs at 600 mm (Figure 1). In a modern house, it is unlikely that fully sheeted segments of wall without openings exceed 3.6 m in length. The end connection of the studs to the upper and lower track, which compose the frame, is made with nails staved along the grains. This connection cannot resist tension, but only compressive forces due to the contact of the two pieces of wood, being modelled using contact elements. In usual wall-panel calculation models the end connection of the studs is often considered to be pinned even if, due to the used detail, this consideration is not fully justified. According to this way of modelling, if a punching verification is necessary then this could be made using the internal force from the contact elements. The considered sheeting was OSB, the modelling being realized using shell elements with the thickness of 15 mm. Between two consecutive sheeting panels there was a space of 0.5 mm to allow the two sheets different displacements, but contact elements were also inserted between the sheeting to avoid the penetration of the plates in one another. According to the aspect ratio of the wall three pieces of vertically placed OSB sheets were used on both sides of the frame. The fasteners, which connect the frame and the OSB sheeting, were modelled as springs with the stiffness and resistance from the component-tests results (i.e. mean values from group P-1, Table 1). The distance between the nails was 200 mm according to the practical construction method used for such walls. The anchorage of the shear wall was modelled as a two-degree-of-freedom spring. The values for stiffness and resistance were from the series of experiments in group P-4 (Table 4). In order not to allow the downwards vertical displacement of the frame, a contact element which was blocked to compression and free to tension was placed parallel to each anchor. At the bottom of each vertical stud an anchor was placed. 3.3 Results of the modelling Due to the large uplift forces on the lateral studs, observed during the preliminary modelling, the replacement of the anchoring on these studs became necessary. In the original corner detail the uplift force from the stud was transmitted to the L profile trough the sheeting, because the connection between the stud and the horizontal track is not capable to transmit tensions. To improve the corner detail a much higher L profile is used for the anchoring of the wall ends, which enables the connecting of the L profile directly to the stud (BMF – KR 285). Based on the catalogue values 132

Force - F [kN]

50 40 30 20 10 H 0 0

10

20

30

40

50

Displacement [mm]

Figure 13. Resultants of the analytical model.

Figure 14. Load–displacement diagram (horizontal load only).

for this anchoring it is expected that the new L profile has a load bearing capacity and rigidity 1.4 times higher than the tested one. According to this correction value, in the FEM the stiffness of the first anchorage was taken 2000 N/mm and the resistance 10 kN. Different densities of sheeting-to-framing connections have been considered in the FE modelling, but in the following only the standard case of 20 cm nail intervals will be discussed. In case of providing nails at closer intervals, the failure mode of the panel will not be the one presented in Figure 12, but a rigid body rotation involving the failure of the anchoring system only. This failure mode is not advantageous from the point of view of the ductility of the panel, and should not be encouraged by excessive nailing of the sheeting. Two sets of FE models have been used: one without vertical loads and the second applying a vertical load of 8 kN to the top of each stud (a total of 56 kN for the entire panel). By loading the panels with horizontal loads push-over analysis has been performed on the models. In case of loading the panel with horizontal loads only, the distribution of forces in the anchoring from Figure 13 and the characteristic curve presented in Figure 14 has been obtained. It can be observed that the resultant forces V2 are smaller than V1 by 17.27 times which emphasizes again the importance of the end support anchorage (Figure 13). The yielding of the shear wall appears at FH = 33.6 kN (dH = 11.1 mm). At FH = 36.4 kN (dH = 18.3 mm) the anchorage at the end stud yields also, which corresponds to the failure of the wall. The failure of wall under horizontal loads occurred due to overturning. Loading of the sheeting-to-framing nails was observed but the load bearing capacity of the nails has not been reached. The total deflection of the shear wall consists of the sum of the: (1) sheeting shear (s ); (2) rigid body rotation (r ) and (3) slip of sheeting fasteners (n ). To asses the influence of the dead load of the upper structure a FE model with vertical load has been used. The value of the vertical load has been evaluated supposing that the wall supports one supplementary level. For this model the vertical force has been kept constant while the horizontal load was gradually increased. The load–displacement diagram of the shear wall with constant vertical load and variable horizontal load is presented in Figure 15. In this case the reaction forces are not similar like for the first model. Here, only in the first two anchorages (1 and 2 left) is tension and in the others there is compression force. The tension force in the edge anchorage is four times smaller than in the case without vertical force. The elastic yielding of the shear wall appears at FH = 39.9 kN (dH = 11.4 mm). At FH = 42 kN (dH = 17.2 mm) the anchorage yields at the edge and after that the other anchorages. The yielding force in this case is 1.18 times greater than the one with only horizontal load. A more important difference is that in this case the yielding of the sheeting-to-framing nails preceded the yielding of the anchoring system, facilitating the desired failure mode of the wall panel. The deformation modes of the shear wall can be seen in Figure 16 (where one of the OSB panels was hidden to see the deformation of the framing). In Figure 16 the deformations of sheeting, chord rotation, slip of sheeting fasteners, slip of anchorage and bending of horizontal stud can be observed. 133

Force - F [kN]

50 40 30 20 10 H+V 0 0

10

20

30

40

50

Displacement [mm]

Figure 15. Load–displacement diagram (horizontal and vertical load).

Figure 16. Deformation pattern of the wall panel (horizontal and vertical load).

4 CONCLUSIONS In this study tests on critical components of wood wall-panels are presented. The experimental results on nailed sheet-to-frame connections, but also those of the anchoring detail, showed surprising homogeneity. With the size parameters in use the nailed connection fails by bending of the nail and cutting of the nail in the grains of the wood piece. This has to be the preferred failure mode of these fasteners, because edge failure of the OSB would provide less ductility for the connection. The tests on the anchoring system emphasized again the need to avoid introducing eccentric load transmission paths in these detail. Otherwise, the anchoring system is excessively weakened both in terms of load bearing capacity and rigidity. The finite element modelling proved to be a useful tool for predicting the behaviour of a wall panel based on the characteristics of the components, even if FE results still need experimental validation. The wall-panel experiments to validate the FE results are underway. An interesting conclusion of the FE analysis is that even a relatively weak, external anchorage is capable to assure sufficient strength and rigidity to force the sheeting-to-framing connections to yield if the nailing pattern in usual. In fact, it is not advantageous to increase excessively the number of sheet-to-frame fasteners, because then again the failure of the anchoring system will govern the behaviour of the panel, leading to reduced ductility. It is important to emphasize that the FE results of this study have to be treated with some caution until they are backed with experimental evidence. REFERENCES *** 2000. Axis VM Version 5.0 – Finite Element Program, Users manual, Inter-CAD Ltd. Budapest. Foliente, G.C. & Zacher, E.G. 1994. Performance Tests of Timber Structural System under Seismic Loads. In Foliente G.C. (ed), Analysis, Design and Testing of Timber Structures under Seismic Loads, Proceedings of a Research Needs Workshop. University of California Forest Products Laboratory, Richmond, CA. Foschi, R.O. 1977. Analysis of wood diaphragms and trusses, Part One: Diaphragms, Canadian Journal of Civil Engineering, 4(3): 345–352. Salenikovich, A.J. 2000. The Racking Performance of Light-Framed Shear Walls. PhD thesis – Faculty of the Virginia Polytechnic Institute and State University. Blacksburg, VA. Tissell, J.R. 1993. Wood Structural Panel Shear Walls, APA Research report 154, American Plywood Association: Tacoma, Washington. Wolfe, R.W. & Moody, R.C. 1991. North American Structural Performance Tests of Low-Rise WoodFramed Building Systems. Proceedings of Workshop on Full-Scale Behavior of Wood-Framed Buildings in Earthquakes and High Winds, Walford, UK, XXII-1-19.

134

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Seismic retrofitting of steel and concrete structures using low-yield strength shear panels G. De Matteis Department of Structural Analysis and Design, University of Naples Federico II, Naples, Italy

E.S. Mistakidis Laboratory of Structural Analysis and Design, Dept. of Civil Engineering, University of Thessaly, Volos, Greece

A. Formisano Department of Structural Analysis and Design, University of Naples Federico II, Naples, Italy

S.I. Tsirnovas Laboratory of Structural Analysis and Design, Dept. of Civil Engineering, University of Thessaly, Volos, Greece

ABSTRACT: The paper addresses the problem of seismic retrofitting of existing concrete structures through an innovative methodology based on low yield strength metal shear panels. These panels are introduced in specific places of existing RC frames enhancing both the strength and the stiffness of the base structure. However, the most important improvement concerns the energy dissipation capacity of the retrofitted structure. The paper presents the design methodology which takes place within the performance based design framework and details for the practical application of the intervention strategy. Nonlinear finite element models are used in order to verify the reliability of the results obtained from simpler models used in everyday engineering practice.

1 INTRODUCTION Among the new seismic design approaches, there are those based on the passive structural control systems. The main aim is to reduce the damage of the primary structure through the use of special devices acting as hysteretic and/or viscoelastic dampers. Many types of devices are based on the use of metallic yielding technology. Low yield metallic shear panels could be very effective for this purpose because they allow the plastic deformation to be spread throughout the entire surface of the device rather than to be concentrated in a limited portion as it happens for many other systems (Nakashima et al, 1994; Nakagawa et al, 1996). Recently, a wide numerical and experimental research has been undertaken at the Department of Structural Analysis and Design of the University of Naples “Federico II” in order to investigate the energy dissipation capacity of shear panels made of either low-yield steel (LYS) or pure aluminium and suitably reinforced by ribs to prevent shear buckling both in the elastic and plastic field (De Matteis et al, 2003a,b). On the other hand, low yield metal steel panels could be profitably used as new methodology for the seismic retrofitting of existing concrete structures (De Matteis and Mistakidis, 2003; Mistakidis et al, 2004). In this paper, a design procedure is proposed, based on the performance design framework. This design procedure takes into account the characteristics of the original structure and requires a moderate level of engineering judgment with respect to the quantification of the displacement that would be acceptable for the original structure. The application of the preliminary design procedure gives the global characteristics of the required shear panels, i.e. the total stiffness, the total strength and the total displacement ductility. As, until now, there are no design charts 135

available, the paper gives also instructions on the selection of the appropriate panel configuration and on the modeling required in order to carry out the required static nonlinear (pushover) analysis required by seismic retrofitting codes. Moreover, on the basis of a sophisticated FEM model, which has been set up by using the non-linear code ABAQUS, a numerical study is carried out, determining the effect of the main geometrical parameters (plate local slenderness and stiffness of stiffeners) on the shear capacity of the system. The numerical analyses are performed under both monotonic and cyclic loading procedures, aiming at determining the maximum strength of adopted shear panels. Besides, useful information on the hysteretic performance of shear panels is provided. In the sequel, the results obtained using non-linear F.E. modelling, are compared against the results of the static nonlinear analysis. Finally dynamic nonlinear analyses are performed for a variety of ground motions recorded in Greece during the last 20 years. The results of the dynamic analyses, expressed in terms of maximum acceleration capacities for the retrofitted and the original (non-retrofitted) structures, are compared with those from the static nonlinear analyses. 2 USE OF LYS PANELS FOR SEISMIC RETROFITTING Due to the exceptional hysteretic behaviour and the capability to early undergo plastic deformations, low-yield metallic shear panels may be also conceived as hysteretic dampers and therefore their employment is essentially based on the concept of seismic structural passive control (Tanaka et al, 1998). To this purpose, low-yield metallic panels should be designed in such a way to avoid any buckling phenomenon, checking that the panel yields under shear. Low-yield strength metal panels can be used either as large panels rigidly and continuously connected along the confining frame members, serving also as basic enclosure system of the building, or as smaller damper plate elements installed in the frameworks of a building at nearly middle height of the storey and connected to rigid support members to transfer shear forces to the main frames (De Matteis et al, 2004a). In both cases, performed experimental tests have shown that the hysteretic behaviour of the system is excellent, provided that suitable ribs are arranged and a rigid panel-to-frame connecting system is adopted (Nakagawa et al, 1996). To this purpose, appropriate materials are low-yield point steel and heat treated pure aluminium, which are characterized by a limited yield strength and a significant strain hardening. Typical stress–strain curves for such materials are presented in Figure 1. In particular, the curves related to aluminium reflects the behaviour of such a material after the execution of adequate heat treatment, which allows a significant reduction of the initial material strength and a considerable increase of ductility. In particular, it can be noted that AW5154A aluminium alloy has a similar strength and hardening of LYS steel but a significant lower ductility, while the ‘pure aluminium’ AW-1050A is characterized by a large ductility (about 45%) and a very limited yield strength (about 25 MPa). In particular, such a limited yield strength allows to compensate the reduced Young’s modulus of aluminium respect to one of the steel. The most common method to connect metal shear panel to the existing concrete structure is with drilled or adhesive anchors (Robinson and Ames, 2000). Recently, a connection scheme has been 250

250


(N/mm2)

200

200

150

150

100

100

50


(N/mm2)

AW-1050A (Pure aluminium) AW-5154A

50 



0 0,0

0,1

0,2

0,3

0,4

0,5

0 0,0

1,0

0,2

0,3

0,4

Figure 1. Typical σ −ε-curves for (a) low yield point steel and (b) low-strength aluminium alloys.

136

0,5

devised within a research project currently being conducted at the University of Alberta (Driver and Grodin, 2001) which is not only able to transfer the interfacial forces but also to provide confinement and shear reinforcement to the concrete columns, thereby enhancing their ductility. The steel infill plate is connected to the concrete columns using a series of tube collars. It has been also shown that the application of low-yield shear panels is particularly effective in case of primary structures characterized by limited lateral rigidity and reduced overstrength. In fact, in such a case, shear panels may behave as stiffening devices also, rather than as hysteretic dampers only. Therefore, the contribution provided by shear panels could be profitably taken into account when performing the serviceability limit state check of the primary structure, which usually has a strong impact in the case of old existing structures (De Matteis et al, 2003a). 3 THE APPLICATION OF THE PROCEDURE 3.1 General A simple numerical application is provided in the following in order to demonstrate the way that LYP metallic panels can improve the seismic resistance of existing structures. The considered building is a Greek concrete structure which was built in 1968. Due to the simplicity of the structural system, the analysis can be done using the equivalent frame shown in Figure 2. The column and beam cross sections are shown in Figure 3. The steel quality was found to be equivalent to S220 and the quality of the concrete was found to be equivalent to C12. The total vertical load that participates to the seismic combination is equal to 3112 kN. 3.2 Preliminary design The analysis of the original structure using the non-linear static procedure gave the diagram shown in Figure 4. The performance point obtained for the structure gives ultimate spectral acceleration for the unretrofitted structure Sa,ini = 0.083g and a target displacement of about 6.2 cm. The initial stiffness of the unretrofitted structure is Kini = 15.500 kN/m and the initial period Tini = 0.89 sec.

Figure 2. Equivalent frame. Second floor beam 6.55 cm2 50 cm

50 cm

9.32 cm2

Column cross section

40 cm

First floor beam

2.26 cm2

1.58 cm2 Stirrups Ø6/30

20 cm

6Ø14

20 cm

Figure 3. Cross sections and reinforcements.

137

30 cm

0.5 Capacity curve of the original structure Retrofitting capacity curve from preliminary design

0.45 =5%

Spectral acceleration (g)

0.4 0.35

Required stiffness =10%

0.3 Desired target displacement 0.25

=20%

0.2

Performance point for the initial structure

=30%

0.16 0.15 0.1 0.083 0.05 0

0

0.01

0.02

0.03

0.04

0.05

0.055

0.06

0.07

0.08

Spectral displacement (m)

IO limit

LS limit

SS limit

Figure 4. Preliminary retrofitting design.

The ultimate base shear capacity of the initial structure is Vini = 258 kN. In order to perform a preliminary design, it is necessary to decide what target spectral displacement is desired for the retrofitted structure, and then, to determine the characteristics of the LYS panels that will shift the performance point to this spectral displacement. After the examination of the damage state that exists at each of the various points on the capacity curve, the design engineer may easily determine the spectral displacement levels that correspond to each performance level. In order to achieve a “Life safety” performance level, a maximum displacement of about 3 cm is required. Therefore the target displacement is Sd = 0.03 m and the spectral acceleration corresponding to the intersection of the target √ displacement line with the elastic response spectrum is Sae = 0.4g. From the formula Tret = 2π Sd /Sae , where Sae is the spectral acceleration corresponding to the intersection of the target displacement line with the elastic response spectrum, we find the required period for the retrofitted structure Tret = 0.55 sec. In order to determine the strength of the retrofitted structure, it is necessary to make an assumption about its damping properties. As a simplifying assumption it is reasonable to consider that the retrofitted structure will be able to provide at least the same level of damping as that of the initial structure. Therefore, the approximate solution for the performance point of the retrofitted structure is obtained as the intersection of the vertical line at the desired target spectral displacement with the demand spectrum that corresponds to the damping level of the initial structure. Assuming that the structure will be able to provide the same level of damping as the one of the original structure (about 25%), a required spectral acceleration capacity Sa,ret = 0.16g can be obtained from the diagram. Using the above values we find the required stiffness Kret and shear capacity Vret for the retrofitted structure from the following formula:

Therefore, due to the fact that Kret = Kini + Kp and Vret = Vini + Vp , the minimum required stiffness of the shear panel is Kp = 25110 kN/m and the required strength is Vp = 239.9 kN. 138

Figure 5. Proposed arrangement of the retrofitting system.

3.3 Design of the retrofitting system Using the results of the preliminary analysis, the following structural configuration is assumed (Figure 5), where the steel shear panels are installed in the central bay of the existing frame acting as a bracing system throughout the longitudinal direction. Concerning the characteristics of the material used for the LYSP, the elasticity modulus is E = 210000 Nmm−2 , the yield stress is f0.2 = 85.8 MPa and the ultimate stress is fu = 236.20 MPa. The stress–strain relationship of the adopted material has been defined according to material test presented in Nakashima, 1994. The assumed σ −ε diagram is the one presented in Figure 1a. In order to find a panel configuration that will result to the values obtained by the preliminary design procedure, a trial and error procedure was followed involving three different numerical models according to the following steps. – Determination of appropriate panel geometry for retrofitting purposes by means of simplified analysis. – Check of the mechanical behaviour of the selected shear panels under monotonic loading by means of sophisticated FEM analysis (using the ABAQUS F.E. code). – Check of the hysteretic behaviour of the selected shear panels submitting the aforementioned F.E. model in cyclic loading. At the end of the above procedure, it was finally decided to adopt panels having a width of 1.2 m. The height of the panel is 3.75 m and its thickness t = 4.0 mm for the first floor. For the second floor, the height is 2.75 m and the corresponding thickness t = 1.7 mm. These panels are inserted in steel frames (one for each floor) composed by HEB180 columns and HEB160 beams, which are installed in one of the ten bays of the reinforced concrete frame. For the surrounding steel frames, pinned beam-column connections have been considered. All the members of the surrounding steel frame have been designed such that they remain fully elastic for the forces transferred by the internal panels. The elements of the surrounding frame are made by Fe430 steel. 3.4 Detailed description of the procedure followed for the selection of shear panels In this first step, nonlinear analysis of the whole structure has been carried out, by using SAP 2000 analysis program. The panels have been modeled by using the “strip model” theory (SabouriGhomi et al, 2003). In particular, ten 45◦ inclined members have been used to model the shear plate pin-joined to the beams and the columns throughout the tensile principal direction (Kulak et al, 2001). In order to account for the contribution of the compression diagonals (due to adopted stiffeners and low-yield material, the considered panel should not undergo buckling phenomena), the characteristics of the tensile members of the strip model have been appropriately modified. In particular, the elastic modulus for these members was increased by 50% (this value provides the same equivalent stiffness of a system working in a pure shear mechanism). Moreover, the yield strength of each member was taken equal to 1.17·1.5 f0.2 (the factor 1.17 is needed to provide the same equivalent shear strength of a system working in a pure shear mechanism, while the factor 1.5 has been included to account for the strain hardening of the considered material). Besides, in 139

0.5 Capacity curve of the original structure Retrofitting capacity curve from preliminary design Panels capacity curve Final retrofitting capacity curve

0.45 =5%

Spectral acceleration (g)

0.4 0.35

=10% 0.3

Performance point of the retrofitted structure Desired performance point

0.25

=20%

Performance point of the initial structure

0.2 =30%

0.16 0.15 0.1 0.083 0.05 0

0

0.01

0.02

0.03

0.04

0.05

0.055

0.06

0.07

0.08

Spectral displacement (m)

IO limit

LS limit

SS limit

Figure 6. Final retrofitting design.

order to have a structural model more similar to the actual one, where the shear panels are endowed with appropriate ribs to avoid buckling phenomena, for each floor, the steel panel has been split into two separate sub-panels by means of an additional horizontal HEA160 beam, which is pin connected to the columns and placed at the mid height. The diagram shown in Figure 6 provides the Sa –Sd behaviour of the retrofitted structure. It can be observed that the response of the structure is very similar to the ideal behaviour evidenced by the initial retrofitting design. The required spectral displacement for the retrofitted structure has been calculated equal to 29 mm using the “Displacement Coefficient” method. It has been also verified that this displacement is compatible with the strength and ductility of the existing reinforced concrete members. In the second step, the behaviour of the preliminarily selected panel has been checked by means a more sophisticated analysis. For this reason, the load–displacement curve of the panel installed at the first level of the reinforced concrete structure, as it was obtained by means of the strip model, has been compared with the one obtained by the FEM analysis carried out by ABAQUS. The F.E. numerical model is composed by a steel rectangular frame, with beams and columns linked together with pinned joints, in which low yield steel shear panels are inserted. The frame members have been represented by using two-nodes BEAM T33 finite elements, having the same section as the ones employed in the SAP 2000 model, while both the shear panels and their stiffeners have been modeled through SHELL S4R elements. The connection system adopted for joining the two structural components has been the TIE constraint, through which each of the beam element is linked with the corresponding surface of the plate. In such a way the two connected parts undergo the same deformations and the employed connection does not represent a deformability source for the system. Moreover, in order to avoid shear buckling phenomena of the selected panel, appropriate stiffeners have been used. The stiffeners have a rectangular cross-section with the same thickness of the steel plate. In particular, 3 longitudinal and 11 transversal stiffeners have been alternatively placed on the two surfaces of the panel, defining in this way 300 × 300 mm plate fields. Each one of these fields has been discretized by means of a 50 × 50 mm mesh (De Matteis et al, 2004b). 140

700000 F (N)

ABAQUS with low yield material model

600000

500000

Strip model

400000

ABAQUS with elastoplastic material model 300000

Displacement range of interest

200000

100000

d (mm) 0 0

10

20

30

40

50

60

70

80

90

100

Figure 7. Comparison of the shear panel behaviour obtained by ABAQUS model and SAP 2000 model.

The minimum height of the stiffeners so that they can be considered as “rigid” (and therefore not participating in the possible shear buckling phenomena occurring in the plate) was calculated equal to 55 mm according to the recommendations of Eurocode 3-Part 1.5 (2003). An initial configuration based on the superposition of the two first positive eigenvalues critical mode has been assumed in order to represent the initial deformed configuration of the panel, with a maximum out-of-plane displacement in the order of 1.5 t. Two different analyses have been carried out, considering two different material models for the low-yield steel panels: – an elastic-perfectly plastic law with elastic limit strength equal to 1.5 f0.2 ; – the actual stress–strain relationship for considered LYP steel material. While the former allows a better and more direct comparison with the results obtained by using the simplified strip model, the latter allows the actual behaviour of the panel to be derived. The obtained results are plotted in the load–displacement diagram of Figure 7. The comparison between the curve obtained by using simplified strip model theory and the one obtained by ABAQUS adopting the elastic-perfectly plastic material having a yield stress equal to 1.5 f0.2 is very satisfactory. The above proves that the adopted strip model for the modelling of the panel provides results in good agreement with the ones obtained by using a more refined finite element model. It is worth mentioning that the difference between the simplified elastic-perfectly material model and the actual one is quite limited for the displacement range which is of interest in the specific analysis performed here (29–35 mm). Moreover, it is important to observe that the adopted stiffener configuration for the selected panel is appropriate, as it prohibits the development of buckling phenomena even for deformations larger than the required one. In order to verify the design of select panel geometry and in particular of the adopted stiffeners, a cyclic analysis of the panel has been carried out as well. The analysis has been refereed to constant displacement amplitude equal to ±35 mm, with a number of cycles equal to 20. In the cyclic tests, due to the lack of experimental evidence, the isotropic strain hardening was defined according to the default values suggested in the Abaqus User’s manual (Hibbit et al, 2001). The results are obtained in the force–displacement diagrams of Figure 8. Figure 8a corresponds to the case of the 141

(a)

(b) 800

F (kN)

800

F (kN)

600

600

400

400

200

200

0 -40

-30

-20

-10

0 0

10

20

30

40

-40

-30

-20

-10

0

-200

-200

-400

-400

-600

10

20

30

40

-600 ¬ (mm)

-800

¬ (mm)

-800

Figure 8. Cyclic behaviour of the panel by assuming (a) the simplified elastic-plastic material model and (b) the actual low yield steel material model.

Table 1. Summary of the strong motions used in the analysis. No

CODE

Location

ML

PGA (g)

No

CODE

Location

ML

PGA (g)

1 2 3 4 5 6 7 8 9

ARGO183-1 ATHENS-2 ATHENS-3 ATHENS-4 ARGO183-7 ZAK188-4 KAL186-1 EDE190-1 ARGO183-8

Argostoli Athens Athens Athens Argostoli Zante Kalamata Edessa Argostoli

6.5 5.9 5.9 5.9 5.7 5.5 5.5 5.4 5.1

0.171 0.159 0.302 0.121 0.192 0.170 0.273 0.101 0.305

10 11 12 13 14 15 16 17

PAT393-2 LEF194-1 KYP187-1 ARGO192-1 PYR193-8 KAL286-2 LEF188-2 IER183-3

Patras Leykas Kyparissia Argostoli Pyrgos Kalamata Leykas Ierissos

5.1 5.1 5.0 5.0 5.0 4.8 4.5 4.4

0.401 0.136 0.127 0.204 0.165 0.263 0.245 0.178

elastic-perfectly plastic law, while the diagram of Figure 8b corresponds to the actual stress–strain law. It can be observed that the hysteretic behaviour of the system is excellent, confirming the possibility to adopt selected shear panels as dissipative devices for improving the seismic behaviour of the existing structure. 3.5 Prediction of the structural behaviour by means of incremental dynamic analysis Finally, an incremental dynamic analysis was performed in order to verify the results of the static nonlinear analysis concerning the global structural behaviour under dynamic loading. Both the retrofitted and the initial structure were subjected to a number of 17 ground motions recorded at various Greek sites. The records used here were selected from a database of about 220 earthquakes recorded in Greece in the period between 1980 and 1999 with the following criteria. – Magnitude ML > 4.4 in the Richter scale; – PGA > 0.1g. The used records are summarized in Table 1. The pseudo-acceleration spectra corresponding to these earthquake records are presented in Figure 9 for 5% critical damping. The above records were scaled in order to achieve 20 different spectral acceleration levels ranging from 0.03g to 0.2g. These spectral acceleration levels were calculated according to the vibration period corresponding to the first eigenmode of each structure. Then nonlinear time history dynamic analyses were carried out using the SAP2000 program. The structural models were exactly the same with those used for the static nonlinear analysis. For each analysis, the maximum spectral displacement was calculated and the results were plotted in a spectral acceleration-spectral displacement 142

900 800 700 600 500 400 300 200 100 0

1000 800 600 400 200 0

0.5

1

1.5 2 Period (sec)

2.5

3

1800 1600 1400 1200 1000 800 600 400 200 0

500 400 300 200 100 0

0

0.5

1

1.5

2

0.5

1

2.5

3

Spectral acceleration (cm/sec2) 0.5

1

1.5

0

0.5

1

2

2.5

3

1000 800 600 400 200 0.5

1

1.5

2

2.5

3

Spectral acceleration (cm/sec2)

PAT393-2

0

2

2.5

3

400 300 200 100 0

0

0.5

1

1.5

2

2.5

3

LEF194-1

600

1800 1600 1400 1200 1000 800 600 400 200 0

0.5

0

400

300

300

200

200

100

100

0

0 0.5

600

500

500

400

400

300

300

200

200

100

1

1.5 2 Period (sec)

2.5

3

2.5

3

0

0

0.5

2

2.5

3

2 1.5 Period (sec)

2.5

3

2

2.5

3

2

2.5

3

1

1.5

0.5

1

0

0.5

1

1.5

KAL286-2

1

Period (sec)

1.5

2

2.5

3

Period (sec)

LEF188-2

3

Period (sec)

100 2

2.5

KYP187-1

600

PYR193-8

1.5

0

400

0

1.5 2 Period (sec)

ZAK188-4

500

600

1

1000 900 800 700 600 500 400 300 200 100 0

500

700

0.5

1

ARGO183-8

500

ARGO192-1

0

0.5

Period (sec)

600

700

900 800 700 600 500 400 300 200 100 0

1.5

700

Period (sec)

0

0

Period (sec)

1200

0

3

EDE190-1

800

Period (sec) 1400

2.5

Period (sec)

KAL186-1

0

2

ARGO183-7

Period (sec) 2000 1800 1600 1400 1200 1000 800 600 400 200 0

1.5 Period (sec)

ATHENS-4

600

0

2000 1800 1600 1400 1200 1000 800 600 400 200 0

Spectral acceleration (cm/sec2)

0

ATHENS-3

ATHENS-2

ARGO183-1

1200

1000 900 800 700 600 500 400 300 200 100 0

0

0.5

1

1.5 Period (sec)

IER183-3 700 600 500 400 300 200 100

0

0.5

1

1.5 Period (sec)

2

2.5

3

0

0

0.5

1

1.5

2

2.5

3

Period (sec)

Figure 9. Acceleration (cm/s2 ) response spectra (5% damped) of the considered ground motions.

143

0.2 Median, initial structure Median, retrofitted structure Percentile 15%, retrofitted Percentile 85%, retrofitted Percentile 15%, initial Percentile 85%, initial

0.18

Spectral acceleration (g)

0.16 0.14 0.12 0.1 0.08 0.06 0.04 0.02 0 0

0.02

0.01 Spectral displacement (m)

Figure 10. Median and 15% and 85% percentile IDA curves for the initial and for the retrofitted structure.

Table 2. Acceleration level after which dynamic instability was noticed. No

CODE

Initial structure

Retrofitted structure

1 2 3 4 5 6 7 8 9

ARGO183-1 ATHENS-2 ATHENS-3 ATHENS-4 ARGO183-7 ZAK188-4 KAL186-1 EDE190-1 ARGO183-8

0.115 0.105 0.083 0.080 0.145 0.090 0.125 0.165 0.080

0.190 0.160 0.150 0.150 0.160 0.240 0.170 0.250 0.200

No

CODE

Initial structure

Retrofitted structure

10 11 12 13 14 15 16 17

PAT393-2 LEF194-1 KYP187-1 ARGO192-1 PYR193-8 KAL286-2 LEF188-2 IER183-3

0.105 0.095 0.100 0.060 0.100 0.110 0.070 0.070

0.180 0.160 0.170 0.130 0.180 0.190 0.130 0.130

Average

0.100

0.180

diagram. Therefore 17 curves were obtained for each structure, corresponding to the considered ground motions. The median curves as well as the 15% and 85% percentiles corresponding to the aforementioned two sets of curves are presented in Figure 10. It was noticed that after a specific spectral acceleration level, we had the onset of dynamic instability which was manifested through the inability of the program to achieve convergence, although different algorithms with various parameters were tested. The value of the acceleration after which this instability was noticed for each ground motion for the initial and for the retrofitted structure, is given in Table 2. From the above table, we notice that the average acceleration value after which instability occurs, is 0.10g and 0.18g for the initial structure and for the retrofitted structure, respectively. The above are in good agreement with the results for the maximum acceleration capacity obtained by the static nonlinear analysis presented in Figure 6. 144

4 CONCLUSIONS An innovative method is presented in this paper for the seismic retrofitting of existing concrete structures. The method is presented in detail, following all the steps of engineering analysis, from preliminary to final design. Various numerical tools are used from simple nonlinear static analysis to cyclic inelastic and incremental dynamic analysis. The results of the sophisticated analyses that were performed in this paper validate the applicability of the proposed simple model that can be used for everyday engineering practice.

ACKNOWLEDGMENTS Part of this study is framed within the research project “Innovative steel structures for the seismic protection of new and existing buildings: design criteria and methodologies” financially supported by the Italian Ministry of Education, University and Research (MIUR) for the years 2003–2005. REFERENCES De Matteis, G., Addelio, F. & Mazzolani, F.M. 2004b. Sul dimensionamento dei pannelli di alluminio puro per la protezione sismica di telai di acciaio, Proc. XI Convegno Nazionale ANIDIS L’Ingegneria Sismica in Italia. Genova. De Matteis, G., Landolfo, R. & Mazzolani, F.M. 2003b. Seismic response of MR steel frames with low-yield steel shear panels, Journal of Structural Engineering 25: 155–168. De Matteis, G., Mazzolani, F.M. & Panico, S. 2003a. Pure Aluminium shear panels as dissipative control system for seismic protection of steel moment resisting frames, Proc. Behaviour of Steel Structures in Seismic Areas (STESSA 2003): 609–614. The Netherlands: Balkema. De Matteis, G., Mazzolani, F.M. & Panico, S. 2004a. Seismic protection of steel buildings by pure aluminium shear panels, 13thWorld Conference on Earthquake Engineering (Paper No. 2704). Vancouver B.C., Canada. De Matteis, G. & Mistakidis, E.S. 2003. Seismic retrofitting of moment resisting frames using low yield steel panels as shear walls, Proc. Behaviour of Steel Structures in Seismic Areas (STESSA 2003): 677–682. The Netherlands: Balkema. Driver, R.G. & Grondin, G.Y. 2001 Sept. Shear walls now performing on the main stage, Modern Steel Construc. AISC. ENV 1993-1-5, Eurocode 3. 2003. Part 1.5: Plated structural elements, Design of steel structures. Hibbitt, Karlsson & Sorensen Inc. 2001. ABAQUS Standard version 6.1. Pawtucket, RI, U.S.A. Kulak, G.L., Kennedy, D.J.L., Driver, R.G. & Medhekar, M.S. 2001. Steel plate shear walls – an overview, AISC Engineering Journal 38(1). Mistakidis, E.S., De Matteis, G. & Formisano, A. 2004. Seismic Upgrade of Concrete Structures using Low Yield Metal Shear Panels. In B.H.V. Topping & C.A. Mota Soares (eds), Proceedings of the Seventh International Conference on Computational Structures Technology (paper 279). Civil-Comp Press, Stirling, United Kingdom. Nakagawa, S., Kihara, H., Torii, S., Nakata, Y., Matsuoka, Y. Fujisawa K. & Fukuda, K. 1996. Hysteretic Behavior of Low Yield Strength Steel Panel Shear Walls: Experimental Investigation, Proc. 11th WCEE (Paper No. 171) CD-ROM. Elsevier. Nakashima, M., Iwai, S., Iwata, M., Takeuchi, T., Konomi, S., Akazawa, T., Saburi, K. 1994. Energy Dissipation Behaviour of Steel Panels made of Low Yield Steel, Earthquake Engineering and Structural Dynamics 23: 1299–1313. Robinson, K. & Ames, D. 2000 Jan. Steel plate shear walls: Library seismic upgrade, Modern Steel Construction: 56–60. AISC. Sabouri-Ghomi, S., Ventura & Kharrazi, M.H.K. 2003. Shear analysis and design of ductile steel plate walls, Proc. Behaviour of Steel Structures in Seismic Areas (STESSA 2003): 189–195. The Netherlands: Balkema. Tanaka, K., Torii, T., Sasaki, Y., Miyama, T., Kawai, H., Iwata, M., Wada, A. 1998. Practical Application of Damage Tolerant Structures with Seismic Control Panel Using Low Yield Point Steel to a High-Rise Steel Building, Proc., Struct. Eng. World Wide (Paper T190-4). Elsevier.

145

Session 3: General methodology

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Autosystems in probability-based durability prediction A. Kudzys KTU Institute of Architecture and Construction, Kaunas, Lithuania

ABSTRACT: New methodological approaches on probability-based time-dependent reliability assessment and durability prediction of load-carrying structures are presented. The design on durability of deteriorating structures is based on the concepts of autosystems representing multicriteria failure mode of members. The long-term survival probabilities of members as components and autosystems are analyzed. The technical service life as quantitative durability parameter of autosystems is considered. Peculiarities of the target reliability indices and their effect on technical service life of autosystems are discussed.

1 INTRODUCTION A wide range of applied reliability issues of deteriorating materials due to steel corrosion, concrete degradation or wood putrefaction caused by random aggressive environmental conditions or other adverse circumstances can be neither formulated nor solved by deterministic analysis methods [CEB Bulletin 238 1997, Leira et al. 1999]. At the same, time-dependent quality of deteriorating structures of buildings and works may be closely defined only by quantitative durability parameters using available probabilistic approaches, concepts and methods [Kudzys 1992, ISO 2394 1998, Melchers 1999]. JCSS (2001) presents probability-based durability formulation taking into account stationary and non-stationary processes of material deteriorations, external actions and loads dedicating failure probability calculations. The durability prediction analysis is based on consideration of mechanical properties, action duration effects and probabilstic responses of deteriorating members [Vrowenvelder 2002, COST Action C12 2003]. But this analysis is rather complicated not only because of uncertainties of degradation processes and variable actions but also due to peculiarities of prediction models. At all cases, it is expedient to base the probabilistic durability analysis of the structure on the concepts of the intended service period [CEB Bulletin 238 1997] and the life span assessment [ISO 2394 1998, Melchers 1999]. According to these concepts, the probability-based design on durability and the time-dependent safety are in large extent similar methods. Contrary to traditional structural safety analysis, it is expedient to consider the durability prediction issues not for cross-sections, connections and other components but for whole load-carrying members representing by the autosystems of member components. The technical service life is the main durability parameter of the autosystems. This parameter is the random period of time at which the autosystems can actually perform according to the requirements based on intended purpose without major repairs at a preset target reliability index. To durability prediction problems may be ascribed an analysis on durability of non-deteriorating autosystems for which a failure probability is time-dependent random value due to random reiterated extreme loads or fatigue actions. The intention of this paper is to present some new methodological approaches on the durability prediction of structures turning an attention of researchers and engineers on merits of probabilitybased methods in the progress of building science. 149

2 RELIABILITY ANALYSIS OF MEMBERS 2.1 Time-dependent safety margin According to the probability-based approaches [ISO 2394 1998, JCSS 2000], the time-dependent safety margin or the limit state process as performance function of members may be presented as:

Here θ and X (t) are the vectors of additional and basic random variables, representing design model uncertainties and random parameters, respectively;

is the member resistance process, where vR (τ) is a rate of resistance decrease; Sg , Sq1 , Sq2 (t) and Sv (t) are the action effects caused by permanent (g), sustained (q1 ) and extraordinary (q2 ) gravity and wind, snow or earthquake (v) actions (Fig. 1). For the sake of design simplifications, the expression (1) may be presented in the form:

Here

is the conventional resistance process modelled by the normal distribution law:

is the process of extreme gravity and climate action effects described by the Type 1 distribution [Rosowsky & Ellingwood 1992, ISO 2394 1998, JCSS 2000]. The means and coefficients of variation of additional variables are: θim ≈ 1.0 and δθi ≈ 0.10.

S, P R(t) Sg+Sq

1

Rc(t) Sq (t)

Sv(t)

2

Sq

1 2 3 4

1

Sg

1 0

2

3 r

tr Tm

Figure 1. Model for time-dependent reliability analysis of deteriorating members.

150

P{Ttr}

t

2.2 Survival probability analysis The instantaneous survival probability of members at any cut k of the safety margin process, assuming that they were safe at the time less than tk , is calculated by the formula:

where fRck (x) is the cumulative density function of the resistance Rck by Equation 4; FSk (x) is the distribution function of the action effect by Equation 5. Under the action of two extreme loads a member failure may occur not only in the case of their combination but also when a level of one of two loads is extreme. So in reliability analysis three safety margin sequences are considered as follows:

where r1 , r2 and r12 are the reiteration numbers of extreme actions and their combinations during the period of time [0, tr ]. The coefficient of autocorrelation of sequence cuts is:

where Cov(Mk , Ml ) and σMk , σMl are the autocovariance and standard deviations of the safety margin at these cuts. The long-term survival probability of members is:

Here T is member lifetime as random variable. 3 RELIABILITY ANALYSIS OF STRUCTURAL AUTOSYSTEMS 3.1 Autosystems and their modelling The durability prediction analysis of load-carrying structures should be considered as a prediction of the technical service life of members as the autosystems representing multicriteria failure mode due to various actions and responses of components. The autosystems are characterized by stochastically dependent conventional elements in series, parallel and mixed connections (Fig. 2). The failure of steel joints are caused by welding stresses in parent metal and faulty welds. Foundation piles may lose a bearing capacity as geotechnical supports or compression columns (the series systems with two elements). Reaching of the limit state in any bar of two-bar hangers does not necessarily mean the failure of whole system (the parallel system with two elements). Continuous beams have two normal and one oblique design sections (the mixed system with three elements). Homogeneous in nature but different in structural reliability properties the autosystems need of different approaches in assessments of the reliability index and predictions of the technical service life. 151

(a)

(b) N

(c) N

N

N

N 2N N

R1 R2

R1

R2

R1

(d) R1

R2 R3

R2

Figure 2. Structural autosystems and their models: (a) joint of steel members; (b) single pile; (c) two-bar hanger; (d) continuous reinforced concrete beam.

3.2 Survival probability of autosystems The survival probability of the autosystems may be calculated by the simple method of transformed conditional probabilities (TCPM). For the series and parallel autosystems consisted from two elements the instantaneous survival probability may be presented, respectively, as follows:

Here Pik and Pjk are the components of survival probability calculated by Equation 6; Pi/j is the less value from two components.

is the coefficient of correlation of i and j elements of the autosystems. The long-term survival probability of autosystems PS {T ≥ tr } may be calculated by the Equations 12 and 13 changing the components Pik and Pjk into the values Pi = Pi {T ≥ tr } and Pj = Pj {T ≥ tr } defined by the Equation 11. In this case, the coefficient of correlation ρijk = ρijr characterizes the statistical relation of the safety margins at last cut r of random sequences as follows:

Thus for instance, the long-term survival probability of continuous reinforced concrete beams (Fig. 1, d) may be calculated analytically as follows:

where the survival probability of the autosystems with two parallel elements is:

The value P3 is the survival probability of oblique sections as components as components. 4 DURABILITY ANALYSIS OF AUTOSYSTEMS 4.1 Originality of technical service life The conditional failure rate of the autosystems as the risk function at time t may be expressed as:

where PS (t) is the reliability function or the survival probability for a period of time [0, t]. Increasing in the value λ(t) characterizes an intensity in decrease of the structural durability. The technical 152

100%

2 Ps(t) 1

Pmin

0

t

tt1 tt2

Figure 3. Effect of autosystem types on the technical service life with members of the same (1) and different (2) initial survival probabilities. 100%

1

P(t)

2

P{Ttt2}

P{T>tt1}

P{Tt}in Pmin

tin

t

tpr tt2 tpr

tin tt1

Figure 4. Reliability functions P(t) versus the life tt for autosystems exposed to permanent (1) and extreme (2) variable actions effects.

service life tt as quantitative parameter represents the structural reliability of the autosystems more objectively than deterministic ultimate state verifications of member components. The predicted durability value tt is in close relation to the target survival probability of structures Pmin which may be presented as:

Here  is the cumulative distribution function of the standardized normal distribution; βmin is the target reliability index. The index βmin must be calibrated taking into account not only three reliability classes [EN 1990 2002] and economic optimization factors [ISO 2394 1998] of structures but also three functional working classes [Kudzys et al. 2004] and types of the autosystems. Easily repairable members (slabs, columns, walls) belong to 1-st, not easily repairable (beams, foundations, joints) to 2-nd and not repairable (piles) to 3-th working classes. A necessity to base a prediction of the technical service life tt of structural members on the autosystem models is illustrated in Figure 3. It is not difficult to be convinced of great effect of types and initial survival probabilities of the autosystems on the value tt . These two facts are fatal in the probability-based durability prediction. 4.2 Technical service life prediction It is expedient to divide life cycles of deteriorating structural autosystems into the initiation and propagation periods of mechanical degradation processes [Kudzys 1992, JCSS 2000]. The propagation period tpr is late for metals which are protected by concrete covers in reinforced concrete structures or antirust coats in steel ones. A length of the initiation period tin is random variable depending on a quality of protected covers or coats and environment aggressiveness. At all cases, the technical service life tt of the autosystems depends on the intensities and uncertainties of actions (Fig. 4). 153

Technical service life values of the autosystems may be computed by numerical integration, Monte Carlo simulation and analytical methods using the Equation 16 and the methodological concepts are presented in this paper. The parameter tt may be calculated from this equation using the survival condition as follows:

The computation is iterated until the value tt corresponds the target probability Pmin by Equation 19. 5 CONCLUSIONS The durability prediction as one of the main design task is indispensible in order to guarantee the time-dependent performance of deteriorating structures and avoid unexpected accidents. Design on durability of structures should be based on the concepts of autosystems representing multicriteria failure mode of load-carrying members. It helps to make the technical service life of members and their rational design working life for which the structure is to be used with balancing reliability indices more exactly. The design on durabilty of the autosystems needs of different approaches in the target reliability indices taking into account not only reliability classes and economic optimization factors of structures but also the types and the functional working classes of the autosystems of deteriorating members. REFERENCES CEB Bulletin 238. 1997. New approach to durability design. Stuttgart: Sprint-Druck. COST Action C12. 2003. Improvement of buildings structural quality by new technologies. Proc. inter. sem., Lisbon, 19–20 April 2002, EUR 20728. Luxemburg. EN 1990:2002E. 2002. Eurocode – Basis of structural design, CEN. Brussels. ISO 2394:1998E. 1998. General principles on reliability for structures. Switzerland. JCSS. 2000. Probabilistic Model Code: Part 1: Basis of Design. Joint Committee on Structural Safety. Kudzys, A. 1992. Probability Estimation of Reliability and Durability of Reinforced Concrete Structures. Vilnius: Technika. Kudzys, A., Kvedaras, A. K., Kliukas, R. 2004. Methodological approaches on structural durability assessment. Probabilistic Safety Assessment and Management; Proc. PSAM7-ESREL 04, 2167–2173. Berlin: Springer. Melchers, R.E. 1999. Structural reliability analysis and prediction, 2nd ed. Chichester: John Wiley & Sons. Rosowsky, D. Ellingwood, B. 1992. Reliability of wood systems subjected to stochastic live loads. Wood and Fiber Science, V. 24, N◦ 1, 47–49. Vrowenwelder, A. C. W. M. Developments towards full probabilistic design codes. Structural Safety, V. 24, N◦ 2–4, 2002, 417–432.

154

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Real-time model for earthquake prediction S. Radeva University of Architecture, Civil Engineering and Geodesy, Sofia, Bulgaria

D. Radev University of Rousse, Rousse, Bulgaria

ABSTRACT: The paper is devoted on the problem of real-time earthquake prediction. Different kinds of stages of earthquake prediction are observed, and implementation of existing algorithms M8 and MSc for intermediate-term middle-range prediction for is discussed. An approach for real-time prognoses, based on neural network and vector quantization is suggesting. As input information for the neural network are given the parameters of recorded part of accelerogram, principle axis transformation and spectral characteristics of the wave. With the help of stochastic long-range dependence time series analyses and scene oriented model are determined the boundaries of destructive phase of strong motion acceleration. For selected diapason of transformed accelerograms is implemented one-dimensional and two-dimensional vector quantization. With self-organized map are determined weight centers of selected classes. The prognoses are realized with the help of neural network, learned and trained to optimize selected target classes and determine probability density function. 1 INTRODUCTION A very promising method in earthquake engineering for protection of height – risk and very important structures against destructive influence of seismic waves is anti-seismic structural control. One of the critical problems there is the problem of forecasting in real-time of the behavior of seismic waves. Prognoses for further development of the waves can be made from recorded in real-time data for certain part of destructive seismic wave’s registrated in three directions. These prognoses are based on classification of strong motion seismic waves made on general, tectonic, seismic and site parameters. During these prognoses is supposed that waves can be classified as destructive or non-destructive and can be taken decision for switching on the devices for structural control (Radeva et al., 2004). Another application of classification and real-time prognoses of development of seismic waves is prognoses of further seismic activity after primary ones. For making such prognoses it is necessary to develop different kind of models. Modeling gives possibility to study the behavior of seismic waves and relationships between their parameters during their spread in soil layers, where for each point the parameters of her displacement are presented with three components in three directions of the orthogonal axes. For practical purposes of possible records for displacements, velocities and accelerations as time history, most often accelerograms are used, which are characterized with certain duration, frequency and peak ground acceleration. They are involved in models and systems for estimation of elasticity response spectrum. The most practical usage in structural engineering and design has their peak values, independently of their sign and direction. That’s why the modeling of the behavior of seismic waves is used as input information in the process of calculation of the structural response spectrum. Nowadays increase implementation of artificial intelligence methods for describing the behavior of seismic waves. Most of them are based on neural networks for analyzing of earthquake records, which are trained, with real strong motion seismic records. Other models are based on the fact 155

that crisp values as earthquake parameters can be successful described with the help of fuzzy logic models (Radeva & Radev, 2003). One of the very promising trends is creating models, which combines different approaches like Neuro-Fuzzy models, models combining stochastic and artificial intelligence approach etc. The main purpose of this work is to develop an approach for quick estimation and prognoses of characteristics of destructive phase of strong motion seismic waves with combined stochastic and neural modeling.

2 PROGNOSES CLASSIFICATION AND EXISTING PROGNOSES ALGORITHMS The prognoses of earthquake occurrence can be classified according to the prognoses time duration. The most popular is their dividing into long-term (for next ten years), intermediate-term (for next few years), short-term (for next months-weeks) and real-time. Other kind of prognoses is prognoses of the area of occurrence of earthquake excitation of certain magnitude. Both approaches for prognoses are connected with difficult problems, when are applied the traditional stochastic time series analyses instead of applying methods for crash prognoses, where is dealing with reaching of certain critical threshold (Kossobokov et al., 2000). Existing algorithms M8 and MSc are developed for intermediate-term middle-range prediction. The prediction algorithm M8 is designed for intermediate-term earthquake prediction by retroactive analysis of dynamics of seismic activity preceding the greatest, magnitude 8.0 or more earthquakes worldwide. The prediction algorithm MSc (Mendocino Seismic catalog) is designed by the retroactive analysis of the detailed regional seismic catalog prior to Eureca earthquake (1980, M7.2) near Cape Mendocino California. The phenomenon, which is used in the MSc algorithm, might reflect on second (possibly shorter term, and definitely narrow range) stage of the premonitory rise of seismic activity near the incipient source of main shock. This prediction method was designed by retroactive analysis of dynamics of seismic activity preceding the greatest, magnitude 8.0 or more earthquakes. Its original version (Keilis-Borok & Kossobokov, 1900) were tested retroactively at 143 points, of which 132 are recorded epicenters of earthquakes of magnitude 8.0 or greater from 1857–1983. The catalog of main shocks can be described by {ti , mi , hi , bi (e)}, i = 1, 2,…, where ti is the origin time, hi is the focal depth, mi is the magnitude and bi (e) is the number of aftershocks with magnitude Maft or more during the first e days. According to M8, the prediction is aimed at earthquake of magnitude M0 and larger from the range M0 += [M0 , M0 + DM ], where DM < 1. The magnitude scale should reflect the size of earthquake sources. The algorithm is realizing via calculating overlapping circles, with diameter D(M0 ) = (exp(M0 − 5.6) + 1)◦ in degrees of the Earth meridian, scanned seismic region under study. The received sequence of overlapping circles is normalized by the lower magnitude cutoff m = Mmin (N˜ ), N˜ being the standard value of the average annual number of earthquakes in the sequence. Several running averages are computed for this sequence in the trailing time window (t − s, t) and magnitude range M0 > Mi3 m. They depict different measures of intensity in earthquake flow, its deviation from the long-term trend and clustering the earthquake. The averages determine range and acceleration of activity. The algorithm recognizes criteria, defined by extreme values of the phase space coordinates, as a vicinity of the system singularity. When a trajectory enters the criteria, probability of extreme event increases to the level sufficient for its effective provision. The MSc algorithm outlines such an area of the territory of alarm where the activity, from the beginning of seismic inverse cascade recognized by the first approximation prediction algorithm is continuously high and infrequently drops for a short time. The phenomenon, which is used in MSc might reflect on second stage, where the seismic activity may rise near the incipient source of main shock. Let given a TIP diagnoses for a certain territory U at the moment T , the algorithm attempt to find within U a smaller area V , where the predicted earthquake can be expected. The algorithm 156

requires a reasonably complete catalog of earthquakes with magnitudes M = 3 (4), which is lower than the minimal threshold used by M8. The territory U is coarse-grained into small squares (s × s) and determines the centers of the squares (i, j). Within each square the number of earthquakes and aftershocks is calculated for consecutive, short time windows and k is the sequence number of a trailing time window. Finally, the time-space considered is divided into small boxes (i, j, k) of the size (s , s , u). “Quiet” boxes are singled out of each small square with coordinates of its centre (i, j) and clusters of q or more quiet boxes connected in space or in time are identified. Area V is territorial projection of these clusters. The prediction is localized to a spatial projection of all recent “sufficiently large” clusters of squares being in state of “anomalous quiescence”. The standard values of parameters are u = 2 months, q = 4 and s = 3D/16, where D is diameter of the circle used in algorithm A8. The algorithm MSc outscores simple alternatives of narrowing down the area of first approximation alarm, in which are included “nonempty cells” and “most active cells” that contain a part of recent seismic activity. At second approximation is improving accuracy by more detailed determination of the “weight centers” of the squares. 3 DETERMINING OF DESTRUCTIVE PHASE BOUNDARIES FROM ACCELEROGRAM The purpose of stochastic modeling for this investigation is the defining of the three phases of the earthquake wave and identification of the main parameters for each phase, such as resonance frequency, damping ratio, peak value. According to implemented stochastic model, each wave is dividing into three separated phases: primary (P-waves), transversal or secondary (S-waves) – on the second phase, and converted and guided waves (C-/G-waves) on the third phase. For evolutionary power spectrum estimation were used the time dependent stochastic principal axes method (Scherer and Zsohar, 1998). According to this method earthquake accelerograms are delivered as representations of the three-dimensional acceleration vector in a Cartesian coordinate system, generally with axes parallel to east–west, north–south and vertical direction, as shown on Fig. 1, where is presented registrated records from Loma Pierta Earthquake, 1989. On Fig. 1 is shown as well determining the boundaries of three separated phases, which is realized with scene-oriented model. For determining boundaries between separated classes were analyzed 4300 strong motion seismic records, registrated in Europe and North America and to this records were implemented different stochastic models. We are suggesting the scene-oriented model as best fitting for determining boundaries of destructive S-phase. The scene-oriented model is a modification of simple Markov chain model, where the time series {xt } was transformed into discrete states {yt }, where the number of states is the same as the number of target classes, and the size of the model yi for each state is determined. At the scene-oriented model as three scenes are separated the three phases of the seismic waves. Consider the S-phase as a second scene. The target values in the classes of the second scene are determined with (1).

The next step is forecasting the resonance frequency of S-wave on the base of prognoses made with the help of principle axes transformation and further estimation of probability density with neural network and vector quantization. 4 PROGNOSES ZONE ON THE BASE OF PRINCIPLE AXES TRANSFORMATION Principle axes transformation is based on composing the components corresponding to the maximum, medium and minimum eigenvalue from all time windows will result in accelerogram time 157

1 0 0.5

S phase C/G phase

0 -0.5 -1

P phase

Acceleration (m/sec/sec)

0

10

20

30

40

50 Time (sec)

10

20

30

40

50 Time (sec)

10

20

30

40

50 Time (sec)

90 1 0 -1 0 0.5 UP 0

-0.5 0

Figure 1. Recorded accelerograms with axes parallel to east–west, north–south and vertical direction.

histories that are ordered by seismic energy for every chosen time interval (Scherer & Zsohar, 1998). These transformed components are called the stochastic principal axis accelerogram T1, T2 and T3, presented after the original ones in three directions on Fig. 2. This process of time-series transformation gives possibility to use for empirical hazard analyses T1 and T2. For searching the most dominant and energetic component for every phase were used window parameters and principal axes transformation. For a certain given time t0 time delay τ and window length L the cross-covariances are presented in equation (2) and would consist of statistically independent components in selected intervals (Scherer & Bretschneider, 2000).

Principle axes transformed accelerograms can be visualized in the coordinate system of the original record. For the covariance matrix of the three components of an acceleration record, its squared variances of the respective components and transformation of the 3D accelerogram into the coordinate system of the eigenvectors of the covariance matrix yields an accelerogram with statistically independent components. As most significant in seismic hazard analyses, for vector quantization and determination of basic classes further are used accelerograms T1 and T2. 5 ONE-DIMENSIONAL VECTOR QUANTIZATION FOR T1 AND T2 BASIC CLASSES The values of transformed accelerograms T1 and T2 can be presented as a couple of time series {x(1) } and {x(2) }, for which the probability distribution is determining with the help of vector quantification (VQ). We can associate the clustering problem with a k dimensional discrete distribution with (1) (1) (2) (2) independent margins, where the random values X1 , . . . , Xk and X1 , . . . , Xk are distributed 158

T1

S phase

1 0 -1 Acceleration (m/sec/sec)

0 1

T2

6.1

11.6

20

30

40

50 Time (sec)

20

30

40

50 Time (sec)

20

30

40

50 Time (sec)

S phase

0.5 0 -0.5 0 0.4

6.2

11.56

T3

S phase

0.2 0 -0.2 0

6.3 29.16

Figure 2. Determining of S-phase boundaries for principle axes transformed records.

into M classes S1 ,…, SM (Radeva & Radev, 2003). Further we consider only selected S-phase for transformed accelerograms T1 and T2. With one dimensional VQ of is determined number of classes for destructive phase analyses, where values about 276 points of T1 and T2 are in diapason from 6.1 to 11.6. One realization with 6 classes for T1 with equal number of target values distributed in each class is shown on Fig. 3. The values from T1 have higher weight in prognoses than values from T2, which are distributed into 4 classes, and received model of VQ for T2 is shown in Fig. 4. Further, determining the number of target classes, which will be used in further prognoses, can be optimized. Moreover, by using a two-dimensional VQ, neural structure with learning rules, and adjusting of target classes boundaries (Radeva et al., 2004), we suggest a solution where the weight centers of target classes could approximate the Markov chain. 6 PROBABILITY DENSITY ESTIMATION AND 2D VECTOR QUANTIZATION For probability density estimation of destructive phase in this research is suggesting a modification of two-dimensional VQ, where on axes are absolute values of T1 and T2. Afterward with one layered neural network and self-organizing map (SOM) (Kohonen, 1997) is determined the function of density distribution with amplitudes, received from transformed accelerograms. Self-organizing neural networks have one-layered neural competitive structure, which can learn to detect regularities and correlations in the input patterns. The neural maps learn both, the distribution and topology of the input vectors, to recognize neighboring clusters of the attribute space. Kohonen’s network algorithm provides a tessellation of the input space into patches with corresponding code vectors. It has an additional feature that the centers are arranged in a low dimensional structure (usually a string, or a square grid), such that nearby points in the topological structure (the string or grid) map to nearby points in the attribute space. The Kohonen learning rule is used when the winning node represents the same class as a new training pattern, while a difference in class between the winning node and a training pattern causes the node to move away from training 159

Acceleration, m/s/s

T1 VQ Model 150 class 6 100

54.1

class 5

31.36 class 4 -7.87

class 3

-31.32

class 2

-48.03

class 1 -100 6.5

7

7.5

8

8.5

9

9.5

10

10.5

Time, s

11.5

Acceleration, m/s/s

Figure 3. One-dimensional vector quantization for transformed accelerograms T1. 100

T2 VQ Model

class 4

22.82

class 3 -5.14 class 2 -21.3

class 1 -50 6.5

7

7.5

8

8.5

9

9.5

10

10.5

Time, s

11.5

Figure 4. One-dimensional vector quantization for transformed accelerograms T2.

pattern by the same distance. In training, the winning node of the network, this is nearest node in the input space to a given training pattern, moves towards that training pattern. It drags with its neighboring nodes in the network topology. This leads to a smooth distribution of the network topology in a non-linear subspace of the training data. 160

102 T2

VQ mod class 12 class 11 class 10

101

101

102

T1

Figure 5. Two-dimensional vector quantization in 12 classes with self-organizing map (SOM).

In two-dimensional output space is expected a map, corresponding to the k-dimensional array of output neurons Cj , which can be one or two-dimensional. The connection between n-dimensional input vector and k-dimensional output neural vector is realized with the weight matrix W. At competitive learning for winner is selected the output neuron j ∗ , which weight vector is closer to the current input according to (3).

The Kohonen learning rule is differ from vector quantization rule and is determined by (4).

The neighborhood function ∧( j, j*) is equal to 1 if j = j*, and decreases with increasing of distance between neurons j and j* in input space. The neurons closer to the winner j*, changes their weights more quick than remote neurons, for which the neighborhood function is very small. The topological information contents in the fact, these closer neurons, which are changing almost in the same way and in this manner corresponds to neighbor input patterns. The learning rule (4) attracts the winner’s weight vector to the point Xm . The self-organizing map is supposed to be an elastic set in input space, which wants to be moved maximal closer to the input values. The set has topology of attribute space and it points have as coordinates weight vectors. Here is suggesting a modification of VQ, with implementation of logarithmic scale and absolute values for T1 and T2. On Fig. 5 is shown modified vector quantization where with black points are 161

PD, % 35.00

32.18

30.00 25.00

0.09

2.56

7

7.01

13.80

6

12.90

13.44

7.69

5.00

15.05

10.00

15.62

15.00

25.53

28.19

20.00

0.00 1

2

3

4

5

8

9

10

11

12

Class

Figure 6. Probability density estimation for destructive phase.

depicted weight centers of target classes. We are interesting of last three classes (10, 11 and 12), because for them is observed higher deviation. With Manhattan distance are determined deviations from trajectory of axis and points in corresponding class according to (5).

On Fig. 6 is shown an estimation of probability density distribution in each class for S-phase. The number of classes can be optimized in dependence of object or construction for which is implemented this approach. Received results can be used for analyses of structural response spectrum and in devices of structural control for very important and high-risk structures.

7 CONCLUSIONS AND ACKNOWLEDGEMENTS An approach for real-time prognoses, based on classification algorithm of strong motion waves with neural network and vector quantization is suggesting. On the base of principle axis transformation and spectral characteristics of the wave, with stochastic long-range dependence time series analyses are determined the boundaries of destructive phase of strong motion acceleration. For selected diapason of transformed accelerograms is implemented one-dimensional and two-dimensional vector quantization. The prognoses are realized with the help of one-layered neural network, learned and trained with Kohonen rules. Two-dimensional vector quantization is presented as self-organizing map and probability density function for destructive phase is determined. The received prognoses of destructive phase of strong motion waves can be used in devices for structural control. Simulation and numerical results are shown. This work is a part of the International NATO research project No: PST.CLG.979333. REFERENCES Keilis-Borok, V.I. & Kossobokov, V.G. 1990. Premonitory activation of earthquake flow: algorithm M8. Phys. Earth Planet. International 61: pp. 73–83. Kohonen, T. 1997. Self-Organizing Maps, Second Edition, Springer Verlag, Berlin.

162

Kossobokov, V., Keilis-Borok, V., Turcotte, D. & Malamud, B. 2000. Implications of a statistical physics approach for earthquake hazard assessment and forecasting. Pure Applied Geophysics 157: pp. 2323–2349. Radeva, S., Scherer, R., Radev, D., Yakov, V. 2004. Real-time estimation of strong motion seismic waves, Acta Geodaetica et Geophysica Hungarica, Vol. 39 (2–3), pp. 297–308. Radeva, S., Radev, D. 2003. Analyze of the Destructive Phase of Seismic Waves with Vector Quantization. Proceedings of the Conference Automatics and Informatics, Sofia, pp. 83–86. Scherer, R., Zsohar, M. 1998. Time-Dependent Stochastic Principle Axes as the Basic Tool to Derive General Shape Functions for Non-Stationary Power Spectra. Proc. of the 11th ECEE, Paris, pp. 116–128. Scherer, R.J., Bretschneider, J. 2000. Stochastic Identification of Earthquake Wave Entities. Proceedings of the 12th WCEE 2000, Auckland, New Zealand, Paper No 2436/4/A.4.

163

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Nonlinear response analysis of different retrofit strategies F. Aras University of Anatolia, Eskisehir, Turkey

G. Altay University of Bogazici, Istanbul, Turkey

ABSTRACT: The major objective of this study is to evaluate different retrofit strategies for a typical, existing reinforced concrete building in Istanbul under a non-linear response methodology and performance based design procedure. Calculation of the capacity spectrum of the system and demand spectrum of the design earthquake and the resulting performance points based on FEMA-356 and ATC-40 documents, using SAP2000. Evaluated building is found to be inadequate in terms of strength and stiffness demands of the design earthquakes as defined in the current Turkish Earthquake Code. Capacity curves and performance levels of the building retrofitted with shear walls and concentric braces are obtained and results are discussed. A two-dimensional frame is retrofitted by eccentric braces, concentric braces and steel plates. Ductility and seismic load reduction factors are calculated based on ATC-19. A preliminary cost analysis is also performed and compared for the retrofit strategies used.

1 INTRODUCTION In recent years significant efforts have been devoted to various strengthening techniques to improve the seismic performance of reinforced concrete frame members and structures. While many of these techniques can efficiently improve the lateral stiffness and resistance of an existing structure, adequate seismic behavior will be obtained only if the retrofitted structure can satisfy the strength and ductility demand imposed by the earthquake. Performance based design is one of the most powerful methods that evaluates the stiffness and strength demands of the earthquake. Created by FEMA-356 and ATC-40, performance based design has some superiorities over the linear analysis. The purpose of this paper is to present the result of an analytical study to investigate seismic behavior of a system retrofitted with steel braces, steel plates and reinforced concrete (R/C) shear walls. The performance of the original building and performance of the buildings retrofitted with different strategies are compared. Benefits and inadequacies of each system are identified and possible design strategies are presented. 2 CASE STUDY A 5-story reinforced concrete building is selected to investigate the seismic performance of different strategies as an example study. It is an existing building in Istanbul, which lies on the first zone according to Turkish earthquake code. The building consists of one basement, ground floor, and three ordinary residence floors. The story height is 2.75 m for the basement and the ordinary floor, while it is 3.6 m for the ground floor. The beam depth is 60 cm and the width is 20 cm through the whole building. The column dimensions are written in the plan view, and they are the same trough 165

Figure 1. Plan of the building, investigated.

Table 1. Components’ hinge properties. Hinge name

a

b

c

IO

LS

CP

Column 1 Column 2 Column 3 Column 4 Column 5 Column 6 Beam 1 Beam 2 Beam 3 Beam 4 Beam 5 Beam 6

0.02 0.016 0.014 0.013 0.012 0 0.025 0.022 0.02 0.018 0.017 0.015

0.03 0.024 0.022 0.021 0.02 0 0.05 0.04 0.04 0.03 0.03 0.02

0.2 0.2 0.2 0.2 0.2 0 0.2 0.2 0.2 0.2 0.2 0.2

0.005 0.005 0.004 0.004 0.003 0 0.01 0.008 0.005 0.005 0.005 0.005

0.015 0.012 0.011 0.011 0.011 0 0.02 0.015 0.01 0.01 0.008 0.0051

0.02 0.016 0.014 0.013 0.012 0 0.025 0.022 0.02 0.018 0.017 0.015

the height of the structure. Specified materials of the building are C25 and S420. Figure 1 shows the plan of the building. 2.1 Modeling and acceptance criteria To perform the non-linear pushover analysis, hinge properties are determined They contain the plastic rotation values that a component’s end can sustain and the acceptable plastic rotation values for the performance levels, immediate occupancy (IO), life safety (LS), and collapse prevention (CP). In Table 1 a, b, c values are the plastic rotation values, defined in ATC-40 and FEMA-356. In these publications hinge properties are given according to component type and failure mechanism. Failure mechanism of all components is assumed as flexural failure. Since all components are confined properly and tie spacing of all components are sufficient to provide conforming section features, shear failure cannot occur. All the necessary values to determine the hinge parameters are obtained from SAP2000. For the values that do not coincide with any specified values in the tables linear interpolation is performed. After the necessary computations, six kinds of hinge properties are determined for the beams and the columns. In this process FEMA-356 is used. Table 1 shows the hinge properties of the building components. These hinge properties are assigned to their components properly. In this study, fivepercent of the component length from the edges is assumed as hinge locations. In other words, the hinges may develop at the defined places. 166

Figure 2. Capacity curve of the building and capacity spectrum method in the x direction.

Figure 3. Capacity curve of the building and capacity spectrum method in the y direction.

Since the site geotechnical characteristics are not known well, type SD is used for the soil profile type, as advised in ATC-40. For the construction of the elastic response spectra, CA is taken as 0.44 and CV is taken as 0.64, due to soil features and the shaking intensity. 2.2 Interpretations of pushover results After SAP2000 performs the pushover analysis in both x and y directions capacity curves in base shear and roof displacement format, capacity and demand spectrum in spectral acceleration and spectral displacement format can be obtained. Figure 2 shows the capacity curve of the building and the capacity spectrum method in the x direction. As it is seen the building has not enough capacity to the earthquake demand. Figure 3 is the capacity curve of the building and capacity spectrum method in y direction. It is seen that the building is also weak in y direction. Since the capacity spectrum does not intersect the demand spectrum, the performance point can not be determined. Deformed shape of the building shows that the soft story mechanism exists in y direction as well. When the building has the maximum top displacement, the hinges occur on the ground floor columns’ end. The most of the hinges are in the life safety performance level, and there are a few hinges that are in the collapse performance level. Although some hinges are developed in the basement, their performance levels are not worse than the ground floor columns’ hinge. For this reason the critical, weak part of the structure seems to be the ground floor. The story height of the ground floor is already more than the other stories’ height. After deciding that the building is not capable to resist the earthquake ground motion, two retrofit ways are chosen to strengthen the building. These strategies are addition of R/C shear walls and steel braces. The new components are added inside of the building. The architectural reasons are taken into account in order not to impair the use of the building. 2.3 Retrofit of the building by R/C shear walls One of the most common ways to improve the seismic behavior of a reinforced concrete structure is addition of reinforced concrete shear walls. R/C shear walls are added to a structure in a way 167

Figure 4. Retrofit plan of the building.

Figure 5. Pushover curve of the retrofitted building by R/C shear walls and capacity spectrum method in the x direction.

Figure 6. Pushover curve of the retrofitted building by R/C shear walls and capacity spectrum method in the y direction.

that not to cause the irregularities, such as torsional irregularity, weak story irregularity and soft story irregularities. For this reason the addition of the R/C shear walls requires careful engineering judgment. Retrofitted building is shown in Figure 4. The R/C shear walls were proportioned according to the current Turkish earthquake code. The minimum reinforcement values were used. The addition of the shear walls increases the building weight to 9176 kN. Seismic load reduction factor is increased to five from four. Pushover analysis of the retrofitted building is performed under revised conditions and significant strength and stiffness increase have been observed in the pushover curves. Figure 5 illustrates the capacity curve of the retrofitted structure by the shear walls and capacity spectrum method in the x direction. The maximum base shear increased to 9415 kN from 2072 kN for the pushover analysis in the x direction. The period of the retrofitted structure decreased to 0.5539 second. The original building’s period was 0.7163 second. Any decrease in the period value indicates the stiffness increase in the building. Since the building has enough 168

Table 2. Section details of the steel braces.

Figure 7. Capacity curve of the retrofitted building by steel braces and capacity spectrum method in the x direction.

Figure 8. Capacity curve of the retrofitted building by steel braces and capacity spectrum method in the y direction.

strength its performance point can be found in the ADRS format. In this performance point the building remains stable and it can provide the immediate occupancy performance level. The hinge locations in the building propagate to all structural frames. The performance point is found at 5936 kN base shear and 5.45 cm top displacement. In the ADRS format, the performance point is on 0.917 Spectral Acceleration/g and 4.101 Spectral Displacement. The effective damping ratio is 0.074. Figure 6 shows the capacity curve of the retrofitted building and capacity spectrum method in y direction. It is seen that; the maximum base shear increased to 9436 kN from 2788 kN. The maximum displacement that the building reach is increased to 12.33 cm. The performance point is found at 6303 kN base shear and 7.68 cm top displacement. In the ADRS formant, the performance point lies on 0.901 Spectral Acceleration/g and 5.935 Spectral Displacement. The effective damping ratio is 0.086. 2.4 Retrofit of the building by steel braces Secondly the building is retrofitted with steel braces. The braces are added to the same location as the R/C shear wall. Thereby the comparison of the effectiveness of the systems is simplified. Rectangular box sections are used for braces. The section details are illustrated in Table 2. All required stress checks are done according to Turkish standards and X bracing is used. Figure 7 shows the capacity curve of the retrofitted building by steel braces and capacity spectrum method in x direction. In this direction the maximum base shear is 5642 kN, and the maximum top 169

Figure 9. Capacity curve of the frame, capacity spectrum method and deformed shape of the frame on the performance point.

displacement is 12.08 cm. Significant strength and stiffness increase are observed in that direction. The performance point is found at 5163 kN base shear and 10.44 cm top displacement. In the ADRS format, the performance point is on 0.762 Spectral Acceleration/g and 8.057 Spectral Displacement. The effective damping ratio is 0.118. In the y direction maximum base shear is 7067 kN, and the maximum top displacement is 16.84 cm. The performance point is found at 5206 kN base shear and 8.91 cm top displacement. Acceleration/g and 4.61 Spectral Displacement. The effective damping ratio is 0.089. On the determined performance points for x and y direction, the building is sufficient to supply immediate occupancy performance level. Soft story mechanism is not observed. Initial period of the building decreased to 0.6279 second from 0.7163 second. 2.5 Analysis of the frame A representative frame from the building is selected and analyzed. D axis of the building seems to be most appropriate axis in order to reflect the common features of the building. All modeling and acceptance criteria were determined on the three dimensional analysis previously mentioned. Therefore modeling and acceptance criteria of all components are available. Figure 9 illustrates the pushover curve, capacity spectrum method and deformed shape of the frame in the performance point. Maximum base shear that the frame can resist is 555 kN, and the maximum displacement is 8.45 cm. The performance point of the frame is found on 544 kN base shear and 7.8 cm top displacement. As seen, the performance point is close to maximum base shear and the maximum displacement values. On the performance point, all ground floor columns have plastic hinges. Three columns are in the life safety performance level and one is in the collapse performance level. For this reason the frame is assumed to be in the collapse performance level hence the frame should be retrofitted by appropriate strategy. 2.6 Retrofit of the frame by steel braces The frame first is retrofitted with steel plates. Steel plates are modeled as inclined steel strips. The technique was presented originally by Thorburn et al. The strips are assigned an area equal to the plate thickness multiplied by the width of the strip. The recommended inclination angle is 45◦ to use, since its results reflect the actual behavior (Lubell, Prion, Ventura & Rezai). Three millimeter steel plate thickness is used to strengthen the frame. Each plate is modeled by 17 steel strips. Figure 10 shows the retrofit scheme, obtained capacity curve of the frame and application of capacity spectrum method. As understood from capacity curve and capacity spectrum method, addition of the steel plates causes significant strength and stiffness increase in the frame. Maximum base shear has increased to 6502 kN from 555 kN. The performance point of the frame has been determined on 2854 kN base shear and 2.2 cm top displacement. There are just a few hinges that take form when the frame has performance point’s displacement and the base shear. These hinges rotation values are even smaller than the immediate occupancy 170

Figure 10. Steel plate retrofit scheme, obtained capacity curve of the frame and application of capacity spectrum method.

Figure 11. Concentric steel brace retrofit scheme, obtained capacity curve of the frame and application of capacity spectrum method.

performance levels rotation. Initial period of the frame decreased to 0.2177 second from 0.5017 second. 2.7 Retrofit of the frame by concentric steel braces The frame is retrofitted by concentric steel braces. Braces are put in the middle bay of the frame. The cross-sections of the braces are determined with the same procedure as in three-dimensional model. Box section is preferred, and 8 cm box height is found adequate. After the pushover analysis performed, the building is found adequate to resist the defined earthquake. Figure 11 illustrates concentric steel brace retrofit scheme, obtained capacity curve of the frame and application of capacity spectrum method. Maximum base shear is 2780 kN and the maximum top displacement is 14.51 cm. Initial period of the frame decreased to 0.3151 second. Performance point is on 1042 kN base shear and 3.52 cm top displacement. Performance level of the frame retrofitted by concentric steel braces is measured as immediate occupancy performance level. 2.8 Retrofit of the frame by eccentric steel braces Finally the frame is retrofitted by eccentric steel braces. In each storey the middle beams have been divided to three parts, since three is another beam in the perpendicular direction. For this reason, the joint is the most suitable place that the brace should be connected. The braces are designed as box section. Because of the buckling and stress problems on the basement and ground floor 10 cm height was chosen. For the upper storey eight centimeter box height is selected. The maximum base shear value increase to 2033 kN and 16.92 cm top displacement has been obtained. The initial period of the frame decreased to 0.3894 second. Figure 12 shows the capacity curve and capacity spectrum method for the strategy. The performance point of the frame is found on 967 kN base shear and 5.2 top displacement value. The frame supplies the immediate occupancy performance level. 171

Figure 12. Eccentric steel brace retrofit scheme, obtained capacity curve of the frame and application of capacity spectrum method.

Figure 13. Comparison of the capacity curves for the building in x and y direction.

7000 Steel plate 6000

Base shear (KN)

5000 4000 Concentric brace

3000 2000

Eccentric brace 1000 Original Building 0 0 -1000

2

4

6

8

10

12

14

16

1

Displacement (cm)

Figure 14. Comparison of the capacity curve of different retrofit strategies in the frame.

3 COMPARISON OF RETROFIT STRATEGIES IN THE BUILDING In order to compare the effectiveness of the systems the pushover curve of the systems are plotted in the same graph. Figures 13 and 14 shows the pushover curves of the retrofitted building by different systems in the x and y direction respectively. One of the most important criteria in the comparison stage is the number of the hinges developed on the performance point. Period of the structure is another key to interpret the effectiveness of the 172

Table 3. Comparison of the retrofit strategies in the building. Number of hinge on performance point Strategy name Shear wall Steel plate

Direction

Performance point (V;D) (KN;cm)

IO

LS

CP

x y x y

5936;5.45 6303;7.63 5163;10.4 5206;8.91

70 120 136 128

– – – –

– – – –

Period (second) 0.5539 0.6279

Table 4. Comparison of the retrofit strategies in the frame. Number of hinge on performance point Retrofit system

Period (second)

Performance point (V;D) (KN;cm)

IO

LS

CP

C

Original frame Steel plate Concentric braces Eccentric braces

0.5017 0.2177 0.3151 0.3894

544;7.9 2874;2.2 1042;3.5 967;5.2

2 6 6 9

6 – – –

– – – –

2 – – –

strategies. Table 3 shows the number and type of the developed hinges on the performance point and the period of the structure. 4 COMPARISON OF RETROFIT STRATEGIES IN THE FRAME Figure 14 shows the pushover curves of the frames. Table 4 shows the hinge performance level and numbers, period and the performance point of the frames to summarize the comparison. After the frame retrofitted by the systems, it is capable to resist the earthquake and provide immediate occupancy performance level. Number of the hinges is different for each strategy. 5 DETERMINATION OF SEISMIC LOAD REDUCTION FACTOR In order to see the non-linear behavior of the systems, the seismic load reduction factors are calculated. The determination of the seismic load reduction factor, R is explained in ATC-19. R is expressed as the product of three factors, namely, period-dependent strength reduction factor (Rs ), period-dependent ductility reduction factor (Rµ ), and redundancy factor (RR ).

Table 5 has been formed to summarize computation. As advised in ATC-19 inelastic displacement is assumed equal to 1.5 times the elastic displacement. 6 BASIC COST ANALYSIS In order to see the cost aspect of the strategies, cost of the materials that added to the structure is calculated. Table 6 shows the added materials to the structures and their cost values. It must be remembered that, two different structures have been retrofitted so comparison of strategies should be done separately. 173

Table 5. Calculation of seismic load reduction factor for different strategies. Strategy name

Vb (kN)

Vu (kN)

T (second)



Rs

Rµ

R

Building Original building-x Original building-y Shear wall-x Shear wall-y Steel brace-x Steel brace-x

1162 1740 1540 1807 1560 2347

2000 2750 8500 6750 4650 5150

0.7163 0.7163 0.5539 0.5539 0.6279 0.6279

0.817 0.817 0.965 0.965 0.888 0.888

1.721 1.58 5.519 3.735 2.981 2.194

1.612 1.612 1.518 1.518 1.563 1.563

2.775 2.548 8.379 5.671 4.659 3.43

447 813 923 708

555 5150 1900 1550

0.5017 0.2177 0.3151 0.3894

1.03 1.43 1.269 1.169

1.242 6.335 2.059 2.189

1.485 1.35 1.394 1.428

1.844 8.549 2.87 3.126

Frame Original frame Steel plate Concentric brace Eccentric brace

Table 6. Cost analysis of strategies. Strategy name

Concrete (m3 )

Unit price ($/m3 )

Steel (kg)

Unit price ($/kg)

Plaster board (m2 )

Unit price ($/m2 )

Total price ($)

3D Shear wall Steel brace

20.44 –

33 –

2000 3260

0.248 0.248

– 102.2

– 6.21

1170.52 1443.142

2D Steel plate Concentric brace Eccentric brace

– – –

– – –

1720 1023 924

0.248 0.248 0.248

73 73 73

6.21 6.21 6.21

879.89 707.034 682.482

7 CONCLUSION The effects of the addition of R/C shear walls, steel plates, concentric steel braces, and the eccentric steel braces to a structure as a retrofit scheme are evaluated and these results are obtained. • The retrofitted buildings by R/C shear walls and steel braces separately have enough strength and stiffness to supply immediate occupancy performance level but their effectiveness is different. The addition of the R/C shear walls is observed to more effective way than addition of steel brace in terms of strength and drift criteria. • The retrofitted frames by steel plates, concentric steel braces, and eccentric steel braces separately have provided enough stiffness and strength to the structure to upgrade its performance to immediate occupancy performance level, although their effectiveness are different. In the study steel plates has shown the better performance than the steel braces do, and the performance of the concentric steel braces is better than the performance of the eccentric steel braces. • At the end of the R factor (Seismic load reduction factor) calculation introduced in ATC-19, it is seen that; addition of steel plate wall is the most effective way to increase the ductility of the structure and addition of steel braces is less effective than addition of RC shear walls. Ductility of the retrofitted frame by eccentric steel brace is better than the one with concentric steel braces. • Addition of steel plate is the most expensive strategy in spite of its good ductile behavior. The cost analysis indicates that the cheapest strategy is the addition of shear wall. Addition of eccentric steel brace is less expensive than the addition of concentric steel brace. 174

REFERENCES Applied Technology Council, “ATC-19”, Components of Response Modification Factors, pp. 17–50, ATC, Redwood City, California, 1996. Applied Technology Council, ATC-40 Seismic Evaluation and Retrofit of Concrete Buildings, Vol. 1, ATC, Redwood City, California, 1996. Aras, F., Nonlinear response analysis of retrofitted structures by different strategies, M. S. Thesis, Boˇgaziçi University, 2001. Canadian Standards Association, Limit States Design of Steel Structures, CAN/CSA-S16.1-94, Rexdale, Ontario, CANADA, 1994. Driver, R. G., G. L. Kulak, A. E. Elwi and D. J. L. Kennedy, “FE and Simplified Models of Steel Plate Shear Wall”, Journal of Structural Engineering, ASCE Vol. 124, pp. 121–130, February 1998. Federal Emergency Management Agency, FEMA-356, Prestandart and Commentary for the seismic Rehabilitation of the Building, FEMA, Washington, 2000. Hamburger, R.O., “Defining Performance Objectives”, Proceeding of the International Workshop on seismic Design Methodologies for the Next Generation of Codes, Bled Slovenia, pp. 33–42, Balkema, Roterdam, 24–27 June, 1997 Jose, A. Pincheira, James O Jirsa, “Seismic Response of Reinforced Concrete Frames with Steel Braces or Walls.” Journal of Structural Engineering, ASCE Vol. 121, pp. 1225–1235, August 1995. Lubell A. S., H. G. L. Prion, C. E. Ventura and M. Rezai, “Unstiffened Steel Plate Shear Wall Performance Under Cyclic Loading”, Journal of Structural Engineering, ASCE Vol. 126, pp. 453–460, April 2000. Ministry of Public Works and Settlement, Specifications for the Buildings to be Constructed in Disaster Areas, ˙IMO ˙Izmir Subesi, ˙Izmir, 1998. ¸

175

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

The behavior of different eccentrically braced frames with short links H. Köber & B. Stef ¸ a˘ nescu Technical University of Civil Engineering, Bucharest, Romania

ABSTRACT: The paper is intended to illustrate some features of different bracing systems used for eccentrically braced frames located in seismic areas. Non-linear static and non-linear dynamic analysis was performed for each of the different braced frames. The maximal values for the displacements, the base shear force, the plastic hinge rotations, the bending moments and axial forces in different structural elements were compared. The history of the formation of plastic hinges was observed.

1 INTRODUCTION A ten-story structure was considered, having the eccentrically braced frames placed as shown in Figure 1. The structure has two spans and six bays of 6.6 m. The eccentrically braced frames were configured according to appropriate Romanian codes. The eccentrically braced frames were designed in six different configurations (using six different bracing systems). Each solution used short links with a unique length of 1.2 m. The six geometry types are presented in Figure 2. The different braced frames were sized so that they could have very close eigenvalues. The values of the first three eigenperiods of the six analyzed braced frames are given in Table 1.

2 INELASTIC DYNAMIC ANALYSIS Each of the six eccentrically braced frames was analyzed under the same base excitation. The excitation was taken from the Vrancea 04.03.1977 earthquake (the most severe earthquake in

2x6.600

Eccentrically Braced Frames

6x6.600

Figure 1. Location of the eccentrically braced frames.

177

Romania in the last sixty years, with a magnitude of 7.2 on the Richter scale) and it consisted of the N-S component acceleration record. The peak acceleration for this record is approximately 0.2 times the acceleration of gravity. The first 22 seconds of this record were used since this period contains all the high acceleration peaks and nearly the whole inelastic activity is expected to occur during this period. The entire analysis was performed with a 0.01 second time step. The results with a larger time step (0.012 seconds) were sufficiently similar to the results with a smaller time step (0.0075 and 0.005 seconds). No damping was considered. Although the six frames were sized so that they could have very close eigenvalues, their behavior was quite different. Table 2 contains the greatest values of the base shear force (Smin and Smax ) and of the horizontal displacement of the last floor (min and max ) recorded during the analysis. Smax is the greatest value of the base shear force and max is the greatest value of the horizontal displacement of the top floor for one sense of motion, whereas Smin and min are the greatest recorded values for the other sense of motion.

K Frame

DC Frame

DM Frame

V Frame

Z Frame

Y Frame

Figure 2. Types of braces geometry.

Table 1. Eigenperiods. Eigenperiod

K frame

DC frame

DM frame

V frame

Z frame

Y frame

T1 (s) T2 (s) T3 (s)

1.125 0.405 0.226

1.111 0.401 0.227

1.067 0.366 0.207

1.104 0.385 0.223

1.109 0.406 0.242

1.112 0.390 0.227

Table 2. Non-linear dynamic analysis results. Frame

Smin (kN)

Smax (kN)

min (m)

max (m)

K DC DM V Z Y

−2304 −2507 −3556 −3232 −3004 −3251

2573 2873 3393 3189 3181 3502

−0.0451 −0.0993 −0.1577 −0.1570 −0.1171 −0.2243

0.3202 0.2987 0.2333 0.2652 0.2823 0.2503

178

3 BEHAVIOR OF K FRAME Except for some small deformations in the central column and in some braces from the upper stories no inelastic deformations were observed outside the potential plastic zones (links and the base ends of the first floor columns and braces). During the analysis plastic deformations were registered in almost all links. These deformations are greater in the first seven stories. The largest recorded plastic hinge rotation was of 0.08345 radians and the largest cumulated plastic hinge rotation was of 0.12241 radians. Beam segments outside the links remained elastic during the analysis although they have smaller cross-sections than similar members of the other analysed frames. Generally the greatest values of the forces were reached at the link end (not at the column end) of the beam segment. Small inelastic deformations could be observed in some braces at different moments during the analysis. In most cases the inelastic deformations were accidental (the plastic hinge rotation was equal to the value of the cumulated plastic hinge rotation, θ = θcum ). The inelastic deformations appeared at the lower end of the first floor braces (in potential plastic zones) and in some braces of the upper stories. Except for the second story central column, no inelastic deformations were registered in the columns outside the potential plastic zones near the base. All these inelastic deformations were small and appeared accidentally. The greatest values of the bending moments on the columns (1411.5 kNm in the marginal column and 569.7 kNm in the central column) were much smaller than the bending moments that appeared in the columns of the other frames. This could be explained in part by the fact that links are not located near the columns in the K frame.

4 BEHAVIOR OF DC FRAME Plastic deformations could be noticed in almost all links. Compared to the values registered in the K frame, the deformations of the links were smaller in the lower stories and greater in the upper half of the frame. The largest recorded plastic hinge rotation was of 0.07902 radians and the largest cumulated plastic hinge rotation was of 0.12199 radians. The beam segments outside the links had an elastic behavior during the analysis. Most of the braces suffered inelastic deformations at different moments during the analysis (see Figure 3). Most of them were accidental (θ = θcum ). These quite small inelastic deformations had comparable values along the height of the frame.Except for a marginal column at the fourth story, no inelastic deformations were recorded in the columns outside the potential plastic zones near the column base. All plastic deformations in the columns were accidental. The values of the bending moments reached in the columns were much greater than those recorded in the columns of the K frame. Considering the fact that most of the braces suffered inelastic deformations at different moments during the analysis, it can be appreciated that the K frame had a more favorable behavior than the DC frame.

K Frame

DC Frame

DM Frame

V Frame

Y Frame

Figure 3. Plastic hinges at 6.38 seconds from the start of the dynamic analysis.

179

Z Frame

Figure 4. Maximum axial forces (white stripes for DC frame; black stripes for DM frame).

5 BEHAVIOR OF DM FRAME During the analysis plastic deformations were registered in all links. Compared to the values reached in the DC frame, the deformations of the links were smaller except for the first two stories. The largest recorded plastic hinge rotation was of 0.06109 radians; the largest cumulated plastic hinge rotation was of 0.11673 radians. No inelastic deformations were observed in any beam segments outside the links. The bending moments and the shear forces were greater than those reached in the DC frame but smaller than the values observed in the beams segments of the K frame. The recorded axial forces in the beam segments had the greatest values of all analysed frames. Plastic deformations appeared in almost all braces at different moments during the analysis (Figure 3). The deformations in the braces were bigger than those reached in the DC frame. The plastic rotations were not accidental especially in the higher stories. This less favorable behavior can be explained through the great axial forces that occurred in the braces and columns. These axial forces were the greatest that appeared in the frames with the braces connected directly to the columns. A comparison between the maximal values of the axial forces registered during the analyses in the braces and columns of the DC- and DM frame is shown in Figure 4.

6 BEHAVIOR OF V FRAME Except for a single brace from the second story, no inelastic deformations were observed outside the potential plastic zones. Plastic deformations appeared in all links during the analysis. The recorded plastic hinge rotations were much smaller than those registered in the K-, DC- or DM frame. The greater number of links in the V frame can explain this basically. The greatest registered plastic hinge rotation was of 0.00387 radians and the greatest cumulated plastic hinge rotation was of 0.0457 radians. A more uniform vertical distribution of the plastic deformations in the links could be observed (compared to the other frames). The greatest cumulated plastic hinge rotations in the links from the stories 2–7 had comparable values. The beam segments outside the links remained in the elastic range. The beam segments outside the links carry together with the braces a great part of the gravitational loads from the floors. That is 180

Figure 5. Deformed shapes of the frames at the end of the analysis (22.08 seconds).

why the values of the bending moments and shear forces are much greater than those that appeared in the beam segments of the K-, DC- or DM frame. The braces had an elastic behavior during the analysis except for a single brace from the second story, where small inelastic deformations appeared (θ = θcum = 0.00043 radians). The columns also remained in the elastic range except for the potential plastic zones from the bottom of the marginal columns where small inelastic deformations could be observed (θ = θcum = 0.00140 radians). The values and the vertical distribution of the bending moments in the marginal columns are similar to the DM frame, whereas those in the central column are similar to the DC frame. The main disadvantage of this frame compared to the frames that had the braces connected to the columns (K-, DC- or DM frame) consists in the greater relative vertical deformations suffered by the floors (Figures 5 and 6). The parts of the floors supported by the braces suffered larger vertical deformations than the parts of the floors supported by the columns. This can be explained by the greater stiffness of the columns compared to that of the beam-braces system when subjected to vertical loads.

7 BEHAVIOR OF Z FRAME During the analysis plastic deformations were recorded in almost all links. These deformations were greater than those reached in the V frame, but smaller than those registered in the K-, DCor DM-frame. The largest recorded plastic hinge rotation was of 0.07185 radians and the greatest cumulated plastic hinge rotation was of 0.07899 radians. The vertical distribution of the plastic link deformations was not as uniform as in the V frame. The greatest hinge rotations appeared in the first five stories. The beam segments outside the links had an elastic behavior during the analysis. The beam segments outside the links carry together with the braces (like in the V frame) a great part of the gravitational loads from the floors. Therefore the recorded values of the shear forces and bending moments are much greater than those that appeared in the beam segments of the K-, DC- or DM frame. Compared to the V frame the bending moments were smaller and the axial forces were greater. Most of the braces suffered plastic deformations at different moments during the analysis. All inelastic deformations were accidental and they were greater than those registered in the braces from the DC- and DM frame. Except for some small plastic deformations recorded in the central column (at the stories 4, 5 and 7) no other inelastic deformations could be observed outside the potential plastic zones from the bottom of the columns. The greatest values of the bending moments in the columns were slightly smaller than those registered in the DC frame. 181

Figure 6. Maximum relative residual deformations registered at the end of the analyses (remaining relative deformations between the ends of the links registered normal to the links axes).

The relative vertical deformations of the floors were larger than those observed in the V frame because the system formed by the braces and beam segments in the Z frame has a smaller stiffness under vertical loads than the one in the V frame (see Figures 5 and 6).

8 BEHAVIOR OF Y FRAME Plastic deformations appeared in almost all links during the analysis. These deformations were smaller than those registered in the DC- and DM frame, but greater than those recorded in the V frame. The largest registered plastic hinge rotation in a link was of 0.03993 radians and the greatest cumulated plastic hinge rotation was of 0.09844 radians. Considering the fact, that many beams, braces and columns from the first five stories of the Y frame suffered inelastic deformations during the analysis, the behavior of this frame can be appreciated to be the poorest of all analyzed frames (Fig. 5). This behavior can be explained by the fact that a favorable global collapse mechanism was difficult to size by design. As long as the plastic deformations remained only in the links the Y frame had a favorable, predictable behavior. When 182

Figure 7. Dissipated energies (Set 1 = energy dissipated through plastic deformations in the links, Set 2 = energy dissipated through plastic deformations in the columns, braces and beam segments outside the links).

the loading level increases and plastic deformations appeared in other members (beams, braces or columns) the behavior of the frame was difficult to control. The influence of the plastic hinges that appeared in the columns, braces and beams (respective beam segments outside the links) can be observed from the graphics in Figure 7. 183

9 CONCLUSIONS • The behavior of eccentrically braced frames during strong earthquakes (quakes which are generating plastic deformations in the frames), is conditioned more on the dynamic characteristics of the structures at the time when the greatest seismic impulses are registered, than on the eigenvalues of the frames (established in the elastic range of behavior). • The main disadvantage of the eccentrically braced frames, which had not the braces directly attached to the columns, consists in the greater remaining relative vertical deformations registered by the floors of the V- and Z frame. • The values of the plastic hinge rotations recorded in the links of the V- and Z frame were smaller than those registered in the other eccentrically braced frames. • The placing of the links near the columns leads to greater values of the bending moments in the columns. • The bracing configuration of the DC frame is more favorable than that of the DM frame, because it leads to smaller axial forces in the marginal columns. • The absence of a favorable global collapse mechanism, conducts after the plastification of the most dissipative members to an unpredictable behavior of the Y frame.

REFERENCES [1] Erdbebenbemessung von Stahlbetonhochbauten, Thomas Paulay, Hugo Bachmann, Konrad Moser, 1990 [2] Normativ pentru proiectarea antiseimic˘a a construc¸tiilor de locuin¸te, social-culturale, agrozootehnice s¸i industriale – indicativ P100-92, 1992 [3] Seismic provisions for structural steel buildings, LRFD-AISC, 1997 [4] Auslegung von Bauwerken gegen Erdbeben, Eurocode 8, 1998 [5] Cod de proiectare seismic˘a P100-2004 (Propunere normativ, redactarea 4) [6] The Behavior of Different Bracing Systems of Multistoried Eccentrically Braced Frames, Helmuth Köber, Bogdan Stef ¸ a˘ nescu Timi¸soara 2003

184

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Cable material consumption depending on the geometrical parameters of hierarchic roof L. Pakrastinsh & K. Rocens Institute of Structural Engineering and Reconstruction, Riga Technical University, Latvia

ABSTRACT: The formation of hierarchic cable roofs and calculation principles have been investigated. The structure is formed by negative Gauss bending type saddle shaped cable roofs by suspending separate saddle roof’s corners to a higher level cable structure. These structures are characterized by all the same advantages, which as well apply to a saddle shaped cable roof, however, with a better ratio between the covered volume and the area. These systems can be intercombined to form wider roof areas. It is suitable to use hierarchic cable structures for long span roofs as well as for completely or partially dismounted provisional coverings. In order to reduce the complexity and the amount of calculation, and to avoid the calculation matrix convergence problem, calculation of stresses and displacements of complicated hierarchically subordinated structures with various geometrical parameters of hierarchic roof could be accomplished by the method of substructuring. The effect of interdependencies of separate structural elements and higher level cable structure can be determined by the iterative approximation method. By using of the proposed principle, the hierarchic cable structure is treated under the different wind and accidental snow load conditions. KEYWORDS: hierarchic cable roof, prestressed orthogonal anticlastic structure, substructuring, lightweight cable net structures. 1 INTRODUCTION Weight reduction and increase of spans of structures are topical tendencies in the development of load-bearing structures, which can be accomplished by application of new high-strength materials. As a result it shows as reduction of ratio between the dead and live loads of a structure from ancient massive structures to contemporary lightweight structures. This ratio can be reduced more than 100 times by the most effective exploitation of the properties of special high-strength materials with much higher specific strength in combination with structural systems where tensile stresses are dominant. Tension structures are characterized by non-linear geometric hardening which results in a less proportional increase of stress in elements in relation to increased external loads. This provides an increased nominal safety factor evaluated at ultimate limit state of structures. Cable roofs with axially tensioned elements and with identical tensile stresses acting in all the cross-section points represent one type of these structures, and it opens possibilities for rational application of advanced materials. The main advantages of cable structures are as follows: new options of architectonic expression, possible translucency, small weight of structures, efficiency of high-strength material application in the production of cables, reduction of the construction time, advantages of transportation (in rolls), good seismic resistance, cheaper fire protection methods by using foaming polymers, low cost maintenance and may be the most essential is the possibility to cover extremely large spans without intermediate columns, which is impossible by using ordinary structures. There are also some drawbacks, i.e., increased deformability mainly of kinematic nature, relatively poor anticorrosion resistance in some cases, the need of supporting structures in taking up the abutment shear, and because of light weight and deformability they require special stabilization arrangements to provide geometric shape invariabilities. 185

Figure 1. Simplified variant of hierarchic cable structure module. *Guy-ropes in longitudinal direction of module have not been conventionally displayed.

The most perspective from the constructional point of view and most expressive regarding architectonical aspect, is a prestressed cable net with negative Gauss bending saddle-shaped hyperbolic surface known as anticlastic form covered with fabric. The structure is formed by orthogonally intersected load-bearing concaved and stressing convexed cables. This model is characteristic by kinematic rigidity and there is no requirement of additional loading for stabilization of structure. The construction has only one statically stable shape, preventing such destructive dynamic instabilities, as flutter or flapping. The model can be constructed on a rectangular supporting contour, which is one of its advantages, and to reach total rigidity of the structure, relatively small radiuses of curvature are required, enabling to provide the minimum material consumption both of the cable net and its supporting contour. Comparing to different cable structures, deformability under exposure of external loading is relatively smaller. From the point of view of material consumption (valuable utilization of load-bearing capacity of high-strength materials) the most rational are saddle-shaped roofing with a compliant supporting contour. Basing on previously accomplished investigations (Rocens et al. 1999, Serdjuks et al. 1999, 2003, Pakrastinsh et al. 2001), were have been determined the rational geometric characteristic values of separate saddle-shaped cable roofing with the compliant supporting contour and sizes in plan from 10 to 50 m, it is shown that the ratio between the roofed volume and the area is unacceptable at larger spans as it causes inexpedient increase of the heated volume. This problem can be solved by application of the saddle-shaped roofing with the compliant supporting contour as standard elements, and by suspending their corners to a higher level cable structure thus obtaining a hierarchic, intersubordinated large-span cable roofing (Pakrastinsh & Rocens 2001) (Fig. 1). 2 FORMATION OF STRUCTURE AND CALCULATION MODEL In comparison with the conventional structures cable structures can be economically suitably used for spans exceeding 40 m. Therefore, for further performance the following dimensions of primary elements have been assumed: 35 × 35 m, 45 × 45 m and 55 × 55 m, thus achieving full structural spans of 140 m, 180 m and 220 m, accordingly. A prestressed cable net with a negative Gauss’ bending saddle-shaped hyperboloid surface covered with fabric has been assumed as the structure of primary element. The primary element is formed by orthogonally crossing curved load-bearing and stressing cables. Lest the roofing should 186

Figure 2. Calculation scheme of primary element. 1, 2 – load-bearing/stressing cables; 3 – edge cable; 4 – LCP Vectran® composite cable; 5 – symmetry boundary conditions; l – span; f – initial deflection of load-bearing/stressing cables; s – load-bearing/stressing cables spacing; fabric covering have not been conventionally displayed.

lose rigidity it must be provided that under disadvantageous load combinations stresses in stressing cables shall not be reduced to zero because cable slack formation can provide damages of roofing or destructive flatter may occur. Primary elements of hierarchic structural module 4 × 4 are modeled and they may be combined thus producing larger roofed areas. Notwithstanding the number of smaller elements, the scheme with slanting pendants has obvious remarkable drawbacks: too long columns, which entail too large top corner vertical displacements of the bottom level standard elements, as well as too large horizontal forces in these due to the angle between the pedants and vertical axis, which in its turn requires strengthening of the standard element. Roofing scheme with vertical pedants is assumed as optimum one. Not to exclude the possibility of expanding the structure in longitudinal direction of the module, transversal direction variant of the higher level cable structure is assumed, and to reduce the total height of structures, the upper part is lowered under the level of primary elements, at the same time lifting the corners of the primary elements by means of vertical latticed post fixed to these cables. Stressing cables change to load-bearing ones and vice versa. Camber values for the numerical experiment are assumed as 1/20; 1/10; 1/5; 1/2.5 part of the span in compliance with the literature recommendations and the existing structures. Calculation scheme of the primary element is illustrated in Figure 2. In the middle part of hierarchic cable structure the board element of the primary element is shaped as an upright cable, which is symmetrically loaded with horizontal forces developed from the load-bearing or stressing cables of adjoining primary elements. In case of uniformly distributed load these forces are intercompensated, which open the way to reduce the material consumption at the expense of the board element. The above-mentioned scheme enables to apply boundary conditions of symmetry during calculations of the primary element and brings closer its action to the element with rigid supporting contour, which considerably reduces the deformability of the primary element. The board element of outer primary elements is modeled as a stretched on cable loaded with the large uncompensated horizontal load coming from the cable net. We choose the camber of these elements equal to 1/15 part of span, basing on the assumption that the board element can be treated as freely suspended cable loaded with distributed load. In this case modeling of an upright rigid board element with length size 35 m is problematic due to large strains. To reduce the total deformability of the primary element and, basing on our previous investigations, the board element is modeled as a cable truss. The stressing value of the board elements is assumed as 50% from the design strength to avoid from slack of the edge element when only permanent load and stressing of the cable net act. 187

The step of the cable net is assumed 1.77 m depending on the constructive condition lest rain bags should form, because the fabric or some other kinds of roofing in cable net structures mainly provide the transfer of external loading to the cable net. Prestressing is assumed for primary dimensioning of primary element of cable net on identical level for all cable nets in compliance with literature recommendations (Walton 1996) which makes 22.5% from the tensile strength (50% from the design strength at material safety factor equals to 2.0). It is required to uniformly resist the opposite direction loads, because each group of cables can work as stressing one depending on the direction of loads. To prevent the increase of the remarkable relaxation effect for steel wires, it is assumed that stresses of permanent load and stressing should not exceed 45% from the cable tensile strength at tension. By combining the primary elements of hierarchic structure, the common board element of two adjoining primary elements remains upright in plan, but in side projection it assumes the form of Sshape curved line, which causes the emergence of cable pendants in corners of the primary element at minimum loadings. To compensate the loss of prestressing, corners of the primary element shall be made from composite cables with enlarged limiting deformation properties, which make up 7% from the total length of cables. One of such materials is recently produced Liquid Crystal Polymer (LCP) Vectran® developed on the basis of polyester molecules with regularly oriented structure along the longitudinal direction of fibers. Fibers made of this material are characteristic by large strength and increased deformation with minimum creep probability. Acting as a component of hierarchic structure, the primary element subjected to the influence of higher level structures loaded with reaction forces of the primary element, which appear in the displacement of the supporting nodes of these elements in the opposite direction. In order to assess the maximum permissible displacements of the primary element supporting nodes without cable slack (which simultaneously shows how strong should be a higher level cable structure to provide the displacements of supporting nodes), a numerical experiment for three level loading of the above mentioned variants of the primary elements was accomplished. It should be observed that the real distribution of aerodynamic coefficients for the structure in question is not established, because model investigations in wind tunnel or computer aided simulation of the exposure of wind flow by using the mathematical appliance of fluid mechanics (Halfmann et al. 2002) are needed. The computing is accomplished by means of finite element method software ANSYS 6.0 University version applying for the cable modeling the universal spatial bar finite element LINK10 with three degrees of freedom in each node with specific bilinear stiffness matrix, which defines that the element works at tension only. The experiment was made by the iterative approximation method assuming the point of reference as the state when deadweight and prestressing loads work on the structure. Since uniform values of intensity of the wind both with positive and negative values can affect the primary element, which is identical with positive values of the snow load of Latvian conditions, structural performance is symmetric relatively horizontal plane and calculations are made only for the displacement of the upper points downwards. Experimental results enable to draw a conclusion that limiting values of the supporting nodes displacements are not dependent on the load value, but on geometric parameters of the primary element. Regularity between the limiting values of supports nodes displacements of the primary elements and the initial curvature of the primary element is shown in Figure 3. In order to prevent the formation of cable slack, a prestressing reserve within the limits of 10 to 40 percent from the total initial stressing shall be provided (suggest the increase of cable crosssections by 1.25 to 5 times relatively primary element condition without nodal displacements). Larger or smaller reserve is not rational because it causes a rapid increase of material consumption for the primary element in case of larger reserve and that for higher level cable structure in case of smaller reserve. Analysis of interaction of the primary element higher level cable structure has proved that rational initial camber area of the primary element lies from 0.05 to 0.2 part of the span, which is beyond rational curvature value of 0.3 part of the span of the primary element as separate structure without nodal displacements. 188

Permissible displacement delta Z, m

0.6

0.5

0.4

0.3

35x35, reserve 10% 45x45, reserve 10% 45x45, reserve 20%

0.2

45x45, reserve 30% 55x55, reserve 10%

0.1

0 0

0.1

0.2

0.3

0.4

0.5

0.6

Initail deflection ratio, f/l

Figure 3. Permissible displacements of supports of primary element depending on initial deflection of primary element.

To reduce the total amount and complexity of the calculation, and to avoid from the convergence problems, the so-called substructuring method is used, which divides the structure into levels. In the first stage the required cross-sections of the primary elements cables and reaction forces on the supporting nodes are calculated, thereby collecting information (boundary conditions) for the next stage calculations of higher level cable structure. As a result of these calculations data for the next iteration are provided for calculation of the primary saddle-shaped element with new boundary conditions. Thus using of iterative approximation method the effect of interdependencies of separate structural elements and a higher level cable structure can be defined. In the first stage of calculations for the initial dimensioning of the primary element cables, prestressing is assumed identical for all cables in the net, which makes up 40% to 10% from the design strength of material, taking into account the reserve accordingly up 10% to 40% to the displacement of the supporting nodes. At first the calculation were made to find cable cross-sections under the primary element under full load by method of stepwise approximation for each type of the primary element and for each load level providing that no stresses exceeding the design strength of the respective material should develop in any group of cables. In the next step calculation was made with the discovered cross-sections of cables and for the load of dead weight and prestressing. To develop the reference deformed state of structure closest to real structure after assembling and before exposure of live load, adjustment of nodal coordinates was made in compliance with this deformed condition. In the next stage a full load is applied to the primary element with corrected geometry thus obtaining the reaction forces of the supporting nodes, i.e., boundary conditions for calculations of higher level cable structures. Using stepwise approximation are obtained the cable cross-sections of higher level structures on condition that at nodes of applying the load from primary elements shall not exceed the values of allowable displacements for the bearing nodes of the primary element.

3 CALCULATION RESULTS AND RECOMMENDATIONS Since mat consumption of whole hierarchic structure is dependent on many parameters, to determine the rational form from the point of view of material consumption, the numerical experiment analysis of the behavior of hierarchic cable structures was accomplished using the values of 189

Table 1. Regression coefficients of response function. bo b1 b2 b3 b4 b5 b12

650.6 −21.2 −80.9 −130.9 −987.6 −1445.0 4.8

b13 b14 b15 b23 b24 b25 b34

2.2 −22.7 −17.3 151.0 −257.7 −236.1 −911.6

b35 b45 b11 b22 b33 b44 b55

−393.3 1973.8 0.4 48.8 741.9 3095.1 3377.8

variable parameters: dimensions of the primary elements: 35 × 35 m, 45 × 45 m, 55 × 55 m, initial deflection values for the primary element are assumed as follows: 1/20; 1/10; 1/5; 1/2.5 part of the span, initial deflection values for a higher level cable structure: 1/10; 1/5; 1/2.5 part of the span. For the 3 levels of loads: 1 – dead load and occasional snow load (0.48 kN/m2 ), 2 – dead load and negative wind suction in accordance with recommendations (Davenport 1995) where the load values proportional to the distribution of actual aerodynamic coefficients, are replaced by uniformly distributed load (0.76 kN/m2 ), 3 – dead load and negative wind suction with twice increased wind speed (1.6 kN/m2 ) in compliance with literature recommendations which exceed by intensity snow load values of building codes valid in Latvia. By using FEM software ANSYS University for each experimental point value determination, and software STATISTICA for the D-optimal experimental plan calculation and for analysis of the results, the correlation between the main geometrical characteristics of the roof and the material consumption required for the roofing of one unit of area is determined as second-order polynomial:

where C – material consumption required for the roofing per unit of the area, kg/m2 ; f – initial deflection ratio for the primary element; F – initial deflection ratio for a higher level cable structure; l – span of the primary element, m; p – calculation load, kN/m2 ; r – reserve of prestressing, %. The response function coefficients (Table 1) are determined on the results of numerical experiments by least squares method. Analysis shows that the important effect is achieved by two following parameters: the initial deflection ratio of higher level cable structure and the prestressing reserve of primary element, influence of which on mat consumption for one case is shown in Figure 4. The rational values for this particular span case 180 m are: Initial deflection ratio 0.288 and Prestrressing reserve 0.302. Comparing the results with other existing long span structures (Fig. 5) it can be concluded that material consumption of such type of structures lies in the rational material consumption area. Summing up the analysis of structural performance, the following recommendations for rational hierarchic structures have been worked out: – from the constructive point of view most rational module of hierarchic cable structure should have dimensions 4 × 4 primary elements, the largest one is characterized by irrational material consumption for the higher level cable structure; – to reduce the total height of structures, the upper higher level structures cables are lowered under the primary element surface by simultaneous lifting up of the primary elements corners by means of vertical latticed posts fixed to these cables; – the bottom higher level structure shall be shaped with curvature upwards, because the plane bottom tie net is typical of irrational material consumption (Pakrastinsh & Rocens 2003); – the roofing scheme with vertical pendants shall be designed since the application of inclined ones may require the strengthening of the primary elements; – not to exclude the probability of extension of structural module in longitudinal direction, transversal direction variant of the higher level cable structure is recommended; 190

Figure 4. Material consumption of hierarchic cable structure depending on initial deflections ratio of higher level structure and the prestressing reserve of primary element (span L = 180 m (45 × 4), load 1.5 kN/m2 ).

Figure 5. Material consumption depending on the span of structures. *Hierarchic cable roof material consumption is shown for three level loading.

– to prevent the appearance of cable slack of the primary element, a reserve within the limits of 10% to 40% from the total initial stressing shall be provided in the prestressing of cables; – for future development of structures, merging of the bottom and upper higher level cable structures into the cable trusses shall be provided, which can reduce the supporting nodes displacements and the proportion of material consumption of these structures; – structural bracing between the corners of the primary elements shall be designed in order to distribute the effect caused by concentrated forces and horizontal wind loads. 191

4 CONCLUSIONS By means of the accessible software internal forses, displacements and material consumption of complicated intersubordinated hierarchic cable roofings can be computed and the effect of displacements of separate structural elements on other structural elements can be determined. Hierarchic cable roofings can be usefully applied for large span structures (e.g. sports grounds, concert halls, parking – lot roofings) in areas with insignificant snow load. These structures can be also used as completely or partially dismountable translucent provisional coverings for covered agricultural areas. A new and rational from the point of view of material consumption hierarchic cable roofing structure enabling to reduce the material consumption by 1.5–2 times comparing with other largespan structures has been worked out. A saddle-shaped primary element structure has been developed and dimensioned, and selection of materials for components is accomplished. A determination method of permissible displacements of corners of the primary element structure acting as a component of hierarchic structure has been developed. It has been proved that the primary elements of hierarchic cable structures cannot be modeled without the application of composite cables with increased limit of elongation. It has been proved that due to the iteration of the primary element and higher level cable structure the rational initial curvature area of the primary element lies from 0.05 to 0.2 part of the span, which is beyond rational curvature value of 0.3 part of the span for a separate primary element. Computing methods of the cable material consumption per covered unit of the area depending on geometric parameters and nodal displacements, which is based on the combination of submodels and iteration methods has been developed. Second-order polynomial correlation between the main geometric parameters of hierarchic cable structures and the material consumption of cables per roofed unit of the area has been determined, and recommendations for modeling rational hierarchic structures are presented. REFERENCES Davenport, A. 1995. How can we simplify and generalize wind loads? Journal of Wind Engineering and Industrial Aerodynamics Vol. 54/55; 657–669. London. Halfmann, A., Rank, E., Glück, M., Breuer, M., and Durst, F. 2002. A Geometric Model for Fluid-Structure Interaction of Wind-Exposed Structures. Proc. of the Ninth Int. Conf. on Computing in Civil and Building Engineering: 309–320. Taipei. Pakrastinsh, L, Serdjuks, D., Rocens, K. 2001. Some Structural Possibilities to Decrease the Compliance of Saddle Shape Cable Structure. Proc. of the 7th Int. Conf. Modern Building Materials, Structures and Techniques: 24–25. Vilnius. Pakrastinsh, L., Rocens, K. 2001. Hierarchic Cable Structures. Scientific proceedings of Riga Technical University, Vol.2 Architecture and construction science: 130–135. Riga. Pakrastinsh, L., Rocens, K. 2003. Evaluation of Cable Material Consumption of Ties Depending on the Nodal Displacements of Hierarchic Roof. Scientific proceedings of Riga Technical University, Vol.2 Architecture and construction science: 182–187. Riga. Rocens, K., Verdinsh, G., Serdjuks, D., Pakrastinsh, L. 1999. Composite Covering Structure. Latvian Republc patent Nr.12191. Riga. Serdjuks, D., Rocens, K., Pakrastinsh, L. 1999. Rational geometrical characteristics of saddle shape cable roof supported by tensioned cables. Proc. of the 6th Int. Conf. SF99: Modern Building Materials, Structures and Techniques: Vol. II, 122–127. Vilnius. Serdyuks, D., Rocens, K., Pakrastinsh, L. 2003. Prestress Losses in the Stabilizing Cables of a Composite Saddle-Shaped Cable Roof. Mechanics of Composite Materials. Vol.39, No.4: 513–522, Riga. Walton, J. 1996. Developments in Steel Cables. Journal of Constructional Steel Research Vol. 39, No.1: 3–29.

192

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Wind and snow loads: some methodological problems of normative regulations B. Snarskis, R. Simkus & V. Doveika Institute of Architecture and Construction of the Kaunas University of Technology, Kaunas, Lithuania

A. Galvonaite Lithuanian Hydrometeorological Service, Vilnius, Lithuania

ABSTRACT: The setting of characteristic and design values of atmospheric loads prescribed by the relevant design codes (e.g., by ENV 1991) involves some serious methodological and practical problems remaining not elucidated adequately. In particular, the estimation of variability of atmospheric actions is problematic; it must be accomplished with appropriate consideration of uncertainties related to spatial («territorial») variance of actions alongside the more self-evident employment of data on the merely temporal variations. We submit some results of analysis performed on Lithuanian climatological data as an illustration of conceivable approach to the problem. In addition, some problems related to the account for wind gusts are mentioned although not analysed in sufficient detail.

1 INTRODUCTION The directions of Eurocodes governing the choice of characteristic and design values of wind and snow loads (especially, directions given in EN 1991-1-3:2002) constitute one of the foundations for structural design and, first of all, for elaboration of various special-purpose normatives. Nevertheless, there are serious methodological problems of interpretation and implementation of these directions formulated somewhat abstractedly and aimed mainly at rather specific conditions of application. This paper presents a brief overview of some such problems based on analysis done by the authors when examinating the normatives recently adopted in Lithuania (the so-called “Technical Regulations of Construction” [Lith. Statybos techniniai reglamentai – STR], including STR 2.05.04:2003, which deals with snow and wind loads) and analysing the pertinent climatological data. A corresponding study is now conducted in the Institute of architecture and construction of KTU. Our main objective here is, however, not a critical revision of these local regulations (although such a revision is actually a necessity) but general methodological remarks following from our analysis and from some unqestionable theoretical considerations. Nevertheless, some particular quantitative results relating to the local Lithuanian regulations and based on our calculations will be presented in the closing fragments of this paper as an illustration of practical meaning of the considerations put forward here. It is hard to formulate immediately all conclusions to be presented here – and the presentation of them will take almost all of the remaining part of this paper – but in a very concise exposition our main assertions may be outlined as follows. Firstly, we emphasize the necessity of a careful taking into account the proper measure of uncertainty associated with the random parameter that is characterized (e.g., the coefficient of variation of the snow or wind load – or any other similar statistical indicator); and, particularly, the necessity of taking into account the «spatial» («territorial») component of variance alongside the more obvious analysis of the «local temporal» variance is stressed. 193

Second, we call attention to the possible paramount importance of wind gusts for the total wind loads – a circumstance occurring at least in some quite not unusual cases but being, in all likelihood, underrated rather frequently. These statements, however, stay in need of more comprehensive explanation.

2 CONSIDERATIONS RELATING TO DISTINCTION BETWEEN VARIOUS COMPONENTS OF UNCERTAINTY IN STATISTICAL MODELS OF ATMOSPHERICAL ACTIONS The Eurocodes describe the characteristic values of wind and snow loads as the quantiles of annual maxima of corresponding actions with probability 0.02 of excess. Nearly the same requirement, in the sense of its intended practical meaning, is presented in STR 2.05.04 adopted now in Lithuania (except for somewhat different formal definition and for some slight misunderstanding in minor details that are not worthy of treatment here). It might appear at first sight as a very clearcut definition but in actuality it is not free of problems, mainly in probabilistic interpretation of “corresponding actions” mentioned above. First (but not foremost), problems arise not infrequently when the influence of specific physical circumstances at the given site (such as configuration of the building, possibility of thawing of the snow on the roof caused by heat transfer from within the building, etc.) must be measured quantitatively. However, it seems likely that we may omit such circumstances here since auxiliary formulae appearing in EN 1991-1-3:2002 and other normative literature allow mostly quite agreeable account of these circumstances – perhaps with only not dangerous distortions in the picture of uncertainties and random deviations that is to be considered when reliability requirements are chosen. A question of prominent importance, however, remains latent and calling for a more attention in this statistical scheme – namely, a question about a correct probabilistic and statistical definition of the random variable mentioned as “annual maximum of the load”. In other words, the question is: what is the amount of probabilistic uncertainty of such a variable (and, respectively, of its statistical variance as reflected in observations) that must be taken into account when analysing the distribution of such variable and estimating its quantiles? There are various interesting aspects of this question – but of special note are problems connected with uncertainties in the probabilistic model of the variable under consideration (maximum of the load) resulting not only from variability of actions at a strictly defined location (for which a more or less representative sample of data is presumably available) but also from the fact that the structure being considered is to function in a location which we are not able to characterize so definitely. (In many cases these uncertainties spring up simply because of scantiness of climatological information related to the individual site – but sometimes even because the localization of the site itself is not determined yet, as it happens, for example, when a design is used repeatedly in numerous places.) And we put forth here the assertion that in typical cases the variance of the analysed variable (e.g., the random level of an annual maximum of the load) must be viewed as including at least two components of variance: (1) the temporal component – i.e., the component of uncertainties related with random properties of a chosen time interval (usually of a chosen annual period and/or a sequence of such periods); (2) the spatial component – i.e., the component of uncertainties related with random properties of a site (and/or a subset of sites chosen from the general set of sites taken into account). Practically, in analysis of data samples, the first component may be observed in comparison of periods surveyed, provided that in any of such confrontations the place of observations is uniformly fixed; and the second may be observed in comparison of all sites treated by official regulations as belonging to the same region – provided that in any of such confrontations the time of observations is uniformly fixed. In reality, neglection of the «spatial» uncertainties in practical statistical investigations of atmospheric loads may be observed in numerous publications where probability distribution of the load 194

is straightforwardly identified with some empirical distribution obtained at the nearest meteorological station – so that unknown special possibilities of the site targeted in such practical application are totally ignored. Nevertheless this simple-minded assumption is reasonable quite rarely – maybe only if a representative sample of data has been obtained in a rather close proximity to the site for which the design values are to be determined. It should be realized here that substantial differences of load data (comparable even with the scale of the load variance throughout all the region for which the normative load levels are defined uniformly) can be observed sometimes even under conditions when the comparison refers to a pair of localities separated by a distance of several kilometres in all. Therefore we may not rely upon such simplifications. 3 CONSIDERATIONS RELATING TO THE TREATMENT OF DATA DESCRIBING THE WIND VELOCITIES We must also notice the problematical character of the fact that in Eurocodes (and as well in the STR), when the observational data are treated, the characteristic and design values of wind loads are related mainly to measurements of “moving” averages of the wind velocity – with averaging on time intervals as long as 10 minutes. The action of more intensive momentary gusts is to some extent taken into account, but the inclusion of parameters referring to gust velocities is only indirect and so organized that the long-term effects are usually prevailing in the end results (at least when the most complicated methods of dynamical calculations are not applied and simplified normative recommendations are followed). Such construction of norms is likely to be sound when rather tall and relatively not very stiff buildings with rather prolonged oscillatory motions are kept in mind – but for relatively simple and stiff structures of low buildings this approach is somewhat misleading since even single gust can make a fatal impact. The analysis of wind observations conducted by us till now is based mainly on data representing maximal velocities in gusts. Therefore it is difficult at this point to compare our conclusions with the recommendations of Eurocodes or the STR – though such a comparison is not meaningless however. (By the way, analysis of data for “moving averages” of wind velocities is being performed by us currently.) The problems mentioned here (and, moreover, some problems associated with a possibility of certain provisional modifications of initial data to account for particular effects of local environment at meteorological stations) are the main causes of the fact that in this publication most attention is concentrated on snow (not wind) loads. Nevertheless the results of statistical analysis of wind velocities obtained so far and briefly presented below seem to indicate definitely that the gust velocities are significantly more dangerous than it is believed to be on the grounds of norms adopted now. 4 SOME REMARKS ANTICIPATING THE REVIEW OF OUR STATISTICAL RESULTS (AND IN OUR OPINION DESERVING NOTICE ALSO FROM THE STANDPOINT OF OTHER SIMILAR INVESTIGATIONS) The properties of empirical distributions of sample data are such that approximative description of these distributions is, as a rule, quite good when the lognormal model is taken as the «theoretical» model.1 Having adopted this scheme of approximative description, let us recollect briefly some important properties of lognormal variables and characteristics of their distributions. (We hope that these remarks, though not new in their origins, can be interesting for authors of other similar investigations. Furthermore, they are necessary for the following presentation of our statistical illustrations.) 1 We do not advocate thereupon that this model has been supposedly established as undoubtedly superior above other

possible models. We state only that the approximation works well at least to the level of quantiles covered here – without manifestation of any evident systematic deviations. In addition, this model is very handy by its mathematical features.

195

To begin with, if a numerical random variable Z has lognormal distribution, then its logarithm L := loga Z is distributed normally – and it is easy to understand that the choice of the base of logarithms (a) results here only in scale transformation of the variable L. It is, moreover, also evident that in efforts to analyse such variable Z and to give its mathematical description it may be advantageous to use such variable L as a convenient auxiliary tool since the properties of normal distributions are very helpful and well explored. To be more specific, in doing so there is good reason to define logarithms mentioned just above as natural logarithms ln Z (with the base of logarithms equal to e = 2.71…). Among other motives, it is pertinent to consider here the fact that such description of lognormal variable leads to a situation where the «logarithmical image» L of the original lognormal variable Z is distributed so that mean square deviation σL of L is asymptotically equivalent (as its value decreases) to the coefficient of variation υL of the original variable Z. This is doubtlessly a great convenience for an easy and clear understanding of the values assigned to the parameters thought of here. Hereafter we shall keep in mind just this way of using the auxiliary variable (logarithmic image) L associated with a given supposedly lognormal Z (that is to be described) by logarithmic transformation (and being supposedly normal). It is needless to say that for description of lognormal random variable Z all usual parameters of its distribution such as mean value µZ , median ηZ , mean square deviation [standard deviation] σZ , coefficient of variation υZ , etc. (or, perhaps, sample estimates of these parameters for a sample Z = (z1 , z2 , …, zn ) presenting the observed values of Z – if empirical distributions and statistical estimations are to be considered explicitly rather than a probabilistic model adopted as a result) may be employed in a quite traditional way. Nevertheless there are grounds for use of some other parameters more specific to lognormal model – in particular, some parameters of the normal image L (µL , ηL , σL , …) similar to the parameters of Z mentioned above (or some sample characteristics of L from the sample L = (ln z1 , ln z2 , …, ln zn )). Of special interest here are two such parameters: (1) the mean µL of the distribution of L (or an estimate of such a mean if statistical estimations are presented explicitly), value of which obviously corresponds to the median ηZ of the original variable (ηZ = exp µL ) if the assumption of lognormality is really applicable to Z; (2) the mean square deviation σL of the distribution of L (or an estimate of such a parameter). The interpretation of these parameters is quite traditional and does not require any explanation; but it might be well to point out the practical consequences following from such treatment of the model (when the relation between Z and its normal image L is put to use). One of these consequences is that we arrive at using the median of Z (or, if we are adhered to strictly statistical point of view – the estimate of this median) in a role closely resembling the usual role of mean values or sample averages (although the difference between the two is worthy of consideration and generally is not negligible2). The second consequence is the wide use of the mean square deviation of the logarithmic image [of the variable under consideration – i.e., use of σL for the given Z] – use of a characteristic that is, as it has been indicated before, similar to υZ , although not identical to it. Bearing in mind the importance and frequent use of σL we shall abbreviate its denomination (“mean square deviation of the logarithmic image”) with brief notation “m.sq.d.l.i.”. 5 SOME PRELIMINARY RESULTS REFERRING TO THE LITHUANIAN STATISTICAL DATA AS AN EXAMPLE OF THE APPROACH PRESENTED ABOVE 5.1 A brief characteristic of the data used and the scope of analysis The data used in this study may be defined briefly as gathered from 19 meteorological stations in 1970–2002. (Minor adjustments of this statement would be appropriate depending on particular 2 More precisely, for a lognormal variable Z the relation between η and µ is given by formula Z Z

µZ = ηZ (1 + υZ2 )1/2 .

196

tasks of analysis and some gaps in observations; however, such accuracy is not strongly needed here.) Supplementing of data with older information is envisaged and realized gradually – but it takes long because of difficulties of handling the old archive materials. So this study is not completed yet; nevertheless some primary results seem to be worth of presenting them already now. 5.2 The results relating to snow loads When the random variable being analysed is defined as a maximum for a winter season amount of water per unit surface (observed at some site which is, however, not fixed a priori and is treated so far as a variable of the model) – let us denote it traditionally as s and express its values in kg/m2 – and the results are concurrently surveyed for all the analysed sites without regional divisions of the country (although grouping of data by individual sites and confrontation of results from various sites is of course a requisite perpetually), the following results are obtained. The most typical estimates of the median value ηs of s are, in a review of data for fixed sites or for single winters, about 45 … 50 kg/m2 – however, with substantial variation from one site to another (from approximately 33 to 73 kg/m2 ) and in comparing of various winters (from very low values as some winters are almost snowless, and to about 122 kg/m2 in 1995/96). The estimate of median for the joined sample of all observations is about 46.0 … 46.3 kg/m2 . (Slightly different results can be obtained depending on one of possible methods of treating data.) Probably yet more interesting are the estimates of the m.sq.d.l.i. σln s of s – the measures of its variance as observed on various grouped samples. These are about 0.56 for the «local» samples (taken at fixed sites and representing the «serial» variation of the variable) – and about 0.35 for the «territorial» samples (gathered in a fixed time period, usually a winter, and representing the «spatial» component of variance). Thus the «local serial» component of variance is distinctly prevailing over the «spatial» («territorial») component; however, both are great (greater than almost all analogous characteristics of variance defined for other factors of reliability essential in structural design) so that no one of them may be neglected. The most necessary characteristic of variance here is, naturally, a characteristic in some way summarizing the tendencies of «serial» and «territorial» variance marked above. Such a combined estimate may be obtained assuming that influences of «serial» and «territorial» variance are mutually independent in the probabilistic sense and expressed multiplicatively by respective factors – random multipliers. Then the overall measure of variance may be represented by m.sq.d.l.i., equal to approximately 0.65 or 0.66 (again with possibility of slightly different results under various assumptions as to the treatment of data). If we were to establish the characteristic values sk of snow load without any regional divisions and expressing the load in terms of water content (as is the case in the remarks above), values of sk are defined by the formula ηs exp(2.054σln s ) with «overall» values of ηs and σln s mentioned above. (The multiplier 2.054 here is the value of standard normal variable which is exceeded with probability 0.02 predetermined for the quantile that we are estimating.) As a result a value about 177 kg/m2 (or, if various minor details of procedure are modificated – about 175 … 180 kg/m2 ) would be chosen. (Conversion to the measurement of acting forces gives here results about 1.75 kN/m2 .) It would be not an optimal decision, however, as some regional differences are unquestionable. So far as we can judge from the data at our disposal, the only part of Lithuania where the snow loads are definitely greater is the hilly region in the west. As for the other regions, there is no firm evidence for clear distinction of any sizable territory – although an experienced meteorologist in all likelihood would be able to indicate some localities where relatively great or small values of the load are probable. (It is plausible, for example, that a narrow strip of hills close by southwestern border of the country is also to be treated as a locality with greater snow loads. Nevertheless there are no observations from Lithuanian territory giving an explicit proof of such conclusion – we can base it only indirectly on some data from Poland.) 197

Assuming the division of territory into regions as marked above, we obtain the following estimates of parameters characterizing the distribution of s and, consequently, the characteristic values sk : (i) for the western region – ηs ∼ = 46 kg/m2 and the values of m.sq.d.l.i. equal, roughly approximately, to 0.62 or even 0.66 for the «local serial» variance and 0.33 or even 0.41 for the «territorial» variance – resulting in overall estimate of m.sq.d.l.i. σln s ∼ = 0.71 … 0.78 and estimate of characteristic value sk in the neighbourhood of 1.92 kN/m2 or even considerably greater levels well above 2 kN/m2 (unfortunately, estimates mentioned here are not exact since the variance is extremely great and the amount of information from the region is rather scant); (ii) for the remaining part of the country – ηs ∼ = 49 kg/m2 and the values of m.sq.d.l.i. equal approximately to 0.51 for the «local serial» variance and 0.26 for the «territorial» variance – resulting in overall estimate of m.sq.d.l.i. σlns ∼ = 0.58 and estimate of characteristic value sk in the neighbourhood of 1.6 kN/m2 . 5.3 The results relating to wind loads If we confine the review, as it has been stated in our introductory remarks, only within the context of wind gusts velocities (for one year long periods), the following results are to be noted – again with dividing the country into two regions, one of them covering the western belt of territory near the sea. (However, the definition of this western region as applied to wind loads must be different from the division of the territory assumed for the snow loads – the belt much narrower than in the former case must be set off as the «seaside» region now.) For the western (seaside) region: estimates of median about 26 m/s and m.sq.d.l.i., most likely, about 0.16 for the «local serial» variance and about 0.17 for the overall estimate of variance. (The paucity of data here is even more acute than in the case of snow loads as there are just three meteorologic stations in this region; however, this shortage is to some extent counterbalanced by the fact that the variance in this case is not so high.) Thus the characteristic levels of the variable under consideration (wind gust velocity) should rather be for this region about 36 or 37 m/s. For the remaining part of the country: estimates of median about 21 or 22 m/s and m.sq.d.l.i., most likely, about 0.1 … 0.14 for the «local serial» variance and about 0.14 … 0.15 for the «territorial» variance – which leads to the overall estimates of m.sq.d.l.i., in a rough approximation, near to 0.18 – and estimates of the characteristic level of wind gust velocity near to 32 m/s. 5.4 Some practical inferences As a practical conclusion regarding the level of reliability requirements in local norms it may be said with reasonable confidence that snow load normatives adopted now in Lithuania are significantly below the level defined by the Eurocode methodology (at least if this methodology is applied with due consideration of the facts mentioned above). The definition of wind load parameters is more problematic and deserves more detailed further investigation; nevertheless, it is very likely that these parameters are also defined not quite appropriately. (Apart from the doubt expressed above and relating to the treatment of wind gusts, there are yet more apparent disadvantages of our Lithuanian normatives stemming from a very risky definition of partial factors of safety γ.) These problems, however, go already clearly beyond the subject of this paper which deals mainly, as it has been stated above, with chosen methodological problems and presents some quantitative results only as an illustration of the suggested approach. REFERENCES EN 1991-1-3:2002. Actions on Structures – Snow Loads. Brussels, CEN General Secretariat, 2002. STR 2.05.04:2003. Poveikiai ir apkrovos [statybos techninis reglamentas]. [Actions and loads – technical regulations for construction (In Lithuanian).] // Valstyb˙es žinios, Nr. 59 (2003).

198

Session 4: Research and development concerning mixed building technologies

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

On the performance-based design of a class of “adaptive” fiber-composites for applications in building structures Y.M. Haddad & J. Feng University of Ottawa, Ottawa, Canada

ABSTRACT: Polymeric fiber-composites offer definite advantages as potential “adaptive” damping material systems that can be tailored to possess a pre-determined “time-memory”. In this paper, simultaneous optimization of damping, stiffness and specific weight of a class of such material is carried out in conjunction with the underlying microstructure. The obtained results show that short aligned-fiber reinforced composites have superior design flexibility and damping properties as compared with those of long aligned-fiber composites. Here, in order to increase the damping, it may be necessary to sacrifice the stiffness and vise-versa.

1 INTRODUCTION Laminated fiber composite materials, with their high specific modulus, high specific strength, and their proven capability of being “adaptive” offer definite advantages when compared with their metallic counterparts. These fiber-composite materials are used extensively in engineering structures. However, their response behavior under various loading conditions is still a concern, since they may experience more complex damage mechanisms than other classes of materials under ordinary service loading. Modeling such response, in terms of the dominant microstructural variables (e.g., Haddad, 1995, 1998), can provide in-depth understanding of the damage mechanisms involved and consequently improve the composite material damage resistance. It is well-known that fiber reinforced polymer composites have higher specific strength and stiffness when compared with conventional materials (e.g., Haddad, 1998). Much effort has been devoted to the improvement and optimization of these properties in various composite structures. Good vibration damping properties are also particularly important for composite structures used under dynamic loading. Due in part to the extensive use of conventional structural materials, which in general have poor internal damping characteristics, the potential for significant improvement and optimization of damping in advanced fiber reinforced composites has not been yet fully realized. Meanwhile, the use of short fiber (discontinuous) reinforcement in composite structures is still quite limited, particularly in carrying load structures. This may be primarily due to a much more involvement of the pertaining industry with the long aligned (continuous) fiber composites. In the conventional damping theory, the damping factor is often assumed to vary comparatively little with frequency for a large class of polymers, particularly at temperatures near their glasstransition temperature. Consequently, a large number of researchers considered the damping factor for this class of materials to be constant. However, in the case of fiber-reinforced composites, and in particular short-fiber composites, the damping factor is frequency-dependant, and varies considerably with the microstructure characteristics, e.g., fiber-aspect-ratio, fiber volume faction and fiber off-axis angle (e.g., Gibson et al., 1982, Sun et al., 1985, Suarez et al., 1986). The damping properties of long aligned fiber composites have been studied by numerous researchers (e.g., Bert and Clary, 1974, Bert, 1980). On the other hand, there are relatively few reports concerning the damping of short fiber composites. In this context, studies reported, for instance, by Gibson et al. (1982) indicate that vibration damping properties of fiberreinforced composites with polymeric matrix may be significantly improved and could be readily 201

optimized by using, as reinforcement, short rather than long fibers. More specifically, these studies examine the effects of fiber-aspect-ratio, fiber spacing, fiber orientation and viscoelastic properties of both matrix and fiber on the damping and stiffness of short-fiber reinforced composites. One possible explanation for the above-mentioned design advantage of short-fiber composites could be the presence of a shear loading transfer mechanism between the fibers and the polymeric matrix at the fiber-segment ends. In this, it is often argued in the literature that shear deformation might be responsible of strain energy dissipation in polymers. The research work of Gibson et al. (1982), for instance, indicates that by varying the fiber-aspect-ratio and fiber orientation, maximum damping and stiffness could be achieved separately. This observation implies that the optimum fiber-aspect-ratio and orientation for high damping would not be necessarily optimal for high stiffness. Consequently, it is important to study the influence of the various governing parameters on both damping and stiffness. The optimization of such trade-off between damping and stiffness constitutes the main objective of this paper. In this context, the effects of selected microstructural parameters, e.g., fiber-aspect-ratio, fiber off-axis angle and fiber volume fraction, on the damping and stiffness of polymeric fiber-composite systems are first examined, and subsequently optimized for a class of E-glass/epoxy composite material. 2 MICROMECHANICAL CONSIDERATIONS 2.1 Force-balance approach There appear to be several primary sources of enhanced damping in polymeric matrix materials, e.g., the viscoelastic nature of the bulk matrix and that of the interface, in addition to the friction mechanism at the interface as caused by the relative motion between the matrix and the fiber. Both of these effects may prove to be significant in the case of short-fiber composites whereas high shear stresses are developed at the fiber-matrix interface. When a short-fiber composite is subjected to a cyclic loading, the matrix at the fiber interface near the ends of the fiber undergoes a high cyclic shear strain, thus, producing significant viscoelastic energy loss. The shear stress concentration at the fiber/matrix interface may also induce plastic effects as well as partial debonding at the interface that would result in a slip between the fiber and the matrix and in an accompanying frictional loss. Such fiber/matrix debonding would, however, affect adversely the strength and stiffness of the composite. It is desirable, therefore, to have a strong interfacial bonding so that slip at the interface may be avoided. Thus, the most viable mechanism of enhanced dissipation appears to be the shear deformation in the matrix caused by shear stress concentration near fiber-segment ends. Based on the stress transfer mechanism between the fiber-segment and the matrix, it is obvious that there are several microstructural parameters that would influence the shear stress distribution at the interface. This situation would become more complex if the interaction between neighboring fiber-segments is taken into account. It is obvious that the ideal situation for designing a short aligned-fiber reinforced composite would be to optimize the damping and stiffness simultaneously with respect to the composite microstructure controlling parameters. In this context, the “Force-Balance Approach” (e.g., Sun et al., 1985) is adopted in this paper to derive an analytical model for optimizing simultaneously both damping and stiffness of short aligned-fiber reinforced composites with a polymeric matrix. The referred-to approach is often regarded as an elastic mechanics-of-materials analysis, namely, Cox’s model (Cox, 1952), combined with the well-known “elastic-viscoelastic correspondence principle” (e.g., Haddad, 1995, 2000). In this context, a particular application is being dealt with, i.e., a short aligned E-glass/epoxy composite system. The assumptions pertaining to the “Force Balance Approach” may be summarized, as follows: – A structural element of a round fiber-segment surrounded by a cylindrical matrix under uniaxial tensile loading is considered. – Both the fiber and matrix are behave in an isotropic linear viscoelastic manner. 202

Figure 1. The damping non-dimensional ratio ηx /ηm vs. fiber-aspect-ratio l/d and off-axis angle θ . Fiber volume fraction Vf is set to be 60%.

– The transfer of load from the matrix to the fiber depends upon the difference between the actual displacement at a point on the interface and the displacement that would exist if the fiber was absent. – The interfacial bonding, between the fiber and the matrix, is considered as perfect, and the interface would behave in a linear viscoelastic fashion similar to the bulk matrix. 2.2 Numerical examples By setting the fiber volume fraction, Vf , at 50%, 60% and 70% and plotting the values of both ratios ηx /ηm and Ex /Em against the fiber-aspect-ratio l/d and the fiber off-axis angle θ, one could identify that with the increase of the fiber volume fraction Vf , the values of both ratios ηx /ηm and Ex /Em change almost linearly, with the values of ηx /ηm are monotonously decreasing and those for Ex /Em are monotonously increasing; e.g., Fig. 1. Meanwhile, by setting the fiber off-axis angle θ as 0◦ , 20◦ , 40◦ , 60◦ , 80◦ and 90◦ , and plotting the non-dimensional damping and stiffness ratios (ηx /ηm and Ex /Em , respectively) of the composite specimen, where the subscript x designates the loading direction, against the fiber volume fraction Vf and the fiber-aspect-ratio l/d, it was observed that with the increase of θ, the values of the ratio ηx /ηm are increasing, meanwhile, those for the ratio Ex /Em are decreasing. When θ reaches a value within the range of 40◦ to 60◦ , the curves corresponding to both ratios ηx /ηm and Ex /Em change their directions. That is, for values of θ between 40◦ and 60◦ , the ratios ηx /ηm and Ex /Em reach their extreme values (maximum and minimum, respectively) almost simultaneously. Further, by setting the fiber-aspect-ratio l/d as 5, 20, 40, 60, 80 and 100 and plotting the obtained values of the ratios ηx /ηm and Ex /Em against the fiber volume fraction Vf and the fiber off-axis angle θ, one observed that the values of the ratio ηx /ηm decrease monotonously as the fiber-aspect- ratio l/d increases. Meantime, for values of l/d > 15, the rate of decrease of the value of the ratio ηx /ηm slows down until the fiber-aspect-ratio l/d reaches 20, but the value of the ratio ηx /ηm maintains a constant value afterwards. It was further observed that the value of the ratio Ex /Em increases monotonously until the fiber-aspect-ratio l/d reaches a value of about 20. With the fiber-aspect-ratio value being within the range of 20 to 60, the value of the ratio Ex /Em increases slowly with the increase of the fiber-aspect-ratio and seems to maintain a constant value from l/d = 60 and upwards. The obtained results confirm that among the three considered variables, the fiber off-axis angle θ is the most significant influential parameter on the damping and stiffness of a fiber reinforced 203

composites. The obtained results appear to be in a good agreement with the observations made by Gibson et al. (1982), Sun et al. (1985), and Suarez et al. (1986). From the above numerical results, one could note that in order to enhance the damping of a short aligned fiber-reinforced composite, it would be necessary to sacrifice its stiffness, and vise versa. Thus, simultaneous optimization of these two properties appears to be a necessity within the scope of enhancing the mechanical performance of this class of material. Further, an important advantage of a fiber reinforced composite over conventional materials is its light specific weight. It, thus, seems meaningful to include this property as a variable in the optimization model. Accordingly, our optimization model is multi-objective and constitutes three objective functions, with the aim of maximizing the damping and stiffness of the composite and simultaneously minimizing its specific weight. The “inverted utility function method”, e.g., Rao (1984), is employed to solve numerically this multi-objective optimization problem. 3 OPTIMIZATION 3.1 Inverted utility function method In this method, a utility function Ui (fi ) is defined for each objective function as

where fi (X ) is said to be the ith objective function, with weighting factor wi , to be minimized. In the process of optimization, one can invert each utility function and try to minimize or reduce the total undesirability. The later is obtained as

where the involved parameters are defined as

The solution of this optimization problem is found by minimizing Ui−1 subject to the pertaining constraints. 3.2 Multivariable non-linear optimization In the present work, the utility functions pertaining to the optimization problem are defined as

where, as identified earlier, ηx and Ex are, respectively, the damping parameter and the storage modulus of the composite specimen in the loading direction, and W is the specific weight which may be defined in term of the fiber volume fraction Vf for the composite as

204

Meanwhile, the variables W f , W m , Em and ηm are set corresponding to the material properties of the considered short aligned-fiber reinforced composite. The total undesirability of the dealt with design problem becomes

In the present optimization problem, both damping and stiffness are considered to be of about the same significance, and each would be more influential than the specific weight variable. Consequently, one selectively set the values of the parameters appearing in expression (6) as a1 = 0.513, a2 = 0.387, a3 = 0.1. 3.3 Implementation of non-linear programming The optimization of the proposed model may proceed as

This is a typical constrained non-linear optimization problem. In order to simplify this problem, one can adopt the mapping technique, often referred to as “variable transformation technique” (e.g., Rio, 1984) to deal with the above-mentioned parametric constraints. By using the referred-to mapping technique, the constrained optimization problem could be solved via a non-constrained optimization technique. In this context, the “Simplex Method”, see Rio (1984) and Haddad (2000), was used in this research to solve the non-constrained non-linear optimization problem being dealt with here. For a given E-glass/epoxy composite material, the results of optimization of damping, stiffness and specific weight show that, approximately at fiber off-axis angle θ ≈ 43.75◦ , by setting fiber aspect ratio l/d ≈ 1.38 or l/d ≈ 85.09, the corresponding fiber volume fraction Vf reaches 62% or 54%, and one could obtain maximum damping, relatively high stiffness and relatively low specific weight for this class of material. The existence of multiple local minima, referred-to above, gives more flexibility in the design of short aligned fiber reinforced composite materials. That is, in view of the results of this research, both the micro-fiber or whisker composites (l/d ≈ 1.38) and short aligned fiber-reinforced composites with longer fiber (l/d ≈ 85.09) may be selected. It is to be mentioned, however, that some of the results presented in this paper are likely to be more of academic than of industrial interest particularly when an optimized composite with very low aspect-ratio fibers are considered. In practice, however, current composite manufacturing processes may not be able to achieve a high volume fraction as 0.60 with a controlled fiber orientation of 43.75◦ for a low fiber-aspect ratio of 1.38. High fiber volume fraction and controlled fiber orientation may be only achieved for parallel long or continuous fibers that are closely packed, e.g., prepreg tapes or filament wound parts.

4 CONCLUSIONS Polymeric fiber-composites offer definite advantages as potential “adaptive” damping material systems that can be tailored to possess a pre-determined “time-memory”. In this paper, the effects of microstructural parameters, such as fiber-aspect-ratio, fiber off-axis angle and fiber volume fraction, on the damping and stiffness of a fiber-composite system, of linear viscoelastic behavior, are first examined. Quasi-static models are, then, developed by using a “Forced Balance Approach” 205

to define the mechanical response properties of short aligned-fiber reinforced composites. Subsequently, simultaneous optimization of damping, stiffness and specific weight is carried out by using the so-called “Inverted Utility Function Method”. The obtained results show that: i. Short aligned-fiber-reinforced composites have superior design flexibility and damping properties as compared with those pertaining to long aligned (continuous) fiber-reinforced composites. Here, in order to increase the damping, it may be necessary to sacrifice the stiffness and vise-versa. ii. Fiber reinforced composites with a lower fiber volume fraction have superior energy absorbing capability compared with the ones with a higher fiber volume fraction. iii. Damping and stiffness of a short aligned-fiber composite are functions of fiber-off-axis-angle, fiber volume fraction and fiber-aspect-ratio. REFERENCES Bert, C.M. 1980. Damping applications for vibrations controls, ASME AMD-38, New York, The American Society of Mechanical Engineers: 53–63. Bert, C.M. & Clary, R.R. 1974. Composite materials: testing and design, ASTM STP 546, Philadelphia, The American Society for Testing and Materials: 250–65. Cox, H.L. 1952. The elasticity and strength of paper and other fibrous materials, British Journal of Applied Physics 3: 72–84. Gibson, R.F., Chaturvedi, S.K. & Sun, C.T. 1982. Complex moduli of aligned discontinuous fiber-reinforced polymer composites, Journal of Materials Science 17: 3499–509. Haddad, Y.M. 1995. Viscoelasticity of Engineering Materials, Dordrecht, Kluwer. Haddad, Y.M. (ed.) 1998. Advanced Multilayered and Fiber Reinforced Composites, Proceedings of the NATO Advanced Research Workshop on Multilayered and Fiber-Reinforced Composites: Problems and Prospects, Kiev, Ukraine, June 2–6, 1997, Dordrecht , Kluwer. Haddad, Y.M. 2000. Mechanical Behavior of Engineering Materials, Volumes I&II, Dordrecht, Kluwer. Rao, S.S. 1984. Optimization: Theory and Applications, New York, Wiley Eastern Limited. Suarez, S.A., Gibson, R.F., Sun, C.T. & Chaturvedi, S.K. 1986. The influence of fiber length and fiber orientation on damping and stiffness of polymer composite materials, Experimental Mechanics 6: 175–84. Sun, C.T., Gibson, R.F. & Chaturvedi, S.K. 1985. Internal materials damping of polymer matrix composites under off-axis loading, Journal of Materials Science 20: 2575–85.

206

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Non linear procedure for the analysis of FRP reinforced frames S. Coccia, U. Ianniruberto & Z. Rinaldi University of Rome “Tor Vergata”, Roma, Italy

ABSTRACT: The effect of the FRP strengthening on the global response of a portal r.c. frame is analysed in this paper. In order to account for the non linear phenomena due to the cracking in concrete, tension stiffening effects, yielding of steel and interaction at the FRP-concrete and steel–concrete interfaces, a suitable iterative methodology is developed and applied. The response of r.c. structures reinforced with different amount and location of FRP sheets, subjected to constant vertical loads and increasing horizontal actions, is judged on the basis of push-over analysis. The obtained results allow highlighting the variation of strength and ductility when different reinforcing schemes are adopted.

1 INTRODUCTION The adoption of external composite FRP materials, particularly in the framework of strengthening and rehabilitation of r.c. structures, has encountered a great favour and it is now considered a common practise, due to the advantages provided by these materials, both in terms of strength increase and easiness of application. The effectiveness of this technique, widely documented by the huge amount of theoretical and experimental works and by application on real structures, is well known and recognized, even if much care has to be devoted to the influence of FRP on the local and global ductility of the reinforced structure. The variation of the dissipative capacity of a structure reinforced with FRP is a problem of topical interest for buildings in seismic zones and requires to be studied and deepened. In previous works the authors pointed out the influence of the linear constitutive relationship of the composite materials on the global plastic behaviour of the structures (Ianniruberto & Rinaldi 2003). The possibility of variation of the strength hierarchy on simple frames through the adoption of FRP sheets has been further investigated by the authors (Ianniruberto et al. 2004). The performance, both in terms of strength and ductility, of existing two-storey concrete frame reinforced with FRP has been recently evaluated with full scale tests by (Della Corte et al. 2004). The main aim of the present paper is the analysis of the influence of different arrangements of FRP reinforcement on simple r.c. frames, through a suitable procedure for the non-linear analysis of externally reinforced r.c. structures. According to the fiber orientation respect to the element axis the FRP sheets can increase the strength or the ductility. In particular the first case is typical of a fiber arrangement parallel to the element axis, while the second one occurs when the sheets are wrapped around the element, with the location of the fibers orthogonal to the element axis. The simulation of the actual behaviour of reinforced statically undetermined structures is not a simple tool, due to the non linear effects induced by the cracks formation, steel yielding, slips at the interfaces and strain localization close to the crack. These phenomena cannot be neglected if the evaluation of the local and global ductility of the structure is required, and then a suitable non linear analysis is necessary. In this paper this aim is pursued through a procedure based on the definition of a non linear constitutive relationship between bending moment and mean curvature of a reference element, overcoming, in this way the sectional aspects and accounting for the tension-stiffening effects (Rinaldi 1998). 207

The developed methodology is validated by means of experimental data available in literature and then extended to the study of simple r.c. frames reinforced with FRP. In particular the effect of different location and amount of FRP on the global strength and ductility is analyzed and discussed.

2 NON LINEAR ANALYSIS OF FRP REINFORCED STRUCTURES The non linear analysis of r.c. structures reinforced with FRP is based on the definition, for a given axial force, of a relationship between the bending moment and the mean curvature of a reference element whose length is equal to the crack distance (Rinaldi 1998, Ianniruberto & Rinaldi 2001). Based on this local constitutive law, the stiffness of each element in which the structure is discretised can be calculated. Obviously this parameter is a function of the internal forces and then an iterative procedure is required for updating the stiffness matrix for each load step and inside the same step. The program, developed with the software MatLab, given the geometrical and mechanical properties, at each force step performs a linear analysis of the structure and the internal actions and the displacements are evaluated. For each element, based on the actual value of axial force, the moment–mean curvature relationship is defined and the element stiffness is evaluated. The global stiffness matrix is then assembled and a new displacement is determined. Finally the updated internal actions are obtained and compared with the previous ones. If the differences of the values are higher than the assigned tolerance the process restarts by defining, on the basis of the new value of the axial force, the new stiffness for the elements. In Figure 1 the flow chart related to the above described procedure is reported. The analysis stops when the convergence between two subsequent steps is reached. F= 0, q = 0 internal action with a linear elastic analysis moment-mean curvature for the actual N updated local stiffness construction of the global stiffness matrix displacements F = F + ∆F

updated internal actions

internal actions

check on the internal forces ok end of the step

ok

check on the strain levels no end of the program

Figure 1. Flow chart of the adopted non linear procedure.

208

no

3 MODEL VALIDATION The reliability of the developed program is checked through a comparison with experimental tests available in literature (Spadea et al. 1998) and with numerical outcomes obtained with a FEM program, named ATENA (Cervenka 2002), particularly devoted to the analysis of concrete structures. In (Spadea et al. 1998) the response of FRP reinforced beams, subjected to four point bending tests is reported. The geometrical details of the analysed schemes are summarised in Figure 2. In this peculiar case of statically determined structure, the proposed procedure does not require an iterative solution as the final results can be easily achieved with analytical formulations. The comparison between the analytical, numerical and experimental outcomes are reported in Figure 2, with reference to the un-reinforced beam A3 and the reinforced beams characterized by different arrangement of composite materials. The calculated curves are very close to the test ones, except for the first linear behaviour, where the numerical and proposed analytical models provide higher stiffness values. It can be noted how the models catch well both the yielding and the ultimate load, related to the concrete crush in the un-reinforced case and to the FRP debonding in the other ones. It is worth noting that in the analytical simulation of the reinforced beams, in order to take account of the FRP debonding in a simplified way, the ultimate strains in the FRP sheets εfu are reduced, as suggested by (FIB Bulletin 2001). On the contrary, the numerical model does not require the limitation of the εfu , as it is possible to simulate the behaviour of the composite–concrete interface.

P/2

150

P

1800

1200

1800

P/2

[kN]

stirrupsØ4

60 300

4700 4800 5000

numerical ATENA

50 analytical

40 30

2Ø16

25

experimental

25 20

300

2Ø16

Beam A3

10

300

0 140

0

80

P

P

[kN]

[kN] analytical

80

20

40

60

δ [mm]

80 100 120 140 160 180 200

analytical

80

numerical ATENA

numerical ATENA 60

60

experimental experimental

40

40 Beam A3.1

Beam A3.2 20

20

Steel anchors

CFRP sheets 0

0

20

40

60

80

0

δ [mm]

0

P

P

[kN]

[kN] analytical

80

40

60

80

experimental numerical ATENA

80

experimental

δ [mm]

100

analytical

100

numerical ATENA

60

20

60

40 Confinement plates

40

Beam A3.3

Beam A3.4

20

20

0 0

20

40

60

80

100

δ [mm]

0 0

20

40

60

Figure 2. Load–displacement relationship: beam A3-A3.3 in Spadea et al. (1998).

209

80

100

120

δ [mm]

q=20 kN/m

600

500 mm² Ø8/200

A

100

60

500 mm² SEZ. B-B 40

40 500

B Ø8/200

5000

20 0

500 mm²

proposed procedure

120

80

500 mm²

3000 B

numerical ATENA

40

A F

P [kN]

SEZ. A-A

0

300

2

4

6

8

10

12

δ [mm]

Figure 3. Portal frame: analytical–numerical comparison.

The very good agreement between the different procedures validates the adopted model, even if the obtain results highlight that the ultimate values of FRP strain suggested in the FIB Bulletin, if applied in presence of anchors, appear too conservative, while the adoption of the actual ultimate strain of the composite material (about 1.6%) is still unsafe. The effectiveness of the proposed non-linear procedure in predicting the behaviour of statically un-determined structures is checked through a comparison with numerical results, once again obtained with the program ATENA (Cervenka 2002). At this aim the simple r.c. portal frame of Figure 3, subjected to a constant vertical load and an increasing horizontal action, is analysed by means of a statical push-over analysis. The concrete (European C20/25) is simulated with a classical Kent and Park relationship, the steel (Italian FeB44K) is represented with an elasto-plastic behaviour with hardening, and the FRP is characterised by the classical linear elastic law. The structural response is expressed by a capacity curve (base shear vs. top displacement) and, as shown in Figure 3, the results obtained with the two models appear almost coincident. 4 FRP REINFORCED FRAMES The developed non-linear program has been used to analyze the effects of different arrangements of the composite reinforcement on the global response of a simple portal frame whose geometry is reported in Figure 4. The strengthening technique is based on the application of FRP sheets at the external surfaces of the columns. The longitudinal sheets are assumed to be fully anchored at the top and bottom sections of the columns and their width is fixed equal to 240 mm. The material properties are related to European concrete C20/25 and Italian steel FeB38K. The debonding is taken into account by means of a fictitious reduction of the ultimate strain value to a limit one equal to 0.85% (Fib Bulletin 2001). 4.1 FRP sheets bonded at the top and bottom of both columns The first analyzed case concerns the FRP strengthening of the top and bottom zone of both columns (Fig. 4) with fibers oriented along the column axes. The length of the reinforced zones is equal to one meter. The choice of this first arrangement is based on the results of the un-reinforced case, characterized by a failure of the section at the bottom of the right column. Figure 4 reports the push-over curves for many values of the FRP sheet thickness t ranging from 0.2 to 1.6 mm. The diagrams show an increase of the post-cracking stiffness of the structure at increasing value of the thickness of the sheets. The ultimate strength of the frame is always larger than the un-reinforced case, but no further increment is obtained as soon as t exceeds 0.8 mm. On the contrary the ultimate displacement of the reinforced frame reduces with respect to the reference structure except for the cases of 0.4, 0.8, and 1 mm. Moreover if the thickness is greater than 0.8 mm, the ultimate displacement starts reducing. The FRP thickness affects the location of the failure section. Indeed for t = 0.2, 0.3 mm the frame collapse is due to the attainment of the limit strain in the composite fibers in the top section 210

q = 20 kN/m

F [kN]

A

1.6 mm

F

40

SEZ. A-A

Ø8/200

un-reinforced

4000

0.3%

40

B

Ø8/200 0.3%

5000

30

SEZ. B-B FRP s=var. br=240

20 300

1000

0.2 mm

40

0.25%

B

0.3 mm

50

500

1000

0.25%

A

1.4 mm 1.2 mm 1.0 mm 0.8 mm 0.4 mm

10

300

0 0

10

20

30

40

50

60

70 δ [mm]

Figure 4. Reinforcement of the top and bottom zones of the columns and push-over curves. F [kN]

q = 20 kN/m

1.6 mm

A SEZ. A-A 40

F A

500

0.25% Ø8/200

B

40

B

Ø8/200

5000

0.3%

0.8 mm 0.4 mm

60 50 40

SEZ. B-B FRP s=var. br=240

un-reinforced

30 20 300

4000

0.25%

0.3%

1.4 mm 1.2 mm 1.0 mm

70

10 0

300

0

10

20

30

40

50

60

70 δ [mm]

Figure 5. Reinforcement of the entire columns of the columns and push-over curves.

of the right column. For t = 0.4 mm the simultaneous failure of the bottom and the top sections of the right columns, once again due to the debonding, takes place. Finally for t greater than 0.4 mm the failure is related to the concrete crush of the section at one meter from the bottom restraint. The best result seems to be obtained when t = 0.8 mm because this value of the thickness gives both the maximum strength increase and the maximum ultimate displacement. 4.2 Complete reinforcement of both columns The second reinforcement scheme is reported in Figure 5 where the sheets are bonded all over the left and right surface of the columns. As in the previous case it exists a value of the thickness, 0.8 mm, which gives both the maximum strength and displacement. The failure section for t = 0.4 mm is placed at the bottom of the right column whilst for t > 0.4 mm the right section of the beam fails in compression. 4.3 Column wrapping The third analyzed case concerns the complete wrapping of both columns. To account for the effect of the transverse composite reinforcement the ultimate strain in the constitutive law of the concrete has been modified according to Sieble et al. (1995). The obtained results are reported in Figure 6. As expected the push-over curves obtained for different values of t are perfectly coincident except for the ultimate displacement which increase with the sheet thickness. 4.4 Comparisons In Figure 7 are reported the push-over curves related to the same FRP thickness (t = 0.8 mm) obtained with the three considered reinforcing schemes. The proposed procedure allows the evaluation of the lengths of the yielded zones and then provides useful indications on the local and global ductility of the structures. In this framework 211

F [kN] 30

F [kN]

1.6 mm

0.4 mm un-reinforced

20

10

0

0

10

20

30

40

50

60

70

total column strengthening

70 60 50 40 30 20 10 0

80 90 δ [mm]

joints strengthening

Wrapping Un-reinforced

0

Figure 6. The push-over curves for different values of the wrapping sheet.

10

20

30

40

50

60 70 δ [mm]

Figure 7. Comparison between the considered reinforcing schemes.

the obtained results, not reported here for brevity’s sake, clearly shows the effects of the FRP in reducing the plastic spread along the elements. Assuming the ratio between the ultimate and yielding displacement (µ ) as a ductility index we found that it reduces from about 4 (un-reinforced scheme) to about 2 almost independently of the longitudinal sheet thickness. On the contrary the wrapping technique leads to an increase of µ which depends on t and in the examined case reaches the value of about 5. 5 CONCLUSIONS A non-linear procedure able to simulate the behaviour of r.c. structures externally reinforced with FRP has been developed and applied to a reference scheme. It is based on an analytical model developed by the authors in previous papers and allows evaluating the influence of different sheet arrangements and properties on the strength and ductility response. Tension stiffening effects, localization of steel strain and slip between the materials are accounted for. The obtained results, while confirm the effectiveness of this techniques in providing increase of structural strength, show some ductility issues when the FRP fibres are bonded along the column axis. REFERENCES Cervenka Consulting. 2002. ATENA Program Documentation, Prague, June. Della Corte G., Barecchia E., Mazzolani F. M. 2004. Seismic upgrading of existing rc structures using FRP: a GLD study case. Proc. Innovative Materials and Technologies for Construction and Restoration IMTCR04, Lecce, Italy, June. FIB Bulletin, 2001. Externally bonded FRP reinforcement for RC structures. Bulletin d’Information N. 14, Lausanne. Ianniruberto, U. & Rinaldi, Z. 2001. “Influence of FRP reinforcement on the local ductility of r.c. elements”; Int. Conference on FRP Composites in Civil Engineering, CICE 2001, Hong Kong, December. Ianniruberto, U. & Rinaldi, Z. 2003. “Global ductility of FRP reinforced frames”; FIB Symposium “Concrete structures in seismic regions”, Athens, May. Ianniruberto, U., Rasulo, A., Rinaldi, Z. 2004. “Comportamento sismico di telai rinforzati con FRP”; XI Convegno Nazionale L’Ingegneria Sismica in Italia, Genova, 25–29 Gennaio (in italian) Park, R. & Paulay, T. 1975. Reinforced Concrete Structures. John Wiley & Sons. Spadea, G., Bencardino, F., Swamy, R.N. 1998. Structural behaviour of composite RC beams with externally bonded CFRP, Journal of Composites for Construction, ASCE. Vol. 2 No. 3, August. Rinaldi, Z. 1998. Duttilità e resistenza di strutture in c.a.: influenza della localizzazione delle deformazioni nell’acciaio”; Dottorato di Ricerca in Ingegneria delle Strutture, Università degli studi di Roma “Tor Vergata”. Seible, F., Priestley, M.J.N. e Innamorato, D. (1995). Earthquake retrofit of bridge columns with continuous fiber jackets, Volume II, Design guidelines, Advanced composite technology transfer consortium, Rep. No. 95/08, Univ. of California, San Diego, USA.

212

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Stiffness of beams prestressed with FRP tendons Z. Plewako Rzeszow University of Technology, Rzeszów, Poland

ABSTRACT: Paper presents objectives and particular results of research work, which main aim is to determine influence of various FRP tendons strain characteristics on main properties of prestressed beams. To obtain results independent on particular properties of considered tendons and beams dimensionless convention of data processing and presenting of results was applied. According to this, some theroretical and numerical calculations have been developed. Their results verified by particular experimental tests were presented.

1 INTRODUCTION Fibre Reinforced Plastic (FRP) reinforcing bars, tendons and strips, which recently have been used in particular numbers of structures, show significant differences in mechanical properties comparing to steel elements. Opposite to steel, FRP tendons have various values of modulus of elasticy constant in full range of strains as well as ultimate strength. Main aims of presented project were to: – determine influence of various FRP tendons strain characteristics on main properties of prestressed beams, – predict static behavior of beams prestressed with FRP tendons, – formulate rules of calculation of beams prestressed with FRP tendons. Program of the project included: – – – – – – –

models of materials, models of considered beams, theoretical calculations, numerical calculations, analysis of calculation results, verifying experimental tests, conclusions, relations and guidelines.

To obtain general results independent of particular properties of applied tendons, concrete and dimensions, convention of relative data preparing, processing and receiving have been engaged. Generally, influence of tendon type, distinguished by its strength, modulus of elasticy and pretensioning force on flexural behavior of prestressed cross-sections and beams was obtain. To verifying these theoretical results, some experimental tests have been performed.

2 MATERIAL PROPERTIES Fibre Reinforced Plastic (or: Polymers) are made of uniaxially oriented parallel thin and strong fibres of different chemical origin, embedded in a resin matrix (Fig. 1). Names of particular types of products follow origin of fibres: Aramide FRP (AFRP), Carbon FRP (CFRP), Glass FRP (GFRP). 213

Axial strength fp and modulus of elasticity Ep of FRP fulfils “rule of mixtures” given by formulas:

where: subscripts p, f, m denotes respectively: composite, fibre, matrix ϕ – relative volume (cross-section area) of fibres σm – stresses in matrix with strains equal to: εm = εfu = ff /Ef Basic precondition of co-operation between fibres and matrix in FRP products is, that εmu ≥ εfu .

Axis of fibres

Figure 1. Scheme of FRP structure. 3500

Legend

7

σ [MPa]

3000 1 2 3 4 5 6 7 8

2500 2000

4

6 3

1500

8

1 5

1000

2

500

ε

0 0

10

20

30

Name & country POLYSTAL D KHS KROSNO PL FIBRA JAP TECHNORA JAP ARAPREE NL CFCC USA BBR CFP CH 7-wire strand UK

Typ GFRP GFRP AFRP AFRP AFRP CFRP CFRP Steel

E [GPa] 51 40 70 54 52 137 165 195

f [MPa] 1520 900 1400 1800 1250 1800 3300 1780

40

Figure 2. Comparison of different prestressing materials. Table 1. Properties of FRP vs steel. Fibre Property

Glass

Aramide

Carbon

Reinf/ prestr steel

Tensile strength Long-term strength Fatigue strength Multiaxial strength Durability in alkaline environment* Durability in acidic environment Durability in carbonated concrete Weight

+ O – – – + ++ +

+ O ++ – O + ++ ++

++ + + O + ++ ++ ++

+ + + + ++ – – –

∗ Like

in concrete.

214

Given formulas shows clearly, that engineers can create FRP product with required mechanical properties (it is a basement of material engineering science and practice). This gives new possibilities in theory and technology of prestressed structures. Also other properties of FRP products differ from steel, and gives new opportunities in application for prestressing of various structures. In order to perform calculations based on non-dimensional (relative) data material models shown in Fig. 3 were engaged.

3 THEORETICAL AND NUMERICAL CALCULATIONS OF CROSS-SECTION Stiffness of beams is correlated with its bearing capacity, and implicates deflections under service loads. Deflection increase as visible signal of reaching the bearing capacity is a very important safety alert. The problem is, how different of steel mechanical properties of FRP tendons, especially various modulus of elasticy, constant if full range of stresses, influence on stiffness of prestressing steel. To solve this problem, some theoretical considerations were carried on. The theoretical analysis of concrete cross-section material models (Fig. 3) and calculating procedures according to EUROCODE 2 was carried out. This analysis was performed for unit geometrical dimensions and acting forces (Figs. 4, 5). Its results allowed determining universal relationships between strain and stress distribution in analysed cross-sections. Using received results, numerical method of dimensionless strain and displacement analysis of prestressed cross-sections and beams was worked out. Engaging this method, influence of tendon types, their bearing capacity and pre-tensioning on properties of cross-section and beams prestressed with and without bond, was

CONCRETE

PRESTRESSING MATERIALS

σc / fc

σp / fp

1

1

CFRP

AFRP

GFRP Steel

0,9 Steel

-2

-3,5

εp0,1

εc [‰]

35

20 24 εp [‰]

33

Figure 3. Material models applied in calculations. c/fc1

ε2

mc

1 p

kc

εcp

c

q

la = 1/3

kp

1

1

Figure 4. Denotes of cross-section and beam calculations.

215

q

500

Q/2

Q/2

FRP beams

Steel beams 25

150

25

1500

100

100

Figure 5. Tested beams appearance. Table 2. Basic data of prestressing tendons in tested beams. Beam symbol S1

GFRP AFRP

Tendon type Steel strand 1Ø12.5 mm (Ø4.5 + 6Ø4.0) GFRP 2Ø8 mm GFRP 2Ø8.5 mm

Area Ap [mm2 ]

Breaking force Fp,min [kN]

Modulus of elasticy Ep [GPa]

Prestressing force P0 [kN]

91.3

162.7

195.0

32.8

48.0 81.6

39.0 47.5

28.4 22.6

2 × 50.0 2 × 56.7

obtained. As reference quantities were assumed: concrete strength – f c , cross-section dimensions d (depth) and b (width), length of beam L, tendon cross-section area – Ap , tensile strength – f p , ultimate strain εu and yield strain for steel – ε 0,1 . Related quantities necessary for calculation and results were defined as functions of these quantities. Calculating procedures were prepared in following stages: 1. Developed procedure calculating ε2 , εcp for given mcp and kp (kp depending on εcp ). This procedure allowed to determine plain state of strain in pre-stressed cross-section subjected to bending moment mcp , considering change in pre-tensioning force kp due to change in reinforcement strain. 2. Developed procedure, which calculates deformations of cross-section for full range of bending moment for given parameters of pre-stressing: tendon type and force. The calculations simulated changing in state of strain (and stress) in discussed cross-section subjected to incremental loading by bending moment. 3. Developed procedure, which calculate deflection of beam for full stage of acting load and numerical simulations of bending test of simple beam (plotting load–deflection curve). All calculations were done for uniform dimensions of beam and cross-section. To compare influence of mechanical properties variety for different kinds of reinforcement, equal value of load bearing of tendons kpf and initial prestressing ratio ωp0 were assumed. 4 EXPERIMENTAL TESTS To verify results obtained in theoretical analysis, experimental tests were provided (Fig. 5, Table 2). Stress–strain curves obtained in load tests showed good conformity with results of calculations for theoretical models (Fig. 6). 5 CONCLUSIONS Figure 7 presents deflection vs load behaviour of theoretical models of four beams. All assumed models had equal geometrical dimensions and concrete strength, were prestressed with tendons of 216

Model

Test 40

GFRP

Q [kN]

35

Test 40

30

30

25

25

20

20 Q

15

10

5

5

0 4

8

12

16

20

24

Q

15

10

0

Q [kN]

35

Model

AFRP

28

32

0

a [mm]

0

4

8

12

16

20

24

28

a [mm]

Figure 6. Deflection vs load curves for beams prestressed with different type of tendons. Test and model results. σp= fs0,1

Bending moment M

Mult

Mcr Steel GFRP AFRP CFRP Deflection

Figure 7. Deflection vs bending moment M curves for theoretical beam models.

different origin and cross-section area, but developed the same prestressing force (pre-tensioned to equal force) and the same bearing capacity (breaking load). The only actual difference in beam models lied in various modulus of elasticy of tendons as an effect of different ultimate strains. Analysis of plots leads to following general conclusions: – beams subjected to loads developing bending moment M which value is lower than cracking moment Mcr of concrete cross-section, have the same stiffness independently of applied prestressing tendon type. – for cracked beams, M > Mcr , stiffness of beams strongly depends in proportion on tendon stiffness. For resulting strain increase of tendons lower than yielding strain of steel, stiffness of beams prestressed with FRP is smaller than prestressed with steel tendons. Oppositely, when strain increase involve yield strains in steel, stiffness of beams prestressed with FRP is smaller than prestressed with steel tendons ones. Consequently, bearing capacity of beam prestressed with FRP tendons could be equal as for beam prestress with steel tendons. Lower and constants in full range of stress modulus of elasticy of FRP tendons and described above course of beam bending under load caused large deformations accompanying to failure. In beam with relatively smaller FRP prestressing reinforcement, failure could be caused by sudden tendon break, differently from beam prestressed with equivalent steel. This situation is determined 217

ωp0 = 70%

ωp0 = 50% 1 2 3 1 2 3

Steel 1 GFRP 2 AFRP* 3 CFRP*

Steel 1 GFRP 2 AFRP 3 CFRP

Figure 8. Course of beam bending example for theoretical models with various ratio of prestressing ωp0 = 50% and ωp0 = 70% (marks * following tendon type denotes tendon break failure mode).

by ratio of prestressing ωp0 = (prestressing force P0 )/(breaking load Fp ), which is approximately irrelevant to bearing capacity of beams with steel tendons (Fig. 8). Consequently, additional conclusions resulting from theoretical analysis are: – Assuming safety of structure as most important requirement, bearing capacity and pre-tensioning of applied tendons should assure failure mode by concrete compression (over-reinforced beams). In this mode large deflection, as a signal of beam overloading occurs. Fulfilling this postulate require application of FRP tendons with bearing capacity Fp higher than of steel once, and with significantly lower pre-tensioning P0 (for FRP P0 = 40%/50%Fp , for steel P0 = 70%Fp ). – Direct relation between tensile strength and ultimate strains for FRP tendons results from constant value of modulus of elasticity. This relation requires exact calculation of structure deformation and corresponding tendon elongation in ultimate limit state of beam subjected to bending. – Verification of theoretical analysis by test results proved that assumption of material models and calculation procedures according to EUROCODE 2 ensure good conformity with experiment in full range of loading. This points, that EUROCODE 2 could be applied for analysis and designing of beams prestressed with FRP tendons. REFERENCES Plewako Z. 2000. Ultimate Limit States of Beams Prestressed with Composite Tendons (PhD Thesis) Rzeszow: Rzeszow UT. Eurocode 2, Dec. 2002. Design of structures. Part 1: General rules and rules for buildings EN 1992-1-1 CEN/TC 250/SC2 N0465.

218

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Strengthening timber beams with prestressed artificial fibres: the delamination problem Maurice Brunner & Marco Schnüriger School of Architecture, Civil and Wood Engineering, HSB, Biel, Switzerlands

ABSTRACT: Delamination is a major problem when glulam beams are to be strengthened by attaching prestressed, high-strength fibres on the tensile face. In a COST Project, the authors analyse two possible solutions. The first solution approach involves the use of a special device to attach the prestressed fibre in stages starting from the centre of the beam. The second approach involves the development of a “ductile” adhesive to spread the prestressing force over a larger area at the ends of the beam. The tests with the prestressing device yielded satisfactory results. The tests with different adhesive mixes yielded some interesting results of a pioneering nature, however the anchorage strength fell short of expectations. Themes could be identified for more research work.

1 INTRODUCTION A method sometimes used to improve the load-bearing capacity of glulam beams involves the inclusion of high-strength artificial fibres on the tensile face ([1], [2]). The practical experience shows that the high strength of these fibres cannot be fully utilised when they are simply glued onto the beam. Thus some researchers are studying how to prestress the fibres in order to make better use of their high strength and at the same time reduce the quantity of fibre needed and the costs. Glulam beams loaded in the direction of the wood grain exhibit high strength. Theoretically, it should be possible to apply very high prestressing forces on relatively small cross sections. In practice, only a small prestressing force can be applied because of delamination dangers caused by the force transmission from the artificial fibres at the ends of the beam into the main beam body. When the usual adhesives on the market are used, the prestressing force is anchored over a short distance at both ends of the beam. The resulting, concentrated force transmission induces high stresses in the direction perpendicular to the grain, where the cracking energy of timber is low. This leads to a premature delamination along the glue line, or to cracking of the timber. As Fig. 1 shows, the danger of delamination is also well known when concrete beams are similarly strengthened.

Crack

τ

Figure 1. Delamination: a prestressed fibre is torn away from the concrete beam.

219

Figure 2. Distribution of shear strees τ in the load-transmission zone.

Figure 3. EMPA Prestressing device with a gradiented anchoring system (Stöcklin & Meier [3]).

Figure 4. Strain distribution of a prestressed CFRP strip showing the gradiented prestressing force along the beam length.

Timber fibre stiff adhesive

τ = shear stress

soft or ductile

Figure 5. Possible influence of adhesive type on the distribution of shear stress at the ends of a beam strengthened with prestressed artificial fibres

Acting within the framework of the COST action E13 “Wood adhesion and glued products”, the authors initiated a research project to analyse two promising approaches to the delamination problem. The research was concerned with the following two themes: • The first theme was that the gradiented anchoring technique developed for concrete structures could also be successfully applied to timber as well. • The second theme was that the transmitted force could be spread over a larger area at the beam ends with the help of “ductile” adhesives. 2 POSSIBLE SOLUTIONS TO THE DELAMINATION PROBLEM 2.1 Gradiented anchoring technique The timber industry can learn from the more advanced research work on the strengthening of existing concrete structures. In Switzerland for example, the EMPA (Swiss Federal Materials Testing Institute) has developed a special prestressing device, the gradiented anchoring system [3], to apply a prestressing force in stages. In the first stage, the middle section of both the prestressed fibre and the beam are attached by activating the adhesive with heat. The force is then slightly reduced and the next section of the fibre is attached to the beam. In effect, the prestressing force is spread over a certain length of the beam, thus reducing the peak stresses (Figs. 3 and 4). 2.2 Development of a ductile adhesive The EMPA device uses an epoxide adhesive which is seldom used in the wood industry. Theoretically, the sharp stress concentration at the beam ends could also be reduced by spreading the transmission force over an increased area with the help of “ductile” adhesives (Fig. 5). The authors decided on a feasibility study using adhesives better known in the wood industry. 220

Experiments have been carried out with glulam beams strengthened with prestressed carbon, aramide or glass fibres. Luggin [4] reports on the practical difficulty to find a simple and reliable method to prestress the fibres: indeed, many of his test specimens failed on delamination. There are calculation models for the delamination phenomenon. Holzenkämpfer [5] gives a fine overview of the many refined mathematical models which have been developed for the analysis of the force transmission from the prestressed fibres into the main body of the beam. Triantafillou & Deskovic ([6], [7]) describe how delamination could be overcome by using “ductile” adhesives which, unlike the usual stiff adhesives on the market, could spread the prestressing force over a larger area at the ends of the glulam beam where the fibres are attached. 3 RESEARCH PROGRAM The research program comprised the following activities: 1. Search for a “ductile” adhesive. Acting on the advice of a well-known expert on adhesives, two basic adhesives were selected: an epoxide and a polyurethane component. New variants were created by mixing the two components in different proportions. The various mixes were used to glue two pieces of wood together which were tested in tensile shear according to the European standard EN205. Of particular interest was the force–deflection-curve and the indication of some ductile behaviour. The most promising mixes were selected for the second test series. 2. Specimens of carbon fibre glued to timber boards with the promising glue mixes selected from the shear tests above were submitted to a tensile test. The aim of the test was to select the mix which would permit a large prestressing force to be safely attached to a glulam beam. 3. A number of glulam beams prestressed with fibres attached by using the selected glue mix were tested in bending. 4. In the final test series, a number of glulam beams were strengthened using the EMPA technique of gradiented prestressing force. The strengthened beams were tested in bending. 4 TESTS 4.1 Analysis of adhesives mixes and tensile/shear tests Two main components on epoxide and polyurethane basis were mixed in different proportions and their characteristics, in particular the hardening times, were determined. Two mixes (4 and 8) exhibited a satisfactory hardening time and were chosen for tensile shear tests. For control purposes, specimens were also made with an adhesive of pure epoxide. Gustafsson [8] describes a number of different possibilities to determine the shear resistance of a glued connection. Fig. 6 shows the set-up chosen for the shear tests. The three different glue mixes were applied, first in thin layers (less than 0.1 mm) as used in industry, and then in thicker (0.6 mm) layers. A total of about 20–30 specimens were used for each of the 6 test series. The results of the shear tests, shown in Fig.7, can be summarized as follows: • The load–deformation-curves of all test specimens display an initial steep rise followed by a gentler slope. After the maximum load has been attained, the force falls abruptly. • All the tested specimens exhibited similar slopes of the force–deformation curve. • Wood failure usually occurred due to shear at the bonding surface. • The failure load depended upon the glue: the best glue mix 8 exhibited failure loads much higher than those achieved by either the 100%-epoxide or mix 4. 4.2 Pull-off tests The glue mix 8, which exhibited the best results for the shear tests, was selected to attach a slack laminate of carbon fibre to a timber board. For control purposes, further specimens were made using 221

Figure 6. Set-up for shear tests according to German standard DIN prEN 205.

Figure 7. Force–deformation-curves of the shear tests (the median values of the different adhesive types with thin glue layers are shown here).

Figure 8. Set-up of tensile tests.

Figure 9. Picture of failed bond.

as adhesive pure epoxide and pure polyurethane. Tensile tests were performed on 10 specimens to determine the pull-off force for the laminate of carbon fibres attached to the timber board. Thin glue layers were used. The specimens were stabilised against lateral deformation in order to avoid tension perpendicular to the grain. The test set-up is shown in Fig. 8, whilst Fig. 9 shows the failed bond. Due to the favorable results of the tensile shear tests, there were high expectations of the mix 8. Unfortunately, the initial promise could not be confirmed in the pull-off tests. All the test specimens of the three different glue mixes yielded similar results. A maximum prestressing force of about 30 kN could be safely transfered without fear of delamination.

4.3 Bending tests with prestressed glulam beams There was a need to verify if it would be possible to attach prestressed artificial fibres safely to a glulam beam of practical size without delamination. Bending tests were performed to demonstrate the increased effectiveness of strengthening glulam beams with prestressed fibres. Due to the positive results of the preliminary tests, the glue mix 8 was chosen to attach a prestressed laminate to a glulam beam. Unfortunately, the attempt was a complete failure: the prestressed force of 30 kN died away within five days because the adhesive could not hold. In a second series of tests, glulam beams were strengthened with a carbon laminate prestressed with the gradiented anchoring system to 60 kN, which is the force normally used in connection with concrete beams. The adhesive prescribed for the system is a specially designed epoxide. The experiment was a complete success because there were absolutely no signs of delamination or of creep. 222

Table 1. Test specimens. Test series

Prestressing force

Adhesive used

Naked beam Carbon laminate (FRP) slack Carbon laminate prestressed

– 0 kN

– Epoxide

60 kN Gradient-system

Epoxide

Dimensions Beam: Height 140 mm Width 200 mm Length 4000 mm

Carbon laminate FRP: 50×1.14 mm length 4000 mm

Figure 10. Four-point bending test according to German standard DIN EN 408.

Table 2. Summary of the bending tests showing the positive results of strengthening glulam beams with slack and prestressed FRP strips. Bending stiffness EI [N.mm2 ]

Ultimate bending moment M [kNm]

Serie

Average values

Increase (%)

Average values

Increase (%)

Naked glulam With FRP slack With FRP prestressed

1.20 E12 1.43 E12 1.46 E12

0 +19 +22

41.6 50.4 54.9

0 +21 +32

In a third series, glulam beams were strengthened with unstressed carbon laminate. For control purposes, some naked glulam beams of the same size were also prepared. Table 1 lists the test specimens used, whilst Fig. 10 shows the set-up of the bending tests which were carried out in accordance with the German standard DIN EN 408. The results of the bending tests are listed in Table 2. They demonstrate how a glulam beam could be strengthened with a carbon fibre laminate both unstressed and prestressed. The tests also demonstrate that the strengthening with a prestressed laminate is more effective than the use of a slack laminate: the bending resistance of the naked glulam beam was increased by 32% when strengthened with a prestressed laminate as against 21% when the laminate is not prestressed. 5 CONCLUSIONS The project aim was to analyse two possible solutions to the delamination problems which may occur when glulam beams are to be strengthened with prestressed artificial fibres of high strength. The first solution proposal was concerned with the gradiented prestressing device developed in Switzerland by the EMPA to strengthen existing concrete beams. Although the prescribed epoxide adhesive is seldom used in the timber industry, the tests demonstrate that the system could also be successfully used to strengthen glulam beams with prestressed artificial fibres of high strength. The second solution proposal was concerned with the adaptation of adhesives which may be more acceptable to the timber industry. It was a feasibility study for the creation of a “ductile” 223

adhesive which, due to its “yielding” at a high stress level, might be able to spread the area of force transmission from the prestressed artificial fibre into the main body of the beam, thus reducing the dangereous peak stresses which occur when the normal stiff adhesives on the market are used. The initial tests yielded interesting results of a pioneering nature, however, the final tests ended in failure. The prestressing force of 30 kN could not be maintained and died away within 5 days. The reasons could not be fully clarified within the framework of the COST E13 project. Possibly, they may have something to do with an inadequate curing of the new glue mix 8.

REFERENCES [1] Tingley D., FIRP Reinforcement Technology Information Packet, Science and Technology Institute, Corvallis OR, USA, 1995 [2] Romani M., Blass H.J., Design model for FRP reinforced glulam beams, International Council for Research on Innovation in Building and Construction, Working Commission W18 – Timber Structures, Meeting 34, Venice, Italy, August 2001 [3] Stöcklin & Meier U., Strengthening of concrete structures with prestressed and gradually anchored CFRP strips, IABSE International Conference, Malta 2001 [4] Luggin W. F., Die Applikation vorgespannter CFK-Lamellen auf Brettschichtholzträger, Dissertation, Universität für Bodenkultur, Vienna, Austria, 2000 [5] Holzenkämpfer P., Ingenieurmodelle des Verbunds geklebter Bewehrung für Betonbauteile, Deutscher Ausschuss für Stahlbeton, Heft 473, Beuth Verlag 1997 [6] Triantafillou T.C., Deskovic N., “Innovative prestressing with FRP-sheets: Mechanics of short-term behaviour”, Journal of Engineering Mechnanics, Vol. 117, Nr. 7, July 1991, pp1652–1672 [7] Triantafillou T.C., Deskovic N., “Prestressed FRP-Sheets as external reinforcement of wood members”, Journal of Structural Engineering, Vol 118, Nr. 5, May 1992, pp1270–1284 [8] Gustafsson J., Tests and Test Results on Mechanical Properties of Adhesive Bond Lines, Chapter 2 of Final Report, COST E13, Version 4, Jan. 2000 [9] Brunner, M. Schnüriger, M. Strengthening of timber beams with prestressed artificial fibres: solution proposals for delamination, WCTE-2004, Lahti, Finnland

224

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Time depended behavior of steel – reinforced glue – laminated timber beams, regarding rheology D.N. Partov & V.K. Kantchev VSU“L.Karavelov”, Sofia, Bulgaria

ABSTRACT: The paper presents analysis of the stress changes due to creep in statically determinate reinforced wood beams. Each beam consist of glue-laminated timber 2 – section, acting compositely with steel rods, or steel-plate; U-profile, symmetrical or unsymmetrical attached to the upper or lower surface of the beams. The mathematical formulation of this problem involves the equation of equilibrium, compatibility and constitutive relationship, i.e. an elastic law for the steel part and an integral-type creep law for the wooden part. For determining the redistribution of stresses in beam section between wood beam and steel part with respect to time “t”, Volterra integral equations of the second kind have been derived, on the basis of the theory of the viscoelastic body of Boltzmann – Volterra. Analytical method, which makes use of Laplace transformation and numerical method, which makes use of quadrature formulae for solving these equations, are proposed. The computer programs are realizes in environment of a high-performance language for technical computing MATLAB®. Some relevant examples with the model proposed are investigated and discussed. In this mathematical model, different creep function are assumed and compared by describing of the time depended behavior of the wood. Finally, this analysis show the way how to be integrated the advantages of the highly perfect model of visco-elastic body, describing the creep of wood, and availability of powerful software products. The proposed methods give us the possibilities for realistic estimates of the behavior of the reinforced glue-laminated wood beams, subjected to sustain service load.

1 INTRODUCTION The deformation of a material over time at constant load is commonly known as rheological properties of the material or creep. Wood is a viscoelastic material (see Rautenschtrauch, K. & Becker, P. 1998; Schniewind, A. P., 1968) and therefore, creep must be accounted for in the design of a wood structure when sustained loads are present. As early as a turn of the XX-century, creep was acknowledged in the design of wood structures. At the 1903 Annual Convention of the American Society of Civil Engineers, Hatt et al. (1903) stated in a presentation that it was generally known “that the deflections under ordinary quickly applied load in a test are only one-half of those, resulting from the continued application of the same load”. The authors, however, recommended further study to better determine a quantitative relation between short-term and long-term behavior. The National Forest ProductsAssociation Design Specification for Wood Construction suggested a creep factor of 1.5 for glued laminated timber and seasoned sawn lumber, and 2.0 for unseasoned sawn lumber. It means that: deflection due to long-term or permanent loads is assumed to be 1.5 to 2.0 times immediate elastic deflection caused by the load. The factor of 1.5 for seasoned sawn lumber was found to be nonconservative for loads of duration greater than approximately 2 to 3 weeks. The factor of 2.0 for unseasoned lumber was found to be sufficient for durations of load of 2 months up to 50 years. 225

b

h

h

h b

b

Figure 1. Cross-section wooden beams with symmetric reinforcement.

Most recently, Fridley et al. (1992) investigated the creep behavior of lumber, tested to failure for load-duration purposes. The tests ranged in duration from a few minutes to 1.5 years. The general creep model developed by Fridley et al. (1992) for lumber is based on the four-element Burger model. Wood is the oldest and one of the most widely used structural materials. Improved design methods and advanced forms of wood construction, involving the use of reinforcement to enhance the mechanical properties of wood members, can enable contemporary and advanced forms of large wood structures (e.g., long-span bridges – Kostelianec, B. A. & Kartopolcev, V. M., 1997) to be at least as reliable and economically competitive as those constructed from other construction materials, such as concrete, steel, and plastic. Several attempts to reinforce wood elements have been reported in the literature – Hoyle (1975) reinforced laminated wood beams with aluminum strips placed between selected laminations. Lantos (1970), Dziuba (1985), Bulleit (1984), Bulleit et al. (1989), investigated reinforcement of wood with both square and round cross-section rods. High-strength steel wire embedded in an epoxy matrix has bean used to replace tension laminations of wood beams – Krueger and Kobetz (1973, 1976). Another method of reinforcing wood is to use prestressed steel reinforcement. One unique attempt has been made to prestressed glu-lam using pretensioned steel plates bonded on the tension face with epoxy adhesive – Peterson (1965). Glass fiber-reinforced plastics have been used as faces of woodcore sandwich beams (e.g. Biblis 1965). In a recent work, Plevris and Triantafillou (1992) provided a comprehensive study of the short-term flexural behaviour of wood beams and beam columns reinforced with unidirectional fiber-reinforced plastics sheets, bonded on the member’s tension face only. Three years later they provided a basic understanding of the creep behaviour of wood members reinforced with fiberreinforced plastic materials epoxybonded to the tension faces. The works done to date on reinforced wood has focus only on short-term and long-term response on deflection. But it is known that the influence of creeping on the behavior of a composed structures, in the case of reinforced wood beam, appeared by the distribution of the internal forces between reinforce and the wood beam (Nowacki, W., 1965). With another words while in the steel reinforcement beam, under the effect of sustained service loads we see only elastic deformation, in the wood beam during the time significant plastic deformation takes place as a consequence of creep of wood. As a result of these deformations and because of the stiff connection between the two elements of the wood and the reinforcement in the beam in every cross section, (Fig. 1, Fig. 2), subjected to the effect of constantly operating outside bending moment M0 in the time t there arises a new additional group of moments Mw (t) and Ms (t) in symmetric reinforcement beam, (see Fig. 3 in Partov, D. et al., 2003) and normal forces. Nw (t) and Ns (t) in non symmetrical case (Fig. 3). The influence of this group of moments and normal forces of the general stress conditions of the statically determinate wood beam is expressed by decrease of the stresses in the wood beam and the increase of the stress in the symmetric or non symmetric reinforcement. 226

h

h

h

b

b

b

b

h

h

h

b

b

Figure 2. Cross-section wooden beams with non symmetric reinforcement. ε ε

ψ

ψ ε

ε ψ

ε

ε ε

ψ

ψ

Figure 3. Model of deformation of cross-section of steel-wood beams during the creep.

2 THEORY The theory implies the following assumptions to be true: (a) (b) (c) (d)

Bernoulli’s hypothesis concerning plane strain of cross section. Wood is uncracked. Hook’s law applies to steel reinforcement as well as to wood under short time loads. In the range of serviceability loads, wood behaves in a way allowing to be treated as a linear viscoelastic body of Bolztman – Volterra (see Prokopovich, Y. E. 1978; Partov, D. et al. 1985, 1986). The stress–strain behavior of wood can be described, with sufficient accuracy by the integral equation (1):

227

(e) (f ) (g) (h)

where ϕ(t − τ ) = ϕ∞ · f (t − τ ) is called a creep function and ϕ∞ – ultimate value of creep coefficient. The function f (t − τ ) – where t is the time interval during which the structure is under observation (τ – running coordinate of time), characterizes the process of creeping. The modulus of elasticity of wooden beam is invariant in time. Ew (τ ) = Ew (t0 ) = Ew (2) The moment of inertia of the reinforcement is slighting small in comparison with moment of inertia of the wooden beam, by symmetric reinforcement. The moment of inertia of the steel plate is considered in the moment of inertia of the composed wood-steel beam.

3 SYMMETRIC REINFORCEMENT CASE 3.1 Solution The Volterra integral equation according to the Figure 1 and its solution for the unknown Mw (t) is derived in Partov, D. et al., (2003):

Figure 4. Changes of the moments Mω (t) = Ms (t) as a result of the creeping.

Figure 5. Changes of the stresses σω (t) as a result of the creeping.

228

3.2 Numerical experiments Let us consider a practical example applying the described above method for reinforced glued laminated wooden beam. On Fig. 4, it is shown the diagram of changing of the Moment Mω (t) = Ms (t) as a result of the creeping of the wood using two different creeping functions. By using the creeping low ϕ(t − τ ) = 0.5(1 − e−1.3(t−τ ) ) it is analysed a practical example published in Partov, D. et al., 2003. By using the more precise creeping low ϕ(t − τ ) = 0.5(1 − 0.11e−0.08(t−τ ) − 0.89e−5.5(t−τ ) ), it is achieved more realistic describing of the process at its starting. On Fig. 5, it is shown the diagram of changing of the stresses σw (t) as a result of the creeping of the wood, according to the second creep function. From the analytical solution and its graphic it can be seen that the solution in both cases diverges to one and the same expression when the argument t tends toward infinity. 4 NON SYMMETRIC REINFORCED CASE 4.1 Solution As a result of our analysis for the normal force Nw (t) and bending moment Mw (t), according Fig. 3 a system of two linear integral Volterra equations of second kind are derived. The system of the integral equations of Volterra of second kind for the unknown functions are solved with Laplace transformation :

4.2 Numerical experiments We shall compute an another practical example, (Fig. 6) using by reinforcement of existing floor beams, applying the described above method. In Fig. 7 there is the diagram of the normal force Nw (t) = Ns (t), moment Mw (t), Ms (t) and stress changing as a result of the creeping of the wood, using the creeping function ϕ(t − τ ) = 0.5(1 − e−1.3(t−τ ) ) (see Prokopovich, Y. E. 1978). The creep coefficient ϕ = 0.5 is experimentally determined and depends on the moisture content of the wood as well as on the surrounding temperature. 5 CONCLUSION From the results in Table 1, we conclude that the stresses of the steel plate increases with almost 8.62% as a result of wood creeping in comparing with the stresses in time t = 0 (experimentally proved by Kaperski, U. N., Schuko, V. U., 2001). 229

Figure 6. Cross-section of reinforced wood beam.

Figure 7. Changes of the normal forces Nw (t), Ns (t) and moments Mw (t), Ms (t).

From the Table 1a, we conclude that the stresses of the steel profile -14 increases with almost 14.12% as a result of wood creeping in comparing with the stresses in time t = 0. The proposed method gives us the possibilities for realistic estimates of the behavior of the composed wood-steel beam, subjected to sustained service load. 230

Table 1. Stress changing as a result of creeping. In the wood (MPa)

In the steel part (MPa)

Stress

σwu

σwd

σsu

σsd

In time: t = 0 In time: t = ∞

−12.99 −12.47

+3.020 +2.062

63.510 62.877

85.930 93.344

Table 1a. Stress changing as a result of creeping. In the wood (MPa)

In the steel profile (MPa)

Stress

σwu

σwd

σsu

σsd

In time: t = 0 In time: t = ∞

−12.72 −12.02

+3.43 +2.54

−17.59 −41.76

84.30 96.20

REFERENCES Biblis, E. J. 1965. Analysis of wood-fiberglass composite beams within and beyond the elastic region. Forest Products Journal, Vol. 15, No 2: 81–88 Bulleite, V. M. 1984. Reinforcement of wood materials: a review. Wood and fiber Science. Vol. 16. No 3: 391–397 Bulleite, V. M. Sandberg. L. B. and Woods, G. J. 1989. Steel-reinforced glued laminated timber. Journal of Struct. Engineering, ASCE. Vol. 115. No 2: 433–444 Dziuba, T., 1985, The ultimate strength of woodenbeams with tension reinforcement. Holzforschung und Holzverwertung, Vol. 37, No 6: 115–119 (in German) Fridley, K. J., Tang, R. C. and Soltis, L. A. 1992. Creep behavior model for structural lumber, Journal of Structural Engineering, ASCE, Vol. 118, No 8: 2261–2277. Hatt, W. K., von Schrenk, H., Langa, G., Johnson, A., l. and Russell, S. 1903. Timber tests. Transactions of ASCE, Vol. LI: 67–100. Hoyle, R. J. 1975. Steel-reinforced wood beam design. Forest product Journal, Vol. 25, No 4: 17–23. Kaperski, U. N. and Schuko, V. U. 2001. Panels of a coating with the reinforced wooden frame, Proc. of the Odessa State Acad. of Building and Archit., Odessa: 109–113 (in Russian) Kobetz, R. V. and Krueger, G. P., 1976. Ultimate strength design of reinforced timber: biaxial stress criteria. Wood Science, Vol. 8, No 4: 252–262 Kostelianec, B. A. and Kartopolcev, V. M., 1997. Wooden bridges on motor roads in Russia, Izvestia Vuzov, Stroitelstvo, Vol. 40: 89–93. Krueger, G. P., 1973. Ultimate strength design of reinforced timber: state of art. Wood Science, Vol. 6, No 2: 175–186 Lantos, G., 1970. The flexural behavior of steel reinforced laminated timber beams. Wood Science, Vol. 2, No 3: 136–143 Nowacki, W., 1965. Theorie des Kriechens. Fr. Deutige Verlag, Wien. Partov, D., Dimitrov, T. T., Tschernogorov, G. V. and Kaltschev, P. G. 1985. Spannungsaenderugen infolge von Kriechen und Schwinden bei statisch bestimmt gelegerten Stahlverbundtraegern, Der Stahlbau, Vol. 54, No 7: 205–209 Partov, D., Dimitrov, T. T., Tschernogorov, G. V. and Kaltschev, P. G., 1986. Numerical Analysis of Creep and ˇ Shrinkage of Concrete of Statically Determinate Composite Beams, Stavebnicky Casopis, Vol. 34, No 9: 649–662 Partov, D., Straka, V., Kantchev, V. and Marinov I. 2003. Untersuchung von bewehrten Holzträgern unter Berücksichtigung des Einflusses des Kriechens des Holzes, Bautechnik, Vol. 80, No 3: 169–173 Peterson, J. 1965. Wood beams prestressed with bonded tension elements. Journal of Structural Engineering, ASCE, Vol. 91, No 1: 103–119 Plevris, N. and Triantafillou, T. 1992. FRP – reinforced Wood as Structural Material, Journal of Materials in Civil Engineering, ASCE, Vol. 4, No 3.

231

Prokopovich, Y. E. 1978. Osnovy prikladnoj linejnoj teorii polzuchesti. (Fundamental studies on alication of linear theory of creep), Visschaya schkola, Kiev, UdSSR, (in Russian) Rautenschtrauch, K. and Becker, P. 1998. Zur Beruecksichtigung des Kriechens bei Druckstaeben aus Holz. Bautechnik, Vol. 75 , No 11: 910–921. Schniewind, A. P. 1968. Recent progress in the study of rheology of wood. Wood Science, Vol. 23, No 2: 188–206.

232

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Timber–concrete-composite with an adhesive connector Maurice Brunner & Marco Schnüriger School of Architecture, Civil and Wood Engineering, HSB, Biel, Switzerlands

ABSTRACT: In timber–concrete-composite structures the connection is traditionally achieved with mechanical means. The Swiss adhesive producer SIKA has developed a special adhesive which is capable of bonding the concrete both wet and hardened. Within the framework of the project, the wet process was researched. The first project phase was concerned with the possible displacement of the adhesive when the fresh concrete is poured onto the wet adhesive. Key parameters such as the concrete type and the falling height of the concrete were analysed. After the adhesive had been freshly applied onto the wood, a certain time delay was allowed for the adhesive to stiffen before the concrete was poured: this “stiffening time” also proved to be a key parameter. After the production parameters had been optimised, a number of timber–concrete-composites slabs were cast and tested in bending. The test results fully met the expectations. In a third phase, the suitability of the system for a reduced gluing area (e.g. T-beams) is scheduled to be tested. 1 INTRODUCTION Timber–concrete-composite structures are becoming important in many parts of Europe because they bring many advantages when compared to traditional timber structures. The positive properties of both materials can be selectively applied: under a bending moment, the timber takes up the tensile force whilst the concrete takes up the compressive force. Furthermore, the timber brings advantages with regard to the appearance and the building physics aspects, whilst the connection with the concrete improves the performance with regard to fire, noise transmission and vibration problems. Up to date, the connection between the timber and the concrete has been achieved with mechanical means such as screws, bolts or with a concrete indentation into the wood. A glued connection is also possible. Indeed, the use of an adhesive could distribute the shear forces uniformly over the entire surface and thus avoid the force concentrations which are inevitable when mechanical connections are used. The adhesive connection is also quite slip-free, which could help to reduce the deflections. Between 1998 and 2000, a feasibility study was made at SWOOD (now renamed “HSB”) using the “wet” production process, where the fresh concrete was poured onto the still wet epoxidebased adhesive. Many timber–concrete-composite slabs with different thicknesses of the timber and concrete members were cast and tested in bending. Most of the test specimens failed at the expected loading capacity. However, a small but significant number of the specimens exhibited premature bonding failure. The feasibility study ended without a full clarification of the reasons, although it was observed that in some cases the adhesive seemed to have been locally displaced. It was thus deemed necessary to carry out further clarifications in a follow-up project within the framework of the COST-Action C12. 2 AIMS OF THE PROJECT, KEY PLAYERS The COST C12-project had the following main aims: • To clarify the reasons for the premature bonding failure observed in some of the tests of the feasibility study in order to optimise the manufacturing process. 233

• To supplement the test results of the feasibility study for timber–concrete-composite slabs. • To study the possible limitations of the adhesive bond when the contact area is reduced such as in T-beams (e.g. concrete slab on timber beam). The project was carried out with the support of industrial partners who not only supplied the test specimens but also ensured that the parameters studied were professionally realistic. The industrial partners already have plans to build a pilot structure using the new technique. The key players in the project were: • Researchers: HSB and EMPA (Swiss Federal Research Laboratories) • Industry: Schilliger AG (glulam manufacturer), Küng AG (timber construction), SIKA AG (adhesive manufacturer) 3 INVESTIGATION 3.1 Project plan The project was divided into 4 main phases: 1. Analysis of the possible displacement of the wet adhesive during the pouring of the concrete. As possible governing parameters, different manufacturing conditions as well as different timber and concrete dimensions were used to prepare small test specimens. These were analysed visually and with bending and shear tests. The main aim of this phase was to help optimise the manufacturing process and to gain some guidelines as to favourable thicknesses of the timber and concrete members. 2. The optimised manufacturing process of the 1st project phase was used to cast large-scale timber– concrete-composite slabs of different dimensions. The aim of this project phase was to clarify if the danger of premature bonding failure had indeed been eliminated and favourable results could be obtained in bending tests. 3. Study of the possible limitations of the adhesive bond in cases of reduced contact area, such as in a ribbed slab. Bending tests 4. Analysis of the long term deflections (creep, shrinkage) At the time of writing the project phases 1 and 2 had been completed whilst the phases 3 and 4 had been started but were not yet complete. The project partners intend to build a pilot structure after the project has been completed. HSB hopes to be able to provide some scientific support including site measurements. 3.2 Study of the displacement of the adhesive 3.2.1 Parameters and materials Many parameters were evaluated which were deemed to have a possibly governing influence on the displacement of the wet adhesive during the pouring of the wet concrete. Finally, test specimens were produced with the following parameters: • Constant parameters: 1. Dropping height of the wet concrete: 50 cm 2. Adhesive amount: 925 g/m2 3. Timber: 3-ply slab, 30 mm • Variables: 1. Concrete quality (consistence, use of vibrators): – Normal concrete B35/25 (C20/25) – SCC (self-compacting concrete: no vibrators needed) 2. Time delay between mixing adhesive and pouring the concrete: – 15 minutes – 90 minutes 3. Concrete thicknesses: 8 cm 16 cm 24 cm 234

The following materials were used to manufacture the test specimens: • Concrete types: 1. Normal concrete B35/25: Material properties according to SIA 162/1993 2. SCC: Self-Compacting Concrete of SIKA AG • Timber: 3-ply wooden slab, quality C24 (SIA 265) • Adhesive: 2-component, epoxide-based, SIKA AG 3.2.2 Manufacture of the test specimens The above-mentioned parameter combinations were taken into account. Large specimens were manufactured which were later sliced into smaller elements for the different tests. Figure 1 shows the manufacturing process. The steel formwork was fixed to the timber. The adhesive – coloured red for a better visualization – was mixed and then applied onto the contact surface of the timber. After a suitable time interval (two parameters were used) the fresh concrete was poured. Figure 2 shows the hardened specimens before they were sliced up for the different tests as shown in Fig. 3. 3.2.3 Visual analysis of prismen sections Different sections from different parts of the specimen in the length were inspected for signs of some disturbance of the adhesive such as some possible mixing with the concrete or the possible displacement from the original contact position with the timber. The results of the visual analysis can be summarised as follows: • The adhesive was markedly displaced only at the points where the concrete was poured in. • The local mixing of the adhesive with the concrete seemed to be more homogeneous when the special concrete SCC was used. In the case of conventional concrete as a whole less adhesive seemed to have been dislocated, but in places the displacement was quite pronounced (Fig. 4). • The time interval between the mixing of the adhesive and the pouring of the concrete seemed to be a key factor: the mixing of the adhesive with the concrete was markedly reduced when the time interval was increased from 15 minutes to 90 apparently because the adhesive could partially harden before the concrete arrived.

Figure 1. Specimens to test manufacturing conditions. Specimen for bending test

Specimens for visual analysis and shear test

Figure 2. Hardened specimens.

Figure 3. Scheme for slicing up the hardened specimens.

235

B35/25

SCC

15 minutes

90 minutes

Figure 4. Locally pronounced mixing of the adhesive with conventional concrete (left), compared with the more homogeneous mixing in the case of SCC (right).

Figure 5. Shear failure and shearing along the adhesive line, local displacement of the adhesive.

3.2.4 Bending tests with small specimens Four-point-bending tests were carried out in accordance with the concrete standards. The increasing load and the corresponding deflection at midspan were measured. As parameter 8 specimens with concrete thicknesses of 8 cm and 16 cm were tested. The tests results varied greatly: it was not possible to come to some clear findings with regard to the load bearing behaviour. Figure 5 shows a specimen which failed in shear. The local displacement of the red adhesive at the point where the fresh concrete had been poured in is clearly visible. 3.2.5 Shear tests with small specimens The shear tests were performed in accordance with the standard EN392 – Glulam, shearing of the adhesive line. A comparison of the test results show that the shear resistance was higher and the variation was also much less when conventional concrete as opposed to SCC was used. The inspection also showed that SCC specimens suffered more from adhesion failure: when conventional concrete was used most of the shearing lines ran through the concrete. 3.3 Bending tests using large-scale specimens 3.3.1 Materials used Based on the parameter study with regard to the optimization of the manufacturing process, the following materials and parameters were chosen to make large-scale specimens for bending tests: • Concrete:

B35/25 (SIA 162) resp. C20/25 (SIA 265) 90 minutes interval between mixing of adhesive and pouring of concrete Careful pouring of concrete, gentle use of vibrators • Timber: C24 rectangular wood, glued side by side to form a slab • Adhesive: 2-component-adhesive on epoxide basis, produced by SIKA AG 236

Figure 6. Left: test specimens for three of the four series. Right: test set-up with specimen.

3.3.2 Bending tests Four-point-bending tests were carried out in accordance with the standard EN 408. Three specimens were tested for each of the four series with the different parameters with regard to the thicknesses of the timber and the concrete. The left half of Fig. 6 shows the three test specimens of three of the four series whilst the on right the test set-up is clearly visible. The test results met all expectations and can be summarized as follows: • No horizontal displacement of the concrete relation to the timber could be established by measurement, which confirmed the assumption that the adhesive connection was rigid. • An evaluation of the load–deflection lines confirmed that in all cases, the timber had a greater stiffness than the value according to the standards: ETimber = 10 000 N/mm2 . This finding further confirmed the rigidity of the adhesive connection. • In all test series the specimens failed exactly as predicted by calculation models, whether it was a case of shear failure, failure of the timber in tension or the concrete in compression. • In three of the four test series, the measured failure load of the different specimens was always greater than the calculated prediction and the variation was relatively small. In the fourth test series, only one of the three specimens attained the predicted failure load. An inspection suggested that the adhesive might have been widely displaced in these cases.

4 CONCLUSIONS In the 1st project phase, the manufacturing process could be optimised. The risk of a large scale displacement of the adhesive could be minimalized when the following parameters were used: • Adhesive: 2-component-adhesive on epoxide basis (SIKA Product) 925 g/m2 • Concrete: B35/25 (old standard SIA 162, now obsolete) resp. C20/25 (SIA 265) Pouring of concrete 90 minutes after mixing the adhesive Falling height of the concrete 50 cm In the 2nd project phase, large specimens were tested in bending to check the usefulness of the manufacturing parameters chosen in the 1st phase. The results were very satisfactory and can be summarised as follows: • A local displacement of the adhesive always occured at the points where the concrete was poured in, but this seemed to have no significant influence on the load-bearing capacity. • Shrinkage of the concrete led to horizontal cracks at the beam ends only in the specimens with a very thick concrete layer (240 mm). However, these cracks did not seem to have a negative influence on the load-bearing behaviour. 237

• In 10 of 12 test specimens, the measured failure load was higher than the value predicted by calculation. In the two specimens which failed prematurely, an inspection of the adhesive surface afterwards showed that some significant displacement had taken place. • The positive results confirmed that the parameters chosen for the manufacture were good. However, great care is needed when mixing the adhesive and pouring the concrete. The project is not yet complete. At the moment, the 3rd phase has started in which slab elements with a reduced adhesive surface will be tested. Should the results be favourable, these elements may be used in a pilot structure where the suitability of the system for site work could be tested and long-term effects could also be studied.

REFERENCES 1. Blass H.J., Schlager M.: Trag- und Verformungsverhalten von Holz-Beton-Verbundkonstruktionen, Bauen mit Holz, Hefte 5 & 6, 1996. 2. Brunner M., Gerber C.: Holz-Beton-Verbunddecken mit Klebeverbindung, 31. SAH-Fortbildungskurs Weinfelden, 1999. 3. Brunner M., Gerber C.: Holz-Beton-Verbundelemente: Entwicklung von Holz-Beton-Verbundelementen durch die Anwendung von Klebesystemen, Interner Bericht der SH-Holz, Biel, März 2000. 4. Gisel Peter: Holz-Beton-Verbund-Elemente: alternative Anwendungen mit Klebesystemen, Diplomarbeit der SH-Holz, 1997. 5. Kenel A.: Zur Berechnung von Holz-Beton-Verbundkonstruktionen, Forschungs- und Arbeitsbericht, EMPA-Debendorf, 1999. 6. Kuhlmann U., Gerold M., Schänzlin J.: Brettstapel-Beton-Verbund – Berücksichtigung von Kriechen und Schwinden, Bauingenieur 77, 6/2000, S. 281–288. 7. Kuhlmann U., Gerold M., Schänzlin J.: Trag- und Verformungsverhalten von Brettstapel-BetonVerbunddecken, Bauingenieur, Band 77, Januar 2002, S. 22–34. 8. Natterer J., Hoeft M.: Zum Tragverhalten von Holz-Beton-Verbundkonstruktionen, Forschungsbericht CERS Nr. 1345, IBOIS, EPFL, 1987.

238

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Composite of board stacks and concrete with integrated slim-floor-profiles J. Schänzlin & U. Kuhlmann Institute of Structural Design, University of Stuttgart, Germany

ABSTRACT: The composite of board stacks and concrete requires similar cross-section heights than pure reinforced concrete slabs. Unfortunately the load transfer of these composite slabs is only unidirectional; therefore a regular support by walls is necessary restricting the free choice of the floor plan. Also a point supported flat slab cannot be realized. By integrating a steel slim-floor profile a very flexible slab system is created that allows a free design of the floor plan including point supported slabs. Comparing concrete and composite slabs with integrated slim-floor-profiles, the composite slab system achieves larger spans, reduces the amount of required reinforcement and minimizes the dimensions of the load bearing structures. Therefore not only ecological reasons but also economical reasons argue for this composite system. KEYWORDS: Timber-concrete-composite structures, integrated continuous support, flat slab.

1 INTRODUCTION The composite slab of wooden board stacks and concrete provides many advantages compared to pure timber or pure concrete slabs, such as increased stiffness and higher ultimate load and improved sound and fire insulation resp. minimized dead load and increased amount of regrowing materials. The board stacks, normally installed in the tension zone of the composite slab, are produced by connecting single boards vertically at each other; the shear forces between the single planks are transferred by nails, screws or timber dowels (comp. Natterer 1997 and Widmann 2001). Figure 2 shows the required total height of a composite slab of board stacks and concrete and of a normal reinforced concrete slab according to Eurocode 2 1992. It is obvious, that the required height of both slabs are comparable, in spite of the small stiffness and the reduced maximum allowable stresses of timber compared to reinforcement steel. But this disadvantage of timber in comparison to reinforcement is compensated by the larger area of timber in the composite structure compared to the reinforcement in normal concrete slabs.

Md = MR,d

concrete slab hc,res

composite slab of board stacks and concrete hc = hres 3 hres (a) Small elements for testing purpose

(b) Load transfer

Figure 1. Comparison of a pure concrete and a composite slab.

239

hc,res 3 Fc

Fc Fc Ft

Fc Mt

required total height

60

BSBVD=Brettstapel-Beton-Verbunddecke

50 40 30 20 10 0 4

6

8

10

12

14

span in m composite slab, l/300 with maximum precamber concrete slab according Eurocode 2 1992 composite slab, l/200 with maximum precamber

Figure 2. Comparison of the required height between a pure concrete slab and a composite slab.

Figure 3. Up-to-now realised continuous supports.

Due to the connection between the single boards and the weakness of timber perpendicular to the grain, the load transfer of these elements is only unidirectional. Therefore a continuous support is necessary. In normal houses the possible span of these composite decks allows the support on the outer walls; but in multiple family dwellings and office buildings the required continuous support often limits the desired free choice of the floor plan. Beside this limitation in the floor plan, cut-outs in the slab for staircases, installations or similar, often require a continuous beam, that prevents a flat ceiling (comp. Fig. 3). Therefore a system is needed, which allows on one hand a continuous support for the composite slab of board stacks and concrete and on the other hand does not increase the overall slab height by a supporting beam. 2 INTEGRATED SLIM-FLOOR PROFILES One possibility to install a continuous support without any additional height for the supporting beams, is the integration of steel slim-floor-profiles such as Fig. 4(a) within the composite slab as an internal support. In this flat slab system two subsystems are combined (comp. Fig. 4(b)). In one direction the loads are transferred by the composite slab of board stacks and concrete to the 240

concrete

(a) Steel slim-floor profile

board stack

timber-concrete composite steel-concrete composite concrete board stack

slim floor profile

slim-floor profile

(b) Integrated slim-floor profile

(c) 3-dimensional view

Figure 4. Composite of board stacks and concrete with integrated slim-floor profile.

slim-floor profile (comp. Fig. 4(c)). For this direction, the concrete has its maximum moment in the middle of the composite slab, whereas at the support the concrete is under-utilized in the composite-slab-span direction. This concrete part can be activated as pressure zone of the steel–concrete-composite beam. Therefore a continuous support within the slab is achieved by only adding the steel-profile for the second subsystem. For this subsystem several steel cross-section types are possible, but to obtain a slender system without any visible beam, the profile shown in Fig. 4(a) appears most reasonable (comp. Kuhlmann et al. 2000 and Kuhlmann et al. 2001). It is produced by welding a common U-profile on a flat bar steel and adding headed studs on the top of the U-profile for creating the composite action between steel and concrete. The advantages of this profile compared to other possible cross-section types are among others the easy manufacturing, the bending and torsional capacity and the shallow height, that allows the installation of tension reinforcement in order to receive a continuous composite slab across the beam. Due to the welding of two cross-sections a pre-camber of the steel profiles can easily be manufactured in order to limit the final deflection of the steel composite beam. For the erection process the steel profiles are installed first. Due to the bending and torsional capacity of the steel profile, the board stacks can be laid on the lower flange of the profile without any auxiliary support. In order to limit the deflection in the middle of the board stacks some props may be added. Finally the concrete is cast over the whole structure and the composite of timber and concrete as well as the composite of steel and concrete are realized within one production process (Fig. 4(c)). As seen, the erection process does not differ from the erection process of a common concrete slab with prefabricated filigree elements as permanent formwork.

3 COMPARISON OF A PURE REINFORCED CONCRETE SLAB AND A COMPOSITE SLAB OF BOARD STACKS AND CONCRETE WITH INTEGRATED SLIM-FLOOR PROFILES In order to determine the potential of this new composite system, several studies were performed. Whereas the subsystem “steel–concrete-composite beam”, the subsystem “timber–concretecomposite-slab” and their connection devices are well known (comp. a. o. Fries 2001, Kürschner 2003, Schänzlin 2003, Kuhlmann and Michelfelder 2001), the load- and deformation behavior of the combination of both subsystems is not known. 241

span of the slim-floor profile ly in m (UPE270)

span of the slim-floor profile ly ly lx lx span of the composite slab

13 12 11 10 9 8 7 6 5 4

concrete slab (d=32cm) composite slab (d=32cm)

4

(a) System

5

6 7 8 span of the slab lx

9

10

(b) Maximum spans of the two systems

Figure 5. Possible systems for a UPE270-slim-floor profile under a live load of 3,5 kN/m2 and a thickness of the slab of d = 32 cm.

Due to the fact, that design methods for this timber–concrete-steel-composite system are missing, the design was performed according to Eurocode 4 1994. Nevertheless following assumptions have to be made: • the determination of the effective width: As visible in (comp. Fig. 4(b)) concrete is installed between the timber and the steel cross-section for reasons of tolerance. Therefore the thickness of the concrete slab is not constant, so influences on the effective width are expected. Nevertheless the effective width according to Eurocode 4 1994

is used, because the concrete is normally cracked in the lower 2/3 of its height in steel–concrete composite beams with integrated slim-floor-profiles. So the influences of this changing concrete thickness are expected to be low. • the transfer of the hogging bending moment of the slab in the range of the slim-floor-profile: In order to get results on the safe side, the timber–concrete-slab is designed as single span girders, whereas the steel–concrete-beam is designed for a continuous slab. So a maximum deflection for the timber–concrete-system is achieved whereas the steel–concrete-beam is loaded maximum. • the shear forces are transferred from the timber-concrete-composite slab to the steel–concrete subsystem only by contact in the flanges of the slim-floor-profiles With these assumptions a case study was performed (comp. Fronmüller 2003). One main conclusion is, that due to the reduction of the dead load of the composite slab of board stacks and concrete with integrated slim-floor-profiles, the allowable span of both directions can be increased compared to a pure reinforced concrete slab with integrated slim-floor-profiles (comp. Fig. 5). As visible in Fig. 5 for the same slim-floor profile UPE270 and the same overall height of 32 cm, the columnfree spanned area reaches round about 37 m2 (5 × 7 m) for the pure concrete slab, whereas for a composite slab of board stacks and concrete with integrated slim-floor profile a column-free spanned area comes up to 60 m2 (5 × 12 m). This means, by only replacing the cracked concrete zone of a pure reinforced concrete slab by board stack elements, the column-free area can be increased to 60%. Therefore less columns and a more flexible floor plan are feasible. In addition to this, the high load bearing capacity of the composite structure leads to a minimization of steel of the reinforcement and of the slim-floor-profile compared to concrete slabs with integrated slim-floor-profiles. For the same imposed load the amount of steel can be reduced of about 60%–70% (comp. Fronmüller 2003 and Fig. 6). 242

related amount of required steel

100%

concrete composite slab

80% 60% 40% 20% 0%

94

75 spanslim-floor profile  spanslab

49

Figure 6. Required amount of steel of the slim-floor-profile for various spans.

(a) Ceiling of a board stack

(b) View of the complete composite slab system (virtual photo)

Figure 7. Ceiling of the composite slab.

Besides the reduction of the required volume of the material the load transmitting systems, such as columns and foundations, can be minimized due to the reduced dead load of the composite systems. Therefore slender and lighter structures can be achieved. In flat slabs made of concrete with point support, the punching of the slab near the columns is often decisive for the dimensioning. In this new type of flat composite slab, the punching is not critical, because the shear forces are transferred by the steel profile. For an efficient joining between steel profile and column several design proposals are given in Lenzen et al. 1999. In addition advantages concerning the erection process can be summarized as follows: • reduction of the efforts for the framework, due to the use of the board stacks as permanent formwork and as replacement of the reinforcement in the cracked zone. • reduction of the required amount of props due to the higher bending strength of the board stacks compared to a prefabricated filigree concrete slab. • reduction of erection time because the amount of in-situ concrete can be minimized to about 1/3 of the amount of a concrete necessary for pre-fabricated filigree slab due to the increased height of the board stacks. • finished ceiling due to the visible board stacks (comp. Fig. 7). Therefore not only material costs and labor costs are reduced by using this type of structural composite system, but also aesthetical demands may be fulfilled. As shown, this system can be an alternative to common reinforced concrete slabs with integrated slim-floor-profiles. 243

CReinforcement z

Cshear CSteel

(a) System

CConcrete

CTimber

(b) Component model

Figure 8. Development of the component model.

4 CURRENT STUDIES The current studies aim at the determination of the load and deformation behavior of this new slab system. Therefore the unknown parameters, such as the influence of the change of the concrete thickness or the transfer of the shear forces from the timber–concrete slab into the steel–concrete beam are being investigated. An other idea to improve the building system for example is to allow for a transfer of the hogging bending moment of the timber–concrete-composite slab across the slim-floor-profile by activating the reinforcement, that is installed in order to prevent large cracks due to shrinkage. As for timber–concrete slabs with a span of more than 6 m the limitation of the deflection determines the required overall height of the in more than 80% of all cases, especially the load transfer of the hogging bending moment at the slim-floor-profile is a very promising step. By changing the structural system from single spans to continuous slabs, slender slabs are possible. The load transfer of the hogging bending moment at the slim-floor-profile is realized by a component model. In difference to steel–concrete-composite structures, where the tension is transferred by the reinforcement as well as by the steel profile due to the (semi-) rigid column-beam-joint, the hogging bending moment is transferred in timber-concrete-composite structures only by a couple of forces – tension in the reinforcement of the concrete slab and pressure in the timber cross section, transferred by contact at the joint timber–concrete and concrete–steel. A first approach for the component model is shown in Fig. 8. The components of this joint are • • • •

reinforcement under tension timber under pressure concrete under pressure steel under pressure.

It is assumed, that the components “concrete under pressure” and “steel under pressure” can be neglected due to the high stiffness and the short dimensions. Therefore the stiffness of these components is assumed to be equal infinity. The component “reinforcement under tension” is determined according to ECCS 109 1999. Unfortunately the component “timber under pressure” is not known. Therefore as a first step of modelling this component was assumed according to following equation.

244

where

b width E|| Young’s Modulus parallel to the grain h height of the timber cross section s0 effective height of the contact at the joint timber–concrete m increase of the load distribution = tan α, according to Lippert 2002 = tan 7◦ α angle of load distribution This equation was obtained by modelling the load transfer of an eccentrically loaded timber cross section under the assumption of an increase of the load distribution with an angle α and the neglect of the deformation due to bending caused by the eccentricity of the load. With this model, the load deformation behavior of this type of joint can be determined. Beside this it is possible to study the influences of the long-term behavior of timber on the deformation behavior. Therefore it is possible to evaluate the effects of the flexibility of the joint on the complete slab not only for t = 0 but also for t = ∞. To determine a design method for this composite system the research work is continued. Within the scope of this research the transfer of the hogging bending moment is studied numerically and experimentally in addition to the component method. In this context the loading of the connection devices due to the different deformations of the continuous concrete slab and the single span timber slab are determined. Beside this, the shear transfer from the timber-concrete-composite slab to the steel-concrete-girder is studied, in order to determine, whether the flange of the slim-floor-profile has to carry the whole vertical load – as assumed – or not. Another question, which is going to be answered within the scope of this research project is the determination of the effective width of this composite structure. Thereby the influences of the change of the thickness of the concrete in the range of the slim-floor-profile are studied. Finally experimental studies will verify the theoretical investigations and the design method. 5 CONCLUSIONS

As shown, composite structures of board stacks and concrete form an alternative to common pure reinforced concrete slabs. The only unidirectional load transfer as an disadvantage of this type of composite structure can be eliminated, if steel slim-floor-profiles are integrated as continuous composite supporting beams. Compared to pure reinforced concrete slabs or even concrete slabs with integrated slim-floorprofiles the column free spanned area can be increased by about 60%, if the composite slab system of board stacks and concrete with integrated slim-floor-profiles is used. In addition it is expected that the erection time and the costs for the finish can be reduced. Therefore common reinforced concrete slabs can be replaced by a slender and lighter construction due to the use of timber as tension zone without higher costs being expected. The research presented here is part of an ongoing research project performed at the Institute for Structural Design, University of Stuttgart. The Deutsche Bundesstiftung Umwelt (DBU) is gratefully acknowledged for supporting the project.

REFERENCES ECCS 109 1999, Design of composite joints for buildings. Technical report, ECCS – Technical Committee 11 – Composite Structures. Eurocode 2 1992, DIN V ENV 1992 Eurocode 2: Planung von Stahlbeton- und Spannbetontragwerken; Teil 1: Grundlagen und Anwendungsregeln für den Hochbau. Eurocode 4 1994, pr DIN V ENV 1994: Bemessung und Konstruktion von Verbundtragwerken aus Stahl und Beton Teil 1-1: Allgemeine Bemesusngsregeln, Bemessungsregeln für den Hochbau. Fries, J. 2001, Beitrag zum Tragverhalten von Flachdecken mit Hutprofilen. Dissertation, Institut für Konstruktion und Entwurf, Universität Stuttgart (Mitteilung 2001-1).

245

Fronmüller, B. 2003, Integration der Brettstapel-Beton-Verbundbauweise in Slim-Floor-Systeme. Diplomarbeit, Institut für Konstruktion und Entwurf. Kuhlmann, U., J. Fries, and M. Leukart 2000, Bemessung von Flachdecken mit Hutprofilen. In U. Kuhlmann (Ed.), Stahlbau-Kalender 2000. Ernst u. Sohn. Kuhlmann, U., J. Fries, and A. Rieg 2001, Composite girders of reduced height. In R. Eligehausen (Ed.), Symposium on Connections between Steel and Concrete (Pro 21 ed.), Volume 2, S.A.R.L, pp. 1371–1381. RILEM Publications. Kuhlmann, U. and B. Michelfelder 2001, Kerven mit Schlüsselschrauben als Verbindung bei Brettstapel-BetonVerbunddecken. 1. Zwischenbericht zum Forschungsvorhaben AiF 13 204, Institut für Konstruktion und Entwurf, Universität Stuttgart. Kürschner, K. 2003, Trag- und Ermüdungsverhalten liegender Kopfbolzendübel im Verbundbau. Dissertation, Institut für Konstruktion und Entwurf, Universität Stuttgart. Lenzen, K., U. Kuhlmann, and J. Fries 1999, Slim-Floor Deckenträger mit UPE-Profilen. FRILO-Magazin, 53–56. Lippert, P. 2002, Biegesteife Rahmenecken mit eingeklebten Gewindestangen. Dissertation, Institut für Konstruktion und Entwurf. Natterer, J. 1997, Concepts and details of mixed timber-concrete structures. In Composite construction – conventional and innovative, Conference Report, Innsbruck, pp. 175–180. Schänzlin, J. 2003, Zum Langzeitverhalten von Brettstapel-Beton-Verbunddecken. Dissertation, Institut für Konstruktion und Entwurf, Universität Stuttgart. Widmann, R. 2001, Screw-laminated timber deck plates. In IABSE (Ed.), Innovative Wooden Structures and Bridges (85 ed.), Lahti, Finnland, pp. 223–228. IABSE: IABSE Bd. 85.

246

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Composite of board stacks and concrete J. Schänzlin & U. Kuhlmann Institute of Structural Design, University of Stuttgart, Germany

ABSTRACT: Timber–concrete-composite structures are strongly influenced by the timedependent behavior of the components timber and concrete. Therefore creep and shrinkage have to be considered. Due to the interaction of both components, stress redistributions take place. This leads to effective creep values, which differ from the material creep coefficients. In order to take these effects into account in the structural design, the determination of effective creep values and the consideration of shrinkage in the design method according Eurocode 5 1994 are shown in the following. KEYWORDS: Timber-concrete-composite structures, time dependent behavior, creep, shrinkage, design. 1 INTRODUCTION Timber–concrete-composite structures provide many advantages compared to pure timber or pure concrete slabs. In these structures, timber elements are connected with a concrete slab by different connection devices (comp. among other Bahmer and Bathon 2003, DIBt 1998, DIBt 2002, Kuhlmann and Michelfelder 2001, Blass et al. 1996 and Fragiacomo 2001). The concrete used is often of grade C20/25 and for the timber different elements are possible, e.g. beams, board stacks, etc. These board stacks are produced by connecting vertically single boards at each other by means of nails, screws or even timber dowels. The aim of this type of timber element is to produce a timber slab without gluing and by using less valuable timber (comp. among others Natterer 1997, Widmann 2001). Compared to a normal reinforced concrete slab, the cracked concrete zone of the concrete cross section is replaced by the board stacks (comp. Fig. 2). So the dead load of these this composite structures is reduced and the amount of regrowing materials in building structures is increased. In addition, by using board stacks as timber elements, the ceiling of the slab is finished, so additional work for the finishing can be saved. Compared to a pure timber slab, the concrete in the pressure zone of the composite structure increases the stiffness and the load capacity and improves the sound and fire insulation of a pure timber slab.

Figure 1. Board stacks.

247

Md = MR,d

concrete slab hc,res

hc,res 3

Fc Fc Fc

composite slab of board stacks and concrete h hc = res 3 hres

Fc Ft

Mt

Figure 2. Comparison of a pure concrete and a composite slab.

Initial state

shrinkage of concrete NS NS

Shortening of concrete

NS

Reaction of the composite section

NS

Resulting state

Figure 3. Influence of shrinkage of concrete.

For the structural design, the connection between timber and concrete is of importance, so several systems have been studied. Therefore the load-slip behavior is well known. These studies aimed at determining the stiffness and the ultimate load of the connection in order to allow an evaluation of the short term behavior according Eurocode 5 1994. The long term behavior however cannot sufficiently be determined because several influences, such as shrinkage, temporal development of the creep strain, composite action, etc. are neglected. Therefore design rules for the long term behavior are needed. 2 LONG TERM BEHAVIOR OF TIMBER–CONCRETE-COMPOSITE STRUCTURES 2.1 General Regarding the time-dependent behavior of timber–concrete-composite slabs the limitation of the deflection of the slab is often decisive for design. Due to the long term behavior the deflection increases because of • the shrinkage of timber and concrete: During shrinkage of concrete the concrete slab shortens resulting in a reduction of the internal concrete stresses, which leads to an increase of the stresses of the timber and to an increase of the curvature and therefore to an increase of the deflection (comp. Fig. 3). The shrinkage of timber influences the composite structure in opposite way, so shrinkage of timber reduces the deflection. If the stresses of external load and due to shrinkage of timber are combined, the internal bending moment of the cross sections decreases, whereas the internal normal forces increase. Therefore the edge stresses decrease whereas the stresses of the centroid increase. • creeping of timber and concrete: In a statically indeterminate system, creeping of one component leads to the decrease of the internal forces of this creeping cross-sectional part. So the stresses are redistributed, which means that the faster and stronger creeping component shows reduced stresses. Due to the global equilibrium the internal forces in the more slowly and fewer creeping cross-section increase. 2.2 Determination of the long term behavior In order to determine the long term behavior of timber–concrete-composite structures a model has been developed, by combining the composite theory of Dabaon et al. 1993 with the rheological models of timber according to Hanhijärvi 1995, of concrete according to Eurocode 2 1992 and of the connection according to Kenel and Meierhofer 1998. With this model the time-dependent 248

0 2 4 6 8 10 12 14 16 18

400

Time in days 800 1200

1600

calculation measurement (Kenel and Meierhofer 1998)

0

200

Time in days 400

600

800

0 Deflection in mm

Deflection in mm

0

10

¨ measurement (Hohmann and Siemers 1998) calculation

20 30 40 50 60

(a) Beam B4

(b) Beam B3

Figure 4. Comparison between calculation and measurement. Increase of σconcrete εc,cr. εc,el.

Creeping of concrete εc,cr. εc,el. εc,res.

εc,res.

Reduction of σconcrete εc,cr. εc,el. εc,res.

Increase of εc,res. resulting bending moment = constant Reduction of σtimber εt,cr. εt,el.

Creeping of timber εt,cr. εt,el.

εt,res.

εt,res.

Increase of σtimber εt,cr. εt,el. εt,res. Increase of εt,res.

Figure 5. Influence of the composite action.

behavior of various systems can be evaluated, taking into account the variation of the surrounding conditions, the cracking of concrete and the load-slip behavior of the connection. This model has been verified by the comparison between calculation and tests according to Kenel and Meierhofer 1998, Höhmann and Siemers 1998 (comp. Fig. 4) and Kuhlmann and Schänzlin 2002 as well as the comparison between calculation and measurements of the deflection of several slabs in a multiple family dwelling in Tübingen (comp. Schänzlin 2003). Beside these verifications the model was compared with the model according to Fragiacomo 2000 (comp. Fragiacomo and Schänzlin 2000), which bases on different rheological models and a different composite theory. So with this verified model the time-dependent behavior of timber–concrete-composite structures can be determined. However this model is too complex for the everyday structural design. For an easy consideration of the time dependent behavior of both components in the structural design creep and shrinkage may be considered by effective creep values and effective shrinkage values. These creep values ϕ are defined as the ratio of the strain caused by creep εcr. and the elastic strain εel. .

Due to the redistribution of the stresses and the effective creep depends on the redistribution of the stresses during the lifetime of the composite action (comp. Fig. 5). This redistribution depends on the end value of the creep coefficient as well as the temporal development. As in Fig. 6 visible, the concrete develops round about 90% of its end creep strain within the first 3 to 7 years, whereas timber creep values reach only around 50%. After this period the concrete hardly creeps, whereas the timber develops its remaining 40% of the end creep value. 249

80% Creep of concrete Shrinkage of concrete Creep of timber

60% 40% 20%

3 years 7 years

ratio ϕ(t)/ϕ(t=50a) resp. ε(t)/ε(t=50a)

100%

0 0

10

20 30 Time in years

40

50

Figure 6. Temporal development of the material creep coefficients ϕi and of the shrinkage εS related to the end values.

So within the first 3 to 7 years the concrete reduces its internal forces due to the faster and stronger creeping. After this time the relation reverses and the timber reduces its stresses by creeping. Therefore it is not sufficient to take only the points in time of t = 0 and t = ∞ into account but also the point in time of t = 3 to 7 years; that point in time, where the difference between the creeping of concrete and of timber reaches a maximum value. At that point the edge stresses in the timber cross section can reach a maximum value especially if shrinkage of timber is reduced by special concrete, by special treatment or if the slab is not within a heated building. 2.3 Determination of effective creep values As seen the elastic strain of the concrete is reduced due to its faster and stronger creeping within the first 3 to 7 years, so the effective creep coefficient according Eq. (1) increases. In general the average composite creep value of concrete is round about 50% larger than the material creep value due to this composite action. For the structural design these influences of the composite action has to be taken into account. First studies were performed by Kupfer 1958 with the aim to determine effective creep coefficient in pure concrete structures, provided that the development of the creep values of both components is affine. Kreuzinger 1994 upgraded this theory for the consideration of the deformability of the connection between both composite partners. Schänzlin 2003 added the different temporal development by dividing the development of the creep values in intervals within which the development of both creep values can be regarded as affine. Therefore composite creep values can be determined by Eq. (2).

where ϕo,V,i ϕo,M,i ψi o i

increase of the effective creep coefficient of the component o within the interval i (comp. Table 1) increase of the material creep coefficient of the component o within the interval i (comp. Table 1) system creep coefficient within the interval i (comp. Kupfer 1958 and Schänzlin 2003) timber cross section, concrete cross section interval according to Fig. 7

and

where ϕo,V effective creep coefficient of the component i. 250

Table 1. Intervals for a simplified determination of the effective creep coefficients. Points in time t = 3−7a

t =∞

Interval

ϕT

ϕC

ϕT

ϕC

1 2 3

0,40 · ϕT ,M,∞ 0,10 · ϕT ,M,∞ 0,0 · ϕT ,M,∞

0,85 · ϕC,M,∞ 0,05 · ϕC,M,∞ 0,0 · ϕC,M,∞

0,40 · ϕT ,M,∞ 0,20 · ϕT ,M,∞ 0,40 · ϕT ,M,∞

0,85 · ϕC,M,∞ 0,15 · ϕC,M,∞ 0,0 · ϕC,M,∞

a = years.

Development of ϕconcrete

1.0 0.8 0.6 0.4 course affine course multi-linear function

0.2 0 0

0.2

0.4

0.6

0.8

1.0

Development of ϕtimber

Figure 7. Development of the creep coefficient of concrete and approach by a multi-linear function in dependence of the development of the creep coefficient of timber.

So the effective creep coefficient, taking into account the composite action and the different temporal development of both materials can be determined according to Eq. (2). Due to the fact that it takes some effort to determine the effective creep coefficient according to Eq. (2), a statistical evaluation of these creep coefficients has been performed. For this study around 4000 cases with different geometries, connection properties and material properties have been evaluated and the 95%-fractile values of the increase of the material creep coefficient have been determined resulting in the values given in Table 2. With these values, the effective creep coefficient in timber–concretecomposite structures can be determined to

where

ϕi,M material creep coefficient of the material i according the relevant standard ψi,V factor considering the composite action ϕi,V effective creep coefficient in a composite structure.

2.4 Effective shrinkage values As Kupfer 1958, Kreuzinger 1994 and Schänzlin 2003 show, the creep coefficient of a member of the composite structure depends on the type of loading; it differs whether the stresses are caused by an external constant load, by shrinkage or by a sudden constraint. In order to avoid creep coefficients in dependence on the type of the loading, an effective shrinkage of concrete has been 251

Table 2. Composite factor for the creep coefficient of timber ψt,comp. , of concrete ψc,comp. and the 95%-fractile factor kc,s for the effective shrinkage. Point in time

ψT,V

ψC,V

kc,s

t =0 t = 3−7a t =∞

0 0,5 1,0

0 1,9 2,0

0 0,5 0,8

a = years.

determined, so that the results due to shrinkage under consideration of the creep coefficients due to shrinkage are equal to the results due to effective shrinkage with the creep coefficients due to external load. Therefore a superposition of different load cases becomes possible. In order to avoid complex equations for practioners, out of a high number of calculated cases the 95%-fractile value of this effective shrinkage value has been defined. With this value, the effective shrinkage strain can be determined to

where εres ks

strains of shrinkage of concrete 95%-fractile value.

3 STRUCTURAL DESIGN The structural design of composite beams with deformable connectors is given in Eurocode 5 1994. In this design method an effective moment of inertia is determined. It is derived by a comparison of the mid-span deflection of a beam with rigid connectors and a beam with deformable connectors, loaded by a sinusoidal distributed load. With this design method only the short term behaviour can be evaluated. For the long term behaviour several modifications are necessary: • modification of the moment of inertia: Due to the superposition of shrinkage and external load, the effective moment of inertia is influenced by the unstressed strain. This influence can be considered by the stiffness factor CJ,sls :

where

and εt,d,∞ εc,d,∞ qd ks,res

strain of timber at time t = ∞ due shrinkage or temperature change strain of concrete at time t = ∞ due shrinkage or temperature change design value of the external load (without the fictive load caused by shrinkage according to Eq. (10)) factor according to Table 2.

With the stiffness factor CJ,sls the moment of inertia can be determined to

252

Table 3. Influence of shrinkage on the internal forces and deflections. Deflection Bending moment1 Normal force1 Maximum loading of the connectors Shearing forces 1 In

Increase Increase Decrease Decrease Decrease

the components of the composite cross section.

required total height in cm

60 50 40 30 20 10 0 4

12 10 span in m composite slab, l/300 with maximum precamber concrete slab according to Eurocode 2 1992 composite slab, l/200 with maximum precamber 6

8

14

Figure 8. Comparison of the required height between a pure concrete slab and a composite slab.

• Consideration of shrinkage as a fictive load: Shrinkage of concrete influences the internal forces and the deformation according Table 3. In order to consider shrinkage, the increase of the deflection and of the bending moment are described by a fictive load. It can be determined by following equation.

where Cp,sls according to Eq. (7) εsls,d according to Eq. (8) • Consideration of creep of both materials: As shown, the temporal development of the creep strain and the composite action influence the effective creep value of both materials. In order to consider these influences the effective Young’s modulus has to be determined by following equation:

where Ei,t,eff ψi,comp. ϕi,∞

effective Young’s modulus of the component i at point in time t composite factor according Table 2 material creep coefficient at t = ∞.

4 COMPARISON OF PURE CONCRETE SLABS WITH COMPOSITE SLABS Figure 8 shows the required total height of a normal reinforced concrete slab according to Eurocode 2 1992 and of a composite slab of board stacks and concrete. It is obvious, that the required height 253

of both slabs are comparable, in spite of the small stiffness and der reduced maximum allowable stresses of timber compared to reinforcement steel. But this disadvantage of timber in comparison to reinforcement is compensated by the larger area of timber in the composite structure compared to the reinforcement in normal concrete slabs. Unfortunately the load-carrying capacity of the composite can not fully be exploited, because the time-dependent behaviour of timber as well as of concrete strongly influences the deflection of the slab (comp. Schänzlin 2003). Therefore the limitation of the deformation is often the decisive criteria, which determines the required height of the composite slab.

5 CONCLUSIONS As shown creep and shrinkage of both materials strongly influence the long term behaviour of timber concrete composite structures. The most decisive design criteria is the limitation of the deformation. In some cases, especially if shrinkage of concrete is reduced by special concrete or treatment, the point in time of t = 3 to 7 years, when concrete has finished around 90% of its end creep value and the creep value of timber has reached only 60%, the edge stresses of the timber section can reach a maximum value, which limits the height of the slab. This non affine development of the creep coefficients and the composite action lead to composite creep values which differ from the material’s creep coefficient. For the practical use 95%-fractile values of the increase resp. the decrease of the material’s creep coefficient are given for the three critical points in time t = 0, t = 3 to 7 years and t = ∞. Beside this additional point in time t = 3 to 7 years, shrinkage has to be considered in the structural design. This can be done by a fictive load in the design method according to Eurocode 5 1994. By this fictive load, the increase/decrease of the material’s creep coefficient and the modification of the moment of inertia, the time-dependent behaviour of timber–concrete-composite structures can be evaluated according to Eurocode 5 1994. The research presented here is part of a research project performed at the Institute for Structural Design, University of Stuttgart. The Deutsche Gesellschaft für Holzforschung, e.V. (DGfH), the Arbeitsgemeinschaft industrieller Forschungsvereinigungen “Otto von Guerike” e. V. (AiF) and the Bundesamt für Bauwesen und Raumordnung (BBR) are gratefully acknowledged for financing and supporting the project.

REFERENCES Bahmer, R. and H. Bathon 2003. Mut zu neuem – 10 m frei spannende holz-beton-verbund-flachdecke. bauen mit holz (1), 21–25. Blass, H.-J., J. Ehlbeck, M. L. R. Linden, and M. Schlager 1996. Trag- und Verformungsverhalten von Holz-Beton-Verbundkonstruktionen. bauen mit holz, 392–399. Dabaon, M., F. Tschemmernegg, K. Hassen, and T. A. Lateef 1993. Zur Tragfähigkeit von Verbundträgern bei teilweiser Verdübelung. Stahlbau 62, 3–9. DIBt 1998. Allgemeine bauaufsichtliche Zulassung Z-9.1-342, SFS-Verbundschrauben. Deutsches Institut für Bautechnik. DIBt 2002. Allgemeine bauaufsichtliche Zulassung Z-9.1-473: Brettstapel-Beton- Verbunddecken mit Flachstahlschlössern. Deutsches Institut für Bautechnik. Eurocode 2 1992. DIN V ENV 1992 Eurocode 2: Planung von Stahlbeton- und Spannbetontragwerken; Teil 1: Grundlagen und Anwendungsregeln für den Hochbau. Eurocode 5 1994. DIN V ENV 1995 Eurocode 5: Entwurf, Berechnung und Bemessung von Holzbauwerken; Teil 1-1: Allgemeine Bemessungsregeln, Bemessungsregeln für den Hochbau. Fragiacomo, M. 2000. Comportamento a lungo termine di travi composte legno-calcestruzzo. Dissertation, Universität Trieste. Fragiacomo, M. 2001. Long term behaviour of a timber-concrete connection system. pp. 263–269. Fragiacomo, M. and J. Schänzlin 2000. Modelling of timber-concrete floor structures. In A. Cecotti and S. Thelandersson (Eds.), Timber constructions in the new millenium. Cost E5.

254

Hanhijärvi, A. 1995. Modelling of creep deformation mechanisms in wood. Dissertation, Helsinki University of Technology. Technical Research Centre of Finland. VTT Publications. Espoo (SF). Höhmann, R. and M. Siemers 1998. Untersuchungen zum Trag- und Verformungsverhalten von Holz-BetonVerbundträgern. Bautechnik 75(11), 922–929. Kenel, A. and U. Meierhofer 1998. Holz/Beton-Verbund unter langfristiger Beanspruchung. Forschungs- und Arbeitsbericht 115/39. EMPA Dübendorf (CH), Abteilung Holz. Kreuzinger, H. 1994. Verbundkonstruktionen Holz/Beton. Kuhlmann, U. and B. Michelfelder 2001. Kerven mit Schlüsselschrauben als Verbindung bei Brettstapel-BetonVerbunddecken. 1. Zwischenbericht zum Forschungsvorhaben AiF 13 204, Institut für Konstruktion und Entwurf, Universität Stuttgart. Kuhlmann, U. and J. Schänzlin 2002. Baukostensenkung durch weiterentwickelte Brettstapel-BetonVerbunddecken. Abschlußbericht zum Forschungsvorhaben; Bundesamt für Bauwesen und Raumordnung BBR (Z6-5.4.00-14/II 13-80 01 00-14), Institut für Konstruktion und Entwurf. Kupfer, H. 1958. Kräfteumlagerung durch Kriechen bei unterschiedlichen Kriechzahl der verwendeten Baustoffe. Monographie der DYWIDAG. Natterer, J. 1997. Concepts and details of mixed timber-concrete structures. In Composite construction – conventional and innovative, Conference Report, Innsbruck, pp. 175–180. Schänzlin, J. 2003. Zum Langzeitverhalten von Brettstapel-Beton-Verbunddecken. Dissertation, Institut für Konstruktion und Entwurf, Universität Stuttgart. Widmann, R. 2001. Screw-laminated timber deck plates. In IABSE (Ed.), Innovative Wooden Structures and Bridges (85 ed.)., Lahti, Finnland, pp. 223–228. IABSE: IABSE Bd. 85.

255

Session 5: Robustness

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Stresses in steel columns under natural fire Z. Sokol & F. Wald Czech Technical University in Prague, Czech Republic

ABSTRACT: The paper presets a part of the results reached during the Structural integrity fire test in Cardington laboratory January 16, 2003. The work summarises the instrumentation and focuses to the measured stresses and temperatures in columns, which was instrumented to observe the tie forces under natural fire. Comparison to finite element simulation, taking into account the connection behaviour is included. The test documents, that finite element models allows presuming the steel to concrete composite frame behaviour with a good accuracy. The composite action and joint behaviour play important role in the structural behaviour under fire. The simple rules for the design of the protected columns ware confirmed in all thee parts of prediction; in fire modelling by parametric fire curve, in transfer of heat to the structure, and in the structural analyses. KEYWORDS: behaviour.

Steel structures, Fire design, Fire test, Global analysis, Natural fire, Column

1 INTRODUCTION Current fire design models for the time-temperature development within structural elements and for structural behaviour are based on tests of isolated members subjected to standard fire regimes. These standard regimes are used as a reference of heating, but do not model the natural fire. Only the tests on a real structure under the natural fire may evaluate the forthcoming models of the temperature development in fire compartment, of the transfer of heat into the structure and of the overall structural behaviour under fire. The BRE’s Cardington Laboratory is a unique worldwide facility for the advancement of the understanding of whole-building performance, see [1]. This facility is located at Cardington, Bedfordshire, UK and consists of a former airship hangar with dimensions 48 m × 65 m × 250 m. The Cardington Laboratory comprises three experimental buildings: a six storey timber structure, a seven storey concrete structure, and an eight storey steel structure. The steel test structure was built in 1993. It is a steel framed construction using composite concrete slabs supported by steel decking in composite action with the steel beams. It has eight storeys (33 m) and is five bays wide (5 × 9 m = 45 m) by three bays deep (6 + 9 + 6 = 21 m) in plan. The structure was built as non-sway with a central lift shaft and two end staircases providing the necessary resistance against lateral wind loads. The main steel frame was designed for gravity loads, the connections consisting of flexible end plates for beam-to-column connections and fin plates for beam-to-beam connections, designed to transmit vertical shear loads. The building simulates a real commercial office in the Bedford area and all the elements were verified according to British Standards and checked for compliance with the provisions of the Structural Eurocodes. The building was designed for a dead load of 3,65 kN/m2 and an imposed load of 3,5 kN/m2 . The floor construction consists of steel deck and light-weight in-situ concrete composite floor, incorporating an anti-crack mesh of 142 mm2 /m in both directions, see [2]. The floor slab has an overall depth of 130 mm and the steel decking has a trough depth of 60 mm. Seven large-scale fire tests at various positions within the experimental building were conducted, see [3], and there is still a place for two more tests. 259

Figure 1. (a) Mechanical load of compartment; (b) fire load around column D2; (c) protection of internal column D2 (after test); (d) protection of external column (after test).

The main aim of this compartment fire tests was to assess the behaviour of structural elements with real restraint under a natural fire. The structural integrity fire test (large test No. 7) was carried out in a centrally located compartment of the building, enclosing a plan area of 11 m by 7 m on the 4th floor [3]. The preparatory works took four months. The fire compartment was bounded with walls made of three layers of plasterboard (15 mm + 12,5 mm + 15 mm) with a thermal conductivity 0,19–0,24 W m−1 K−1 . In the external wall the plasterboard is fixed to a 0,9 m high brick wall. The opening of 1,27 m high and 8,7 m length simulated an open window to ventilate the compartment and allow for observation of the element behaviour. The ventilation condition was chosen to result in a fire of the required severity in terms of maximum temperature and overall duration. The steel structure exposed to fire consists of two secondary beams (section 305 × 165 × 40UB, steel S275 measured fy = 303 MPa; fu = 469 MPa), edge beam (section 356 × 171 × 51UB), primary beams (section 336 × 171 × 51UB, steel S350 measured fy = 396 MPa; fu = 544 MPa) and columns, internal section 305 × 305 × 198UC and external 305 × 305 × 137UC, steel S350. The joints are a cruciform arrangement of a single column with three or four beams connected to the column flange and web by the header plate connections, steel S275. The beam to beam connections were created by fin plates, steel S275. The composite behaviour was achieved by a concrete slab (light weight concrete LW 35/38; experimentally by Schmidt hammer 39,4 MPa) over the beams cast on shear studs (Ø19–95; fu = 350 MPa). The mechanical load was simulated using sandbags, 1100 kg of each, applied over an area of 18 m by 10,5 m on the 5th floor. Sand bags represent the mechanical loadings; 100% of permanent actions, 100% of variable permanent actions and 56% of live actions. The mechanical load was designed to reach the collapse of the floor, based on analytical and FE simulations. Wooden cribs with moisture content 14 % provided the fire load of 40 kg/m2 of the floor area, see Figure 1a, b. The columns, external joints and connected beam (about 1,0 m from the joints) were fire protected to prevent global structural instability. The material protection used was 20 mm of Cafco300 260

E1

D1

PLAN

5th floor

Internal wall of the fire compartment 11,0 m 7,0 m

N

99

97

103

101

83

87 D1 Window 1,27 x 8,70 m E1 UC 305 x 305 x 198 UC 305 x 305 x 137

500

99, 103 97, 101 81 85 83,87

81, 85 500

UC 305 x 305 x 137 (UC 305 x 305 x 198) 20

20

115

113

119

117 115, 119

15,2

(31,4) 21,7 20

309,2 (314,5) 320,5 y (339,9) 13,8 (19,1) z

127 123

107

105

111

109 107, 111 105, 109

91

89

95

93 91, 95 89, 93

4th floor

500 113,117

125 121 127, 123

121, 125 500

20

3rd floor

Figure 2. Location of ambient temperature strain gauges on columns.

vermiculite-cement spray, based on vermiculite and gypsum, see Figure 1c, d. It was applied as a single package factory controlled premix, with a thermal conductivity of 0,078 W m−1 K−1 . 2 INSTRUMENTATION The instrumentation used included thermocouples, strain gauges and displacement transducers. A total of 133 thermocouples monitored the temperature of the connections and beams within the compartment, the temperature distribution through the slab and the atmosphere temperature within the compartment. An additional 14 thermocouples measured the temperature of the protected columns. Two different strain gauge types were used, high temperature and ambient temperature. Nine high temperature strain gauges were used in the exposed and un-protected elements (fin plate and end plate – minor axis joint). A total of 47 ambient strain gauges were installed in the protected columns, see Figure 2, and on the concrete slab. Twenty five vertical displacement transducers were attached along the 5th floor to measure the deformations of the concrete slab. Twelve transducers were used to measure the horizontal movement of the columns and the slab. Ten video cameras and two thermo imaging cameras recorded the fire and smoke development, the deformations and temperature distribution, see [3]. 3 COLUMN TEMPERATURES The temperatures in the columns in the fire compartment were measured at middle of their height, 500 mm from the floor, and 500 mm below the ceiling at both flanges and at the web, and in 261

C402, C405, C407

E2

D2

C404

C488

C410 C401, C403, C406 N E1

D1 Temperature, °C 1000

1348

Gas temperature

C410 Secondary beam D2-E2; midspan, lower flange, measured, C488

G525 800

400

C404

C403, C405

1348 C406, C407

External column D1: 500 mm under slab, C402 at mid height, C405 500 mm above floor, C406

600

C401, C402

500

500 D2

D1

3rd floor

Internal column D2, at mid height measured C410

200

0

0

15

30

45

60

75

90

105

120

Time, min.

Figure 3. Comparison of the measured temperatures along the external column (D1) length to the internal column temperature (D2), gas and beam temperature.

the connections. The columns were fire protected except the joint area, where the primary and secondary beams were connected. A selection of the temperatures recorded at column D1 and D2 are presented in Figure 3, where they are compared to the gas temperature, the beam midspan temperature, the beam end temperature and the column end temperature. As the fire created homogenous gas temperature both columns were heated almost equally. The maximum reported temperature in the insulated part of the middle column of 426,0◦ C, which occurred after 106 minutes of fire. The values reached at the middle of column height and in the upper part of the column are similar. The gradient of the temperatures along the column is changing during the fire. The differences of the measured temperatures across the section were insignificant, see [4]. 4 COLUMN STRESSES The external columns were equipped by the strain gauges 20 mm from the section edge, see Figure 2. Low temperature of the fire protected columns allowed recording the strains using ambient temperature strain gauges at 3rd floor till the 60 minutes of the fire and at 4th floor during the whole experiment. The strains were transformed to stresses using modulus of elasticity 210 000 MPa. Selection of results is presented at Figures 4 and 5 as well as in Tables 1 and 2. 5 STRUCTURAL ANALYSIS The discrete structural analysis is used to predict/verify the findings on Cardington frame from the first to latest tests, see [5] and a comprehensive review of models in [6]. For the numerical simulation 262

Figure 4. Measured stresses at external columns, section 500 mm above the floor at 4th floor. Stress, MPa 150

107 100

97

Column D1 99 97

Column E1 105 107

103

111

101

109

99

50

105 Time, min.

0 0 -50 -100

15

30

45

60

111 101 103 109

75

90

105

150 120

165

180

195

210

135

4th floor

D1, E1

-150 -200

Figure 5. Measured stresses at external columns, section 500 mm below the ceiling at 4th floor.

of sevenths test was utilized finite element code ANSYS 5.7, see [7]. The non-linear analysis of the structure included plasticity of the beam elements, large strains and large deformations and non-linear response of structural joints and shear connectors. Multi-linear isotropic material with strain hardening was used for mild steel. Temperature dependent stress strain relationship (σ − ε) of the steel is based on the Eurocode model EN 1993-1-2: 2004. The structural elements were divided into nine finite elements. Beam element with three degrees of freedom at each node: translation in nodal x and y directions and rotation about nodal z-axis, marked as BEAM 23 in the ANSYS element library, was used for the beams and columns. This element allows linear temperature distribution along its height and length. The structural joints were 263

Table 1. Measured stresses on columns during natural fire, stain gauges 500 mm above floor. Measured stresses, MPa Column D1 Time, min.

C81

15 30 45 60 90 124 160 226

−23,1 −112,3 −113,5 −96,7 −74,1 −35,6 −17,8 −1,7

Max. Min.

33 −121

Column E1 C83

C85

−15,2 −56,6 −78,2 −76,1 −93,3 −95,7 −91,9 −73,8 0 −100

C87

3,0 24,4 42,8 41,0 48,0 68,0 64,3 51,0 68 −1

27,4 109,1 108,1 77,3 67,4 47,9 32,4 17,1 119 −36

C89

C91

C93

C95

−6,0 −38,3 −48,3 −87,6 −118,7 −123,4 −115,3 −95,7

−34,7 −188,7 −168,5 −148,4 −146,1 −87,3 −43,7 −1,1

40,1 180,0 171,5 145,3 125,3 73,7 43,8 12,8

−7,9 14,7 41,1 53,4 90,7 115,1 95,8 64,5

1 −125

68 −201

196 −63

116 −9

Table 2. Measured stresses on columns during natural fire, stain gauges 500 mm under ceiling. Measured stresses, MPa Column D1 Time, min.

C97

Column E1 C99

C101

15 30 45 60 90 124 160 226

26,7 106,3 105,6 91,2 77,1 40,3 10,8 −11,8

14,7 74,2 99,7 109,0 103,5 75,2 50,6 28,5

Max. Min.

120 −54

110 −5

Reinforcement of concrete slab

M, φ F, δ

−8,9 −57,5 −74,2 −96,2 −94,4 −56,2 −37,5 −25,3 0 −103

C103

C105

−28,0 −105,0 −99,1 −78,9 −80,9 −35,7 −7,7 12,6

6,3 51,4 80,6 88,1 70,9 56,6 36,7 20,8

43 −116

88 0

C107

C111

26,8 141,1 137,1 121,5 94,5 43,4 4,7 −22,9

−26,6 −131,8 −119,3 −102,9 −90,0 −29,3 5,3 28,1

−2,7 −33,1 −54,5 −67,0 −64,4 −47,8 −37,3 −33,1

159 −69

60 −146

1 −69

F 1

F Ft,Rd

Ft,Rd

2

δ

1 3

2

Fc,Rd c)

b)

δ 3

Fc,Rd a)

C109

d)

Figure 6. Composite joint modelling; (a) header plate connection, (b) assembly of components, (c) components in tension and compression, (d) reinforcement in tension and concrete slab in compression.

modelled using non-linear element COMBIN39 with different force–deformation relationship in tension and in compression, see [8]. Two separate elements with one degree of freedom were 2 (in necessary for each joint. One element was used to simulate the top part of the connection 3 (in compression). Stiffness and resistance of these parts were tension), the other the bottom part 264

Stress, MPa Column D1 100

Measured, 125/127

Predicted, 121/123 4th floor

50

3rd floor Measured, 121/123

0 0

50

-50

100

Time, min. Predicted, 125/127

127

125

123

121

-100

Figure 7. Comparison of calculated stresses at external column to measured ones.

obtained by component method [9] allowing simulation of joint behaviour at high temperatures. This approach allows for interaction of bending and axial loading of the joint. Coupled degrees of freedom were used to transfer the shear forces from beams to column. The analysis employ three loading steps. In the first step, external forces (dead and live loads) were applied. In the second step, the heating phase was simulated applying the elevated temperature on the beam in the fire compartment. The temperature of the top flange was estimated as 0,70 of the lower flange temperature θa at midspan. The temperature at the connections was estimated as 0,62θa and 0,80θa at the top and bottom of the beam respectively. The cooling phase was modelled in the third step applying negative temperature increment to the elements. Load increment with arc length control method was used for the calculation. The structural response from numerical analysis was transferred to time-dependent relationship to present the results. The stresses in the columns at the position of the strain gauges were calculated based on the heating of the unprotected beam D1-D2 (E1-E2) respectively. The growing of the temperature in the structures is expected till 105 minutes. Figure 7 shows the comparison of the predicted and calculated stresses.

6 CONCLUSIONS The collapse of structure or its parts was not reached during the experiment for the fire load of 40 kg/m2 , which represents the fire load in a typical office building, together with a mechanical load greater than standard approved cases. The structure showed good structural integrity. The test results verified the concept of unprotected beams and protected columns as a viable system for composite floors. The test in Cardington January 16, 2003 documents, that FE models allows presuming the steel to concrete composite frame behaviour with a good accuracy. The composite action and joint behaviour play important role in the behaviour of the structure under fire. Simulation of the frame behaviour may increase the fire safety of the buildings. The simple rules for the design of the protected columns ware confirmed in the temperature as well as stress domain.

ACKNOWLEDGEMENTS The authors would like to thank all nineteen members of the project team working on the large scale experiment in Cardington BRE laboratory from October 2002 till January 2003. Special thanks go to Mr. Tom Lennon, Mr. Nick Petty, and Mr. Martin Beneš for careful measurement of data 265

presented above. The project has been supported by the grant of European Community FP5 HPRI – CV 5535. Paper was prepared as a part of project COST C12 of Czech Ministry of Education, Youth and Sport.

REFERENCES [1] Lennon T.: Cardington fire tests: Survey of damage to the eight storey building, Building Research Establishment, Paper No. 127/97, Watford 1997, p. 56. [2] Bravery P.N.R.: Cardington Large Building Test Facility, Construction details for the first building, Building Research Establishment, Internal paper, Watford 1993, p. 158. [3] Wald F., SantiagoA., Chladná M., LennonT., Burges I., Beneš M.: Tensile membrane action and robustness of structural steel joints under natural fire, Internal report, Part 1 – Project of Measurements; Part 2 – Prediction; Part 3 – Measured data; Part 4 – Behaviour, BRE, Watford, 2002–2003. [4] Wald F., Chladná M., Moore D., Santiago A., Lennon T.: The temperature distribution in a full-scale steel framed building subject to a natural fire, paper accepted for ICSCS’04 Soul. [5] Bailey C.G., Burgess I.W., Plank R.J.: Computer simulation of a full-scale structural fire test, The Structural Engineer, 1996, 74(6), pp. 93–100. [6] Wang Y.C.: Steel and composite structures, Behaviour and design for fire safety, Spon Press, London 2002, ISBN 0-415-24436-6. [7] Beneš M., Wald F., Sokol Z., Pascu H.E.: Numerical study to structural integrity of multi-story buildings under fire, in Eurosteel 2002, Coimbra, pp. 1401–1411, ISBN 972-98376-3-5. [8] Sokol Z., Wald F., Pultar M., Beneš M.: Numerical simulation of Cardington fire test on structural integrity, in Mathematical and computer modelling in science and engineering, ed. Koˇcandrlová M., Kelar V., CTU, 27-30.1.03, pp. 339–343, ISBN 80-7015-912-X. [9] Simões da Silva L., Santiago A., Vila Real P.: A component model for the behaviour of steel joints at elevated temperatures, Journal of Constructional Steel Research, 2001, 57(11), pp. 1169–1195.

266

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

On the fire resistance of aluminium alloy structures B. Faggiano, G. De Matteis, R. Landolfo & F.M. Mazzolani University of Naples “Federico II”, Naples, Italy

ABSTRACT: The paper focuses on the importance of refined material modelling of aluminium alloys under high temperatures for correctly estimating the fire resistance of structures. In fact for such structural materials, the evaluation of the bearing capacity of structures, in general, cannot leave apart the proper modelling of the stress–strain relationship, strongly characterized by the material hardening and ductility. Therefore, firstly the effects of high temperatures are investigated at the material level, with reference to the single mechanical properties and then to the stress– strain relationship of the studied aluminium alloys for structural uses. Accordingly, a mechanical model, defined by tailoring the general Ramberg–Osgood formulation, taking appropriately into account the peculiarities of such materials at high temperatures, is provided. The proposed model is introduced in a finite element program for the analysis of structures under fire. Results are presented for a simple portal frame designed for several aluminium alloys.

1 INTRODUCTION The prediction of the mechanical response of aluminium alloy structures exposed to fire is complicated for two principal reasons: (1) the difficulty of developing accurate structural analyses in post-elastic field, taking correctly into account the mechanical features of the basic material, such as the strain-hardening and the limited deformation capacity; (2) the inadequate knowledge of the material behaviour under high temperatures. As a consequence, first of all the specific mechanical properties and the whole stress–strain curve of the material as a function of temperature have to be accurately defined. Moreover, the methods of structural analysis in fire conditions should hold in due account the influence of the shape of the material constitutive law and thus of the kinematic strain hardening on the global behaviour of the structure. Therefore, for allowing practical analysis of complex structures in fire conditions through advanced methods, such accurate material models should be implemented in finite element programs. In this context, a wide examination of the results of experimental tests (ASM Specialty Book, 1993) carried out on different aluminium alloys exposed to high temperatures has been presented (Faggiano et al., 2004a), aiming at characterizing the behaviour under fire in relation to the series and treatments of the aluminium alloys. The variation laws of the following characteristic parameters has been drawn: the elastic modulus (E), the elastic limit stress conventionally defined as 0.2% proof strength (f0.2 ), the ultimate strength (ft ) and the ultimate deformation (et ). Then, a mechanical model, which appropriately represents the peculiarity of such materials subjected to high temperatures, has been proposed. The obtained simplified constitutive law has been introduced in a finite element program for the calculus under fire of structures (Franssen, 1998), with specific reference to the aluminium alloys selected for structural uses by the Eurocode 9 (CEN, prEN 1999-1-2, 2003). Finally, the results of the structural analyses in fire conditions obtained for a simple portal frame and carried out for all the EC9 aluminium alloys have been presented, clarifying the impact of the material modelling on the global response of the structure exposed to fire, evaluated in terms of time up to collapse for a conventional fire scenario (Faggiano et al., 2003; 2004a,b). 267

2 FIRE ANALYSIS OF ALUMINIUM ALLOY STRUCTURES The more significant effects of a fire on a structure are: 1. the deterioration of materials in terms of strength and stiffness, which implies a reduction of the carrying capacity and a worsening of the deformation state; 2. thermal deformations, which may produce increase of second order effects and/or increase of internal stresses. With regard to the fire modelling, reference is generally made to the so called standard fires, which can be simply related to possible real fires by means of equivalence criteria. The design of structures under fire must essentially assure that in case of fire, according to the chosen thermal program, the static safety is guaranteed for a pre-fixed period of time, to be associated to the class of fire resistance imposed by the standard rules. At European levels, the main references for codifications are given by the Eurocodes: EC 1Part 2-2, for the applied loads in fire conditions (CEN, EN 1991-1-2, 2002); EC 2- Part 1-2, for reinforced concrete structures under fire (CEN, prEN 1992-1-2, 2002); EC 3- Part 1-2, for steel structures under fire (CEN, prEN 1993-1-2, 2003); EC 9- Part 1-2, for aluminium alloys’ structures under fire (CEN, prEN 1999-1-2, 2003). Such provisions can be considered exhaustive and reliable in the case of the most common materials for constructions, such as steel and reinforced concrete, while they result more approximated in the case of aluminium alloys structures, mainly due to the difficulties of accurate and reliable material modelling. Such limitation, together with the generally high vulnerability to fire of structures made of aluminium alloys, represents a strong incentive to check the opportunity of applying for code purposes a better schematization of the mechanical behaviour of the material, so to allow the maximum exploitation of the resistance resource of the material.

3 MECHANICAL FEATURES OF ALUMINIUM ALLOYS AT HIGH TEMPERATURES Common aluminium alloys melt at about 600◦ C and loose the 50% of their original strength at about 200◦ C (Mazzolani, 1994; Lundberg, 1995). The trends of the principal behaviour parameters, such as the conventional yielding stress f0.2,T , the ultimate strength ft,T , the ratio ft,T /f0.2,T and the elongation at rupture et,T as a function of the temperature, is depicted in Faggiano et al. 2004a, on the basis of data of experimental tests, carried out on different aluminium alloys exposed to high temperatures, available from existing technical literature (ASM Specialty Book, 1993). The variation laws were differentiated as respect to the aluminium alloy series: Series 1000 – Pure aluminium and low alloy content alloys; Series 2000 – Aluminium-copper alloys; Series 3000 – Aluminium-manganese alloys; Series 5000 – Aluminium-magnesium alloys; Series 6000 – Aluminium-silicon-magnesium alloys, Series 7000 – Aluminium-zinc alloys. Moreover, curves were drawn for the undergone treatments, such as for alloys in the work hardening state (H), for alloys in the hardening state due to heat treatment (T), for the alloys in the annealed state (O). From the analysis of results, it has been noticed that the conventional yielding stress and the ultimate stress exhibit a significant loss in general for T >100◦ C, in particular at T = 250◦ C the strength reduction is about 70–80% for the alloys in the work hardening state (H) and the ones being in the hardening state by means of heat treatment (T), besides the alloys in the annealed state (O) show a less decay of strength, which is about 30–50%. Moreover heat treated and work hardened alloys (T and H types) are characterized by ultimate strength remarkably larger than the alloys in the annealed state (type O), only up to temperatures of about 150◦ C, whereas for higher temperatures the effects of strength improvement due to hardening treatments is lost. It can be also noted that while at room temperature and up to T ∼ = 150◦ C annealed state alloys (O type) present a strain hardening ratio (ft,T /f0.2 ) and an ultimate elongation (et,T ) about twice larger with respect to the heat treated and work hardened alloys (T and H types), such difference strongly reduces as far as the temperature increases. In particular, for not-treated alloys the strain hardening decreases with the increment of the temperature, while in the case of the treated alloys it increases, 268

Figure 1. EC9 aluminium alloys’ mechanical properties as function of temperatures (f [N/mm2 ], et [%]).

approaching in both cases a value next to 1.5. As far as the ultimate elongation is concerned, it experiences a revival with the increase of the temperature. Then, it can be generally observed that treatments, such as the tempering and the plastic working processes, improve the material strength, but in the meantime they reduce both the effect of the strain hardening and the extent of the ultimate elongation. In Figure 1, diagrams of the mechanical properties are drawn with specific reference to the aluminium alloys selected by the Eurocode 9 for structural uses. In particular the reduction of the conventional strength f0.2,T is given through the reduction coefficient k0.2,T , which provides for each alloy the ratio between the elastic strength at a given temperature (f0.2,T ) and the elastic strength at room temperature (f0.2 ). In Figure also the elastic modulus (ET ), as a function of the temperature is depicted. It can be observed that ET decreases as far as the temperature increases, independently from the alloy and its treatment. From further data available in literature (Conserva et al., 1992), it comes out that the resistance of the aluminium alloys, given in terms of both conventional yielding stress and ultimate strength, decreases as far as the exposure time to an assigned temperature increases. On the contrary, the ultimate elongation, and then the material ductility, increases with the prolonged permanence at high temperatures. Finally, concerning the thermal properties, such as the thermal expansion, the specific heat and the thermal conductivity, they undergo an increment with the temperature increase. Besides, the variation law of such properties, as it is indicated in the EC9, are independent from the alloy. 269

4 STRESS–STRAIN RELATIONSHIPS FOR ALUMINIUM ALLOYS AT HIGH TEMPERATURES The modelling of the constitutive law of aluminium alloys is difficult already at room temperature, due to the high variation in the mechanical behaviour among the different alloys and the adopted fabrication processes. Besides experimentally obtained σ–ε curves are not well fitted by simplified diagrams, like the elastic-perfectly plastic one, due to the remarkably continuous hardening exhibited in the plastic range. For the structural analysis under high temperatures, EC9 suggests the adoption of a simplified relationship of the elastic-perfectly plastic type, exclusively supplying the variation of the elastic modulus and the conventional yielding stress with temperature (Fig. 2a). In order to more correctly interpret the evolution of the mechanical characteristics of the material with the temperature, the general expression of the Ramberg-Osgood law for analyses in the large deformation field can be adapted, by introducing the variation laws with the temperature of all the relevant mechanical parameters, such as f0.2,T , et,T and ft,T , as shown in Figure 2b. The n exponent measures the strain hardening of the alloy, ruling the shape of the curve in the post-elastic field. In particular, for n factor values approaching to 0, the Ramberg–Osgood law gives an indefinitely elastic behaviour, while for large values an elastic-perfectly plastic behaviour is obtained. On the basis of the elaboration of the variation laws for the single mechanical parameters, in Figure 3a, for each EC9 alloy, the strain hardening factor n obtained at growing temperatures are

Figure 2. Typical constitutive laws for the aluminium alloys: (b) Ramberg–Osgood relationship.

Figure 3a. The strain hardening factor n as a function of temperature.

270

(a) elastic-perfectly plastic law

depicted. In particular, in order to evidence the influence of the mechanical properties on the strain hardening factor n, the value obtained considering the actual variation of the single mechanical parameters with temperature (n-analytical) is compared with the value obtained taking the ultimate elongation as constant and equal to the one at room temperature (n-et = const), and with the n value at room temperature assumed as constant with the temperature (n-constant). First of all, it can be observed that for all the alloys the n value at high temperatures is remarkably different with respect to the one at room temperature. Furthermore the n(T) relationship does not present a single trend for the different examined materials, according to the hardening ratio ft,T /f0.2,T (Fig. 1). In particular, it can be noted that for not treated materials (O type) at increasing temperatures the strain hardening factor exhibits an increment larger than 50% with respect to the value at room temperature. On the contrary, for treated materials (H and T types) the n value is larger than the corresponding value at room temperature only up to a temperature of about 200◦ C, over which there is a reversal trend, with values of the n factor lower than the ones at room temperature. Moreover it can be observed that the variation with the temperature of the elongation at collapse et has not a significant influence on the strain hardening factor n. As a consequence, in order to simplify the mathematical expression of the strain hardening factor, the ultimate elongation of the material could be actually taken constant and equal to the one at room temperature. In order to emphasise the influence of the strain hardening factor on the constitutive law of aluminium alloys at high temperatures, in Figure 3b, for every studied alloy, the σ–ε curves are represented at different temperatures (24◦ C, 150◦ C, 315◦ C), considering different hypotheses for the hardening factor evaluation, namely: (a) actual hardening factor n(T) (n-analytical); (b) constant value equal to the one at room temperature (n-constant); (c) constant value equal to 200 (n-200), the latter being representative of an elastic-perfectly plastic mechanical behaviour. As it appears,

Figure 3b. σ (N/mm2 )–ε laws for the aluminium alloys at different T (◦ C).

271

generally, at room temperature and up to T ∼ = 150◦ C for annealed state alloys (O type) the effect of strain hardening is not negligible with respect to the heat treated and work hardened alloys (T and H types), such difference reduces as far as the temperature increases. Moreover the assumption of an elastic-perfectly plastic simplified model is always on the safe side. 5 INFLUENCE OF MATERIAL MODELLING ON THE GLOBAL BEHAVIOUR OF ALUMINIUM ALLOY STRUCTURES AT HIGH TEMPERATURES Aiming at assessing the influence of the material modelling on the fire resistance of aluminium structures, the structural analysis in fire condition of a simple plane portal frame made of different aluminium alloys is presented. The geometric characteristics, the load conditions, the fire event model as well as the exposure condition to fire of the structural members are presented in Figure 4, together with the types of considered alloys and the result of the member sizing. For the analysis of the structure under fire, the finite element program SAFIR 2002, developed by J.M. Franssen of the University of Liége, is used. Such a software has been upgraded by implementing the material models for aluminium alloys provided by EC9. In particular, in order to assess the influence of the strain hardening on the behaviour of the structural complex, different material models based on the Ramberg–Osgood law are considered, corresponding to the above mentioned hypotheses for the hardening factor evaluation, namely: (a) n-analytical, (b) n-constant and (c) n-200. In Figure 5 the results of the analyses, given in terms of fire resistance of the structure (R) expressed in minutes, are specified. The structural resistance R has been also normalised, considering as reference value for each alloy the one corresponding to the material model based on the actual strain hardening value (n-analytical). In the above analyses, the considered collapse condition corresponds to the loss of stiffness of the frame, which, in combination with the deterioration of the mechanical material behaviour induced by temperature, is not more able to balance the applied loads. Figure 6 allows to go more deeply into details of the results, showing for each examined alloy and for each mechanical model assumption the following behavioural parameters: the fire resistance (R) in seconds; the corresponding maximum temperatures (Tmax ) attained within the structural members; the conventional yielding stress at room temperature (f0.2 ) and the degraded values at collapse temperatures (f0.2,Tmax ); the strain hardening factor n at room temperature (n-constant) and at the collapse temperature (n-analytical-Tmax ). Both the analytical and the graphical representation is presented, for a simpler understanding of outcomes. The examination of the obtained results evidences that the effect of the strain hardening factor on the global behaviour of the structure is similar to that already obtained at the material level. It

Figure 4. Study cases.

272

Figure 5. Fire resistance for the study cases.

Figure 6. The results of the structural analysis under fire for the study cases.

can be observed that as far as the mechanical refined model is assumed, the time up to collapse and the corresponding maximum temperature within the members increases. This is strictly evident for treated alloys (H and T types); whereas in the case of annealed alloys (O type), the fire resistance of the portal frame is higher in case of material model with constant value at room temperature of the n factor. In fact the n value at room temperature for the annealed alloy is the lowest one, it growing with the temperature. As a consequence, the assumption of a constant value at room temperature 273

of the n factor determines an overvaluation of strain hardening with respect to the effective one at high temperatures. The worst fire resistance is observed when the simplified elastic-perfectly plastic mechanical model is considered. In particular, in the most unfavourable conditions (3003-O and 5052-O alloys), a fire resistance reduction of the structure of about 30% is gained. It should be also pointed out that even when strain hardening effects are not considered (n-200), the time up to collapse of O type alloys are more favourable as respect to the T and H type alloys, due to the fact that for the latter ones the strength degradation at high temperature is more sharp. After all, it can be concluded that not treated alloys (O) give rise to the best behaviour under fire. However, it is worthy noticing that presented results are somewhat restricted, they being referred to a simple portal frame structures. Moreover it has to be evidenced the limited collapse time, which is equal to about 7 minutes as an average. Anyway, this limited fire resistance refers to a specific structural scheme, considering a standard fire model and the worst exposure condition of the frame members (unprotected, 4 sides’ exposure and inconvenient cross-section shape). As a consequence considerable margins of improvement of the performances to fire of the structure exist. Nonetheless it has to be evidenced that the main aim of the study is more the comparative evaluation of the strain hardening at high temperatures for different aluminium alloys for structural uses and its impact on the structural fire resistance than the optimization of the behaviour of structures under fire conditions. 6 CONCLUSIVE REMARKS In order to set up advanced fire design methods for aluminium alloy structures, in this paper the influence of temperature on the mechanical properties of the material has been analysed, taking into consideration several alloys for structural uses. Particular attention has been focused on material strain hardening, which characterises the mechanical behaviour of the material in plastic range. In fact, as a first result of the study, it has been pointed out that simplified mechanical models, such as the elastic-perfectly plastic one, generally are not able to correctly characterize the material behaviour at the high temperatures, since they disregard the beneficial effect due to continuous material hardening, which is somewhat effective in balancing strength decay due to high temperatures. Therefore, in order to take specifically into account the effect of the strain hardening, a more comprehensive mechanical model for the aluminium alloys has been proposed, based on the well known Ramberg–Osgood law, which is able to represent in an appropriate manner all the peculiarities of such materials exposed to high temperatures. Then, such model has been introduced in the finite element program for global analysis of structures subjected to fire, with specific reference to the aluminium alloys selected by the Eurocode 9 for structural uses. The structural analysis in fire conditions of a study case related to a simple portal frame has pointed out the remarkable effect of material modelling of aluminium alloys, since the adoption of elastic-perfectly plastic model results very conservative and not convenient for a material which exhibits a so rapid strength decay with high temperature. Finally, not treated alloys (O) give rise to the best behaviour under fire due to the beneficial effect of material strain hardening and to the fact that the strength degradation at high temperature is softer than for treated alloys. REFERENCES ASM Specialty Handbook, 1993. Aluminium and aluminium alloys. Edited by J.R. Davis & Associates. CEN (European Communities for Standardisation), EN 1991-1-2, 2002. Eurocode 1: Basis of design and actions on structures – Part 2-2: Actions on structures - Actions on structures exposed to fire. CEN (European Communities for Standardisation), prEN 1992-1-2, 2002. Eurocode 2: Design of reinforced concrete structures – Part 1-2: General rules-Structural fire design. CEN (European Communities for Standardisation), prEN 1993-1-2, 2003. Eurocode 3: Design of steel structures – Part 1-2: General rules-Structural fire design.

274

CEN (European Communities for Standardisation), prEN 1999-1-2, 2003. Eurocode 9: Design of aluminium structures – Part 1-2: General rules-Structural fire design. Conserva M., Donzelli G., Trippodo R., 1992. Aluminium and its application. EDIMET. Faggiano B., De Matteis G., Landolfo R., Mazzolani F.M., 2003. On the behaviour of alluminium alloy structures exposed to fire. Proceedings of the XIX National Congress C.T.A. “III week of steel Constructions”, 28–30 September, Genova, Italia, 393–406 pp. Faggiano B., De Matteis G., Landolfo R., Mazzolani F.M., 2004a. The influence of material modelling on the fire resistance of aluminium alloy structures. Proceedings of the 9th International Conference on Aluminum Structural Design (INALCO 2004), Cleveland, Ohio (USA), 2–4 June. Faggiano B., De Matteis G., Landolfo R., Mazzolani F.M., 2004b. Effects of high temperatures on the resistance of aluminium alloy structures. Proceedings of the 7th International Conference on Modern Building Materials, Structures and Techniques, Vilnius, Lithuania, 19–21 May. Franssen J.M., 1998. SAFIR98a – User’s Manual. University of Liege. Lundberg S., 1995. Design for fire resistance. Training in Aluminium Application Technology (TALAT) EUCOMETT Program, F. Ostermann Ed., Aluminim Training Partnership, Brussels: Section 2500. Mazzolani F.M. Aluminium alloy structures. 2nd edn., E&FN Spon, London, 1994, UK. Wald F., Bosiljkov V., De Matteis G., Haller P. Vila Real P., Structural integrity of buildings under exceptional fire. Proceedings of the 1st Cost C12 Seminar, Lisbon 18–19 April, 2002.

275

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Nonlinear stress–strain behavior of RC elements exposed to fire M. Cvetkovska & L. Lazarov Faculty of Civil Engineering, Skopje, Macedonia

ABSTRACT: Determining the fire response of structural elements and their assemblies is a complex problem of nonlinear analysis in which the strength and the stiffness of the elements as well as the inner forces are continuously modified. To solve this problem the computer program FIRE was developed. The program was verified on the bases of the experimental investigation results available in literature. This paper describes the analytically achieved results for the fire resistance of centrically loaded RC columns. The influence of: element geometry, concrete cover thickness, steel ratio and intensity of axial force are analyzed. Four RC beams, exposed to different fire models are analyzed too, and the predicted results are compared with those experimentally achieved by other researchers. KEYWORDS:

temperature, fire resistance, structural response, RC columns, RC beams.

1 INTRODUCTION Today, as a result of many years of investigations, there are three basic methods for determination of fire resistance of structural elements and their assemblies. The oldest method is the performance of a fire test of loaded elements, in compliance with the national regulations and standards, or comparison of the elements with the results from already performed tests on similar or identical elements. The second method implies the use of empirical formulae that are based on the results from performed fire tests and holds for a certain combination of: structure, material and protective coating. The third method represents an analytically elaborated approach to design elements with a predefined fire resistance and it is based on the principles of structural mechanics and theory of heat transfer. Generally, analytical computation of the whole structure for the case of fire means proof that the structure, or its elements, loaded by a defined load and exposed to thermal effect, satisfies certain functional requirements, expressed through the ultimate state of bearing capacity and usability. To define the fire resistance of structures as assemblies of structural elements, experimental investigations of models are almost impossible. The time dimension of spreading of the temperature field is practically impossible to be simulated on a model of small proportions. Hence, model investigations can hardly be accepted due to the high cost. For the last twenty years, particular importance has therefore been given to analytical definition of the problem. This paper presents a computational procedure for the nonlinear analysis of a reinforced concrete elements and plane frame structures subjected to fire loading. The program FIRE (Cvetkovska 2002) carries out the nonlinear transient heat flow analysis (modulus FIRE-T) and nonlinear stress–strain response associated with fire (modulus FIRE-S). The solution technique used in FIRE is a finite element method coupled with time step integration. The used analysis procedure does not account for the effects of large displacements on equilibrium equations. 2 NONLINEAR FINITE ELEMENT ANALYSIS 2.1 Nonlinear stress–strain analysis The response of a reinforced concrete elements and plane frame structures exposed to fire is predicted by modulus FIRE-S. This modulus that is a modification of the computer program 277

FIRES-RC II (Iding et al. 1977), accounts for: dimensional changes caused by temperature differences, changes in mechanical properties of materials with changes in temperature, degradation of sections by cracking and/or crushing and acceleration of shrinkage and creep with an increase of temperature. To define the fire response of reinforced concrete structure is thus a complex nonlinear analysis problem in which the strength and stiffness of a structure as well as internal forces continually change due to restraints imposed by the structural system on free thermal expansion, shrinkage, or creep. Because linear elements and frames are modeled as an assemblage of members connected to joints, the basic analytical problem is to find the deformation history of the joints U (t) when external loading at the joints R(t) and temperature history within the members T (t) are specified. Since only linear elements and two dimensional frames are considered, each joint has three degrees of freedom, two translations and one rotation. Likewise, there are two forces and a moment at each joint. The overall system stiffness matrix of a structure is assembled by incorporating the stiffness contribution of each member. Each member is treated as a linear beam element modeled by simple beam theory and is composed of a linear elastic material. In a reinforced concrete, the first condition holds if members are of usual properties, but the second condition is violated for virtually all loading conditions. The materials in a reinforced concrete structure are nonlinear and detailed knowledge of the strain states existing within members is necessary in order to obtain the member stiffness matrix. Hence, the member must be further discretized. Structure members are subdivided into a number of segments such that by calculating segment properties, it is possible to determine overall member properties. Each segment is treated as a standard beam element in which axial force is assumed to be constant and bending moment to vary linearly along the length of a segment. However, in order to calculate section properties and internal forces and moments within a segment, it is necessary to discretize cross sections further into subslices (Fig. 1a). The subslices associated with each segment can be envisioned as uniaxially loaded prisms. Therefore, only uniaxial stress states are considered, equivalent to the assumption that the effect of multiaxial stress components is negligible. This type of prismatic model allows only the effects of axial and flexural stiffness to be considered in modeling structural behavior, neglecting the effect of shear in member idealization. Local effects near member ends are not considered. However, in most linear structures these effects are of secondary importance and principal structural action is due to flexure and axial deformation. Internal axial force and bending moment at each end of a segment are found by summing the force and moment contributions of subslices discretizing the cross section. The mechanical strain in each subslice can be found directly from the extension and curvature of a cross section:

r,j

where εi = total mechanical strain in the subslice r for time step i and iteration j; εa = total strain in the reference plane; y = distance from subslice to reference plane; φ = curvature; DCG = distance from the reference plane to centroidal plane (Fig. 1b). In order to find current stress in each subslice, that part of the current total mechanical strain which gives rise to stress, must be determined:

(a)

(b)

Figure 1. (a) cross-section of a segment discretized to subslices, (b) deformation mode of segment.

278

j

f

where εi = stress related strain for time step i and iteration j; εi = free strain for time step i. Free strains are determined for each subslice at the beginning of a time step and do not vary during iteration within a time step. For the concrete, or the steel subslice they are calculated according to the following equations:

f

where εi = free strain for concrete, or steel subslice, for time step i; εicr = free creep strain accumulated over current time step i; εitr,c = transient strain accumulated only in concrete subslice over current time step i; εith = free thermal expansion accumulated over time step i. The “σ −ε” relations, recommended by Eurocode 2, part 1.2, take into account the creep of concrete and steel at elevated temperatures. It is done by moving the maxima in the stress-strain curves to higher strains with higher temperatures. When compressed concrete is heated for the first time, the total strain is different from the total strain measured in constant temperature creep tests and an additional irrecoverable “transient strain” must be taken into account. This transient strain is a function of the level of stress and the thermal expansion (Anderberg et al. 1986). The increment of transient strain at any given time step can be computed as:

where εitr,c = transient strain accumulated over current time step i; ε th,c = free thermal expani sion accumulated over current time step i; σc = applied compressive stress; fc = compressive strength at ambient conditions. When the transient strain is not included in the structural analysis, the result is stiffer structural model at elevated temperatures, in which the computed thermal stresses become very large and failure is predicted to occur much earlier than experimentally observed (Ellingwood & Lin 1991). 3 TEST EXAMPLES TO VERIFY THE COMPUTER PROGRAM FIRE 3.1 Fire resistance of centrically loaded RC column In 1987 a test program was conducted by the Portland Cement Association (PCA) and the National Research Council of Canada (NRCC) to determine the fire resistance of centrically and eccentrically loaded reinforced concrete columns (Lin et al. 1992). Columns were 3.8 m long, and had 38 mm concrete cover to the tie bars. They were fabricated with either siliceous or carbonate aggregate concrete. The specified yield strength of the reinforcement was 400 MPa (deformed bars). The columns were exposed to the standard fire ASTM E119 from all sides. Basic characteristics of the centrically loaded columns are listed in Table 1. For comparative purposes, test loads N were divided by the maximum factored axial load allowed by ACI 318, Nd = Nn , and the load/strength ratio α  was defined as: α  = N /Nd . For columns reinforced with tie reinforcement the equation for Nd is:

where fc is the strength of the concrete, fy is the specified yield strength of the reinforcement, Ag is the gross area of the section, As is the area of the reinforcement,  is the strength reduction factor which is equal to 0.7, for members with tie reinforcement, and to 0.75, for members with spiral reinforcement. The fire resistance of the columns is predicted by the computer program FIRE. No data exist for the temperature dependent thermal and mechanical properties of concrete and steel, so in program 279

Table 1. Test data for centrically loaded columns. Column no.

Concrete strength (MPa)

Test load (kN)

Load/ strength ratio α 

Test duration Hr.:min.

Predicted fire resist. Hr.:min.

Deviation in %

Type of failure (test)

Cross section: 304 × 304 mm, steel ratio: 2.19% (4φ25), siliceous aggregate S1 34.1 0 0.00 4:00 >6:00 / S13 40.3 340 0.15 5:40 5:45 +1.5 S4 35.0 710 0.36 3:40 4:00 +9.1 S25 39.6 800 0.37 4:02 4:00 −0.8 S17 50.3 1070 0.41 3:54 3:51 −1.3 S3 34.0 800 0.41 3:38 3:42 +1.8 S16 52.9 1180 0.43 3:47 3:45 −0.9 S31 41.5 1024 0.45 3:41 3:36 −2.3 S9 38.3 1335 0.63 3:07 2:51 −8.6 S2 36.8 1335 0.65 2:50 2:43 −4.1 S8 34.8 1780 0.90 2:26 1:57 −19.8

none compres. compres. compres. compres. compres. compres. compres. compres. compres. compres.

Cross section: 304 × 304 mm, steel ratio: 2.19% (4φ25), carbonate aggregate S10 40.8 800 0.36 8:30 5:00 −41.2 S11 36.8 1070 0.52 6:06 4:06 −32.8 S12 40.0 1780 0.81 3:35 3:00 −16.3

compres. compres. compres.

Cross section: 304 × 304 mm, steel ratio: 4.38% (8φ25), siliceous aggregate S20 42.5 980 0.36 4:12 3:54 −7.0 S21 37.0 1335 0.53 3:45 3:10 −15.5

compres. compres.

Cross section: 203 × 203 mm, steel ratio: 2.75% (4φ19), siliceous aggregate S6 42.3 169 0.16 3:05 3:27 +11.9

buckling

Cross section: 406 × 406 mm, steel ratio: 2.47% (8φ25), siliceous aggregate S5 40.7 0 0.00 5:00 >5:00 / S22 38.8 2420 0.62 4:22 4:20 −0.8

none compres.

Cross section: 304 × 304 mm, steel ratio: 2.22% (8φ22), siliceous aggregate S27 42.4 1415 0.41 5:56 4:50 −18.5

compres.

Cross section: 203 × 914 mm, steel ratio: 1.22% (8φ19), siliceous aggregate S28 42.0 756 0.16 5:35 5:12 −6.8

compres.

FIRE they are taken as recommended by Eurocode 2, part 1.2. When siliceous aggregate columns were tested temperatures were measured along the centerline, these data are compared with the analytically achieved temperatures (Fig. 2) and a good agreement is noticed. There is no information for the measured temperatures in the carbonate aggregate columns. The difference between the measured and predicted fire resistance for the columns S10, S11 and S12 (Table 1) indicates that the values for the thermal conductivity and the specific heat of the carbonate aggregate concrete, recommended in EC2, are not adequate, but the predicted results are on the side of safety. In the literature there is a considerable scatter in the recommended values for these two parameters. For all other columns there is a good agreement between the calculated and experimentally achieved fire resistance (max. deviation is −8.9%). It is not the case only for columns S4 and S8, but probably the reason is the dispersion of the experimental results. Test data indicate that the end conditions and the effective length of the columns were not significant factors for the fire resistance of the columns. All the columns failed in compression, and only S6 failed in a buckling mode. The reason is the high slenderness of this column. The computer program FIRE does not account for the effects of large deformations, so the numeric results are 12% higher than experimentally achieved. 280

t=1h t=1h* t=2h t=2h* t=3h t=3h*

800 600 400 200 0 0

(a)

3

6

9

distance from exposed surface (cm), (measured along centerline)

800 600 400 200 0

15

12

t=1h t=1h* t=2h t=2h* t=3h t=3h* t=4h t=4h*

1000 Temperature (°C)

Temperature (°C)

1000

0

4 8 12 16 distance from exposed surface (cm), (measured along cemterline)

(b)

20

Figure 2. Predicted* and measured concrete temperatures in the cross section of: (a) 304 × 304 mm, (b) 406 × 406 mm, siliceous aggregate columns. 1

' =N/Nd

0.8 0.6 0.4

experiment FIRE (fc<40MPa)

0.2

FIRE (fc>40MPa) 0 0

1

2

3

4

5

6

fire resistance (h)

Figure 3. Fire resistance of 304 × 304 mm siliceous aggregate columns.

The effect of restraining was also considered. While constant loads were maintained on the reference columns, the initial load was increased as required to prevent expansion. The average decrease in test time due to fully restraining the columns during the fire test was 6%. The effect of concrete strength was evaluated by comparing results for high strength siliceous aggregate column S17 with results for the reference column S3. For the same load/strength ratio the difference is only 4%. The type of the aggregate, the cross sectional geometry and the load/strength ratio α  are significant factors affecting fire resistance of centrically loaded columns. Since it was anticipated that both carbonate and lightweight aggregate concrete columns would have a better fire resistance than the siliceous aggregate concrete columns, most of the tests were preformed on the last one. A relationship of fire resistance of siliceous columns to load/strength ratio was obtained and presented on Figure 3. 3.2 Fire resistance of RC beams In 1987 six reinforced concrete beams were cast at the Construction Technology Laboratories of the Portland Cement Association (Ellingwood & Lin 1991). All beams were designed according to ACI Standard 318. Beams were fabricated using normal-weight carbonate concrete and Grade 60 deformed reinforcing bars. Only four of them are analyzed in this example. Figure 4 provides details of the beam specimens. Beams B1 and B3 were tested using the ASTM E119 fire exposure and beams B5 and B6 were exposed to a short duration, high intensity (SDHI) expose. Specimens were tested to simulate the end span of a continuous beam. This was accomplished by maintaining the cantilever end of the beam at a constant elevation during the course of the fire test by changing the cantilever load Fo as required. The loads applied to the simply supported span were held constant during the test (F = 44.48 kN). 281

Beam B1 and B5

Beam B3 and B6

el.1* el.2* el.3* el.4*

600 400

500

el.1*** el.2*** el.3*** el.4***

Temperature (°C)

Temperature (°C)

Figure 4. Reinforcement details and cross sectional geometry of the beam specimens.

200

400 300 200

el.1* el.2* el.3* el.4*

100 0

0 0

1

(a)

2 time (h)

3

0

4

1

(b)

2 time (h)

el.1*** el.2*** el.3*** el.4*** 3

4

Figure 5. Comparison of reinforcement temperatures: measured(∗) and predicted by program FIRE(∗∗∗) (a) B1 (ASTM fire), (b) B5 (SDHI fire). 15 max. deflection (cm)

max. deflection (cm)

20 measured* T.D.Lin** FIRE

16 12 8 4

9 6 3 0

0 0 (a)

measured* T.D.Lin** Platten*** FIRE

12

1

2 time (h)

3

0

4 (b)

1

2

3

4

time (h)

Figure 6. Comparison of measured maximum deflections of: (a) beam B1 (ASTM fire), (b) beam B3 (ASTM fire), and predicted by different computer programs.

Beams are analyzed using the computer program FIRE. Figure 5 compares the measured and numerically achieved reinforcement temperatures for beams B1 and B5, exposed to ASTM E119 and SDHI fires. The comparison is satisfactory except for the reinforcement No.1 (deviation is max. 40% for ASTM fire, and max. 20% for SDHI fire), but the thermocouples at the exposed surface might not give accurate readings, since they are often damaged by oxidation at extremely high temperatures. Since the actual coefficient of thermal conductivity λc and specific heat ρc for concrete used in fabricating the beams were not known, comparative thermal analysis were performed using the relations recommended by Eurocode 2, part 1.2. Ellingwood & Lin (1991) had modified the computer program FIRES-RC (Iding et al. 1977) by involving the transient strains according to Equation 5 and this program was used to perform the structural analysis of the beams. All six beams developed significant shear cracks near the continuous support early in the fire, but eventually failed from excessive flexural cracking and deformation, so the effect of the shear forces was neglected in this program. The same procedure is used in the computer program FIRE. Huang & Platten (1997) had developed a nonlinear finite 282

8 max. deflection (cm)

max. deflection (cm)

8 6 4

measured* T.D.Lin** Platten*** FIRE

2 0

4

measured* T.D.Lin** Platten*** FIRE

2 0

0 (a)

6

1

2 time (h)

3

4

0 (b)

1

2 time (h)

3

4

Figure 7. Comparison of measured maximum deflections of: (a) beam B5 (SDHI fire), (b) beam B6 (SDHI fire), and predicted by different computer programs.

Figure 8. Isotherms and degradation profiles in the cross section of the beams, at the interior support.

element model based on the “plane stress” theory, so the program FPPRCM-S involved the effect of shear forces. Figures 6 and 7 compare the measured and numerically achieved maximum deflections of beams B1 and B5, using different computer programs. In all three cases the structural analysis models of the beams analyzed were stiffer then the beams actually tested, leading to predicted deflections at ambient temperature (time t = 0) that were less then those measured. The apparent reason for this difference is the presence of several flexural cracks in the beams that appeared in the vicinity of the continuous support during the period before the fire test. This cracking caused a slight loss of rotational stiffness and a concentration of curvature in the vicinity of the support, leading to an increase in deflection in the exposed span. During the fire test the high thermal gradient in the cross section of the beams caused more cracks and the rotational stiffness of the beam models became more close to the real one, so the predicted and measured deflections became close too. The effect of creep at elevated temperatures in the program FIRE is involved by the temperature dependent stress–strain relationships for concrete and steel, recommended in EC2. They are defined while specimens are subjected to ASTM E119 (or ISO 834) fire model, so they are not adequate 283

for SDHI fire model. The other two programs, mentioned above, use a different approach. They involve the creep strains based on a temperature compensated time model that reflected the thermal acceleration of creep. The approach used in FIRE provides better agreement between the calculated and experimentally achieved deflections in case when ASTM fire model is used (Fig. 6), but that is not a case during the cooling phase, when beams are subjected to SDHI fire model (Fig. 7). In the case of ASTM E119 (or ISO 834) fire model the temperatures in the cross section of the beams continually increase (Fig. 8a). In the case of SDHI fire model, during the cooling period, the isotherms are closed curves (Fig. 8c). The concrete directly exposed to fire develops large compression stresses due to thermal gradients. At the interior support fire has the same effect as the load does, so concrete at the bottom of the cross section is in compression, while concrete at the top of the cross section cracks (Fig. 8b, 8d). At the middle of the span the effect is opposite.

4 CONCLUSIONS The model proposed in this study is capable of predicting the fire resistance of planar reinforced concrete structural members with a satisfactory accuracy. The computer program FIRE have been developed as analytical tool to study the fire response of reinforced concrete frame structures. Histories of: displacements, internal forces and moments, stresses and strains in concrete and steel reinforcement, as well as current states of concrete (cracking and crashing) and steel reinforcement (yielding) are calculated subject to temperature field development in the thermal time history of the structure. Since a physical testing program for investigating the response of a large variety of structural elements under differing restraint, loading, and fire conditions is impractical and expensive, analytical studies supported by the results of physical experiments could efficiently provide the data needed to resolve questions related to the design of structures for fire safety. Parametric studies, helping to identify important design considerations, could be easily achieved throughout implementation of this program. The time response capability of FIRE can also be used to assess potential modes of failure more realistically and to define the residual capacity of structure after attack of fire. REFERENCES Anderberg, Y. Magnusson, S.E. Pettersson, O. Thelandersson, S. & Wickstrom, U. 1986. An analytical approach to fire engineering design of concrete structures. International Symposium on Concrete and Structures. ACI Publications SP-55-16: 409–437 Bazant, Z.P. & Kaplan, M.F. 1996. Concrete at high temperature: Material properties and mathematical models. London: Longman Group Limited Cvetkovska, M. 2002. Nonlinear stress strain behaviour of RC elements and RC frames exposed to fire. Phd. thesis. Skopje: University St.Cyril and Methodius Ellingwood, B. & Lin, T.D. 1991. Flexure and shear behavior of concrete beams during fires. Journal of the Structural Engineering. Vol.117, No.2: 440–458 Huang, Z. & Platten, A. 1997. Nonlinear Finite Element Analysis of Planer reinforced Concrete Members Subjected to Fire. ACI Structural Journal, Vol.94, No.3: 272–282 Iding, R. Bresler, B. & Nizamuddin, Z. 1977. FIRES-RC II A computer program for the fire response of structures-reinforced concrete frames. Report No. UCG FRG 77–8. Berkeley: University of California Lin, T.D. Zwiers, R.I. Burg, R.G. Lie, T.T. & McGrath R.J. 1992. Fire resistance of reinforced concrete columns. PCA Research and Development Bulletin RD101B. Skokie, Illinois: Portland Cement Association

284

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Fire resistance of garage plate-wall prefabricated structure ˇ R. Cajka & P. Mateckova Faculty of Civil Engineering, VSB-Technical University of Ostrava, Czech Republic

ABSTRACT: Authors have been engaged in investigation of fire resistance of concrete slab determinate structures. In this paper the research findings are applied in mechanical behaviour of indeterminate wall-plate structure exposed to high temperatures. In indeterminate structures are developed, besides the internal forces caused by mechanical loads, also internal forces caused by non-uniform and uniform cross-section warming. Author’s perspective aim is to find suitable mathematic model for analysis of global structural stress–strain conditions. KEYWORDS: Fire resistance, indeterminate structure, heat exposure model, structural response model, additional internal forces. 1 FIRE RESISTANCE Fire protection of building structures covers a number of requirements, which include also fire resistance of the building structures. Currently, the fire resistance of the building structures is determined mostly by testing in an accredited laboratory (PAVÚS, Fire Attestation and Research Institute for Civil Engineering). Thanks to a number of measurements and experiments which have been carried out in the Czech Republic and abroad, it is possible to calculate correlation between thermal and mechanical properties of materials used in the building structures and the temperature. Those correlations are known also for high temperatures which occur during the fire. New findings have enabled the calculation of the fire resistance of the structure. A progressive method of calculation of the fire resistance is also mentioned in the body of European standards listing applicable dependencies of mechanical and thermal properties on the temperature for individual building materials. The scientists however have not managed yet to determine, with a sufficient accuracy, development of the temperature in the reinforced-concrete cross-section, especially with more complicated geometry and to estimate the mechanical behaviour of the structure with an increased temperature. In the paper the fire resistance of indeterminate plate-wall reinforced prefabricated structure is analysed. 2 PREFABRICATED PLATE-WALL STRUCTURE 2.1 Static model and action on structure The object of fire-resistance analysis is the prefabricated reinforced concrete plate-wall structure, (Fig. 1). The structure is considered as closed frame exposed to self weight, the car on the bottom (qk1 = 2.5 kN · m−2 ), the snow on the ceiling (qk2 = 1.5 kN · m−2 ), and the wind on walls of the frame (pressure qk3a = 0.44 kN · m−2 , admission qk3b = 0.33 kN · m−2 ). With respect to the length of the structure (L = 6 m), the two-way slab effect is neglected. The separate part of actions is the action caused by non-uniform and uniform cross-section warming, which is mentioned in particular chapter. 2.2 Internal forces caused by mechanical load Internal forces on frame structure are calculated on the basis of structural mechanics principals. The object of the further analysis is the ceiling frame part. Extreme bending moment in frame corner 285

D

C Bar 4

Bar 2

Bar 1

A

Bar 3

B

Figure 1. Prefabricated wall-plate structure.

is MEd1 = 2.78 kNm for permanent design situation (PDS) and MEd1,a = 1.95 kNm for accidental design situation (ADS). Extreme bending moment in the mid-span is MEd2 = 2.26 kNm (PDS) and MEd2,a = 1.58 kNm (ADS). Coefficient for reduction of load effects is considered η = 0.7 according to Eurocode 1 (1996) [1]. 2.3 Load bearing capacity Prefabricated plate-wall structure is made of concrete C25/30 and the thickness is 80 mm. The ceiling is reinforced both at the lower and upper part of cross-section with bars R φ 5/150 mm, the concrete cover is at the lower part c = 15 mm and at the upper part c = 20 mm. Ultimate bending moment is for permanent design situation in the frame corner MRd1 = 3.14 kNm and in the midspan MRd2 = 3.14 kNm.

3 HEAT-EXPOSURE AND STRUCTURAL RESPONSE MODEL 3.1 Heat exposure One-dimensional transient temperature array is appointed using ANsys computer program, ANsys guide (2004) [3].The temperature in fire compartment in case of fire is considered according to standard time-temperature curve with respect to both convection and radiation according to Eurocode 1 (1996) [1]. For the purpose of simplicity and on the safe side are all the parts of frame structure exposed to standard heating conditions. However, it is obvious that the heat is divided in particular rate to the ceiling, bottom and walls of the structure. The lab test results of analogical structures have not been available yet. Mathematical modeling of real fire and space temperature distribution is the aim of further research. 3.2 Internal forces caused by temperatures Avoiding temperature deformation on indeterminate structure brings forth additional internal forces (1a), (1b), (Fig. 2).

286

Figure 2. Additional internal forces calculation. Table 1. Additional bending moment MT calculation, time = 5 minutes. Coordinate (m)

Temp. (◦ C)

l/l

Ei (GPa)

σTi (MPa)

0,00 0,05

229 157

0,001422 0,000876

29,79 30,18

42,37 26,44

0,75 0,80

20 20

0,000000 0,000000

30,50 30,50

0,000 0,000

Fi (kN)

zi (m)

MTi (kNm)

1

172,01

0,0375

6,451

15 16

0,002 0,002

0,0325 0,0375

0,000 0,000

MT

−13,440

Layer

Figure 3. Additional internal forces in ceiling frame bar.

Calculation example of bending moment caused by non-uniform cross-section warming is in Table 1, numerical integration is based on rectangular formula. Bending moment caused by nonuniform cross-section warming MT enlarges negative bending moment on ceiling frame bar. It is clear that bending moment MT exceeds ultimate bending moment MRd2 before time t = 5 minutes. Real value of bending moment MT is still the object of investigation and discussion. Part of tension causes probably internal stresses in consequence of rapid temperature gradient. 3.3 Modification of static model Providing that the bending moment caused by non-uniform cross-section warming MT exceeds ultimate bending moment in frame corner C, D MRd2 and providing that the shear load bearing capacity is ensured, the plastic hinges appear in the frame corner C, D, which bring forth the modification of static model and change internal forces in the structure (Fig. 4). Assuming that 287

Figure 4. Static model modification. Table 2. Reduction of ultimate bending moment in mid-span MRd1 . Time (min)

Tr (◦ C)

fyk (MPa)

MRd (kNm)

0 10 20 30

20 145 287 392

490 490 490 438

3,90 3,90 3,90 3,50

Table 3. Reduction of ultimate bending moment in frame corner MRd2 . Time (min)

Tconcrete (◦ C)

fck (MPa)

az (mm)

d (m)

Treinfoc. (◦ C)

fyk (MPa)

MRd2 (kNm)

0 5 10

20 229 392

25 23 20

0 2 3

0,058 0,056 0,055

20 20 26

490 490 490

3,58 3,45 3,37

the plastic hinges appear the bending moment caused by non-uniform cross-section warming MT vanish and the bending moment in mid-span of ceiling frame bar is the sum of primary bending moment in mid-span and the bending moment in frame corner (Fig. 4). 3.4 Reduction of ultimate bending moment Analogical formulas as for permanent design situation and published dependences of mechanical characteristics of concrete and reinforcing steel on temperature are used. The decisive factor for reduction of ultimate bending moment in the mid-span MRd1 is the growth of temperature in the reinforcement. The MRd1 reduction is sequenced in the Table 2. The decisive factor for reduction of ultimate bending moment in the frame corner MRd2 is the growth of temperature in stressed concrete, or rather the depth of damaged layer az , according to Eurocode 2 (1998) [2], example of az calculation is also in Cajka (2004) [4]. The MRd2 reduction is sequenced in the Table 3. Providing that the plastic hinges appear, the decisive factor for moment carrying capacity is the reduction of ultimate moment in the mid-span. The fire resistance is 30 minutes. 4 CONCLUSION In the paper the fire resistance of prefabricated plate-wall indeterminate structure is analysed. On the safe side all the parts of frame structure are exposed to standard heating conditions. However, 288

it is obvious that the heat is divided in particular rate to the ceiling, bottom and walls of the structure. Mathematical modeling of real fire and space temperature distribution is the aim of further research. Cross-sectional warming in case of fire causes not only reduction of ultimate bending moment but also additional internal forces. Additional forces outreach already in early stage of fire ultimate bending moment. The object of discussion is the value of additional bending moment regarding the character of temperature and stress distribution in the structure. Providing that the bending moment in frame corner exceeds load bearing capacity the plastic hinges appear, which cause the modification of static model and internal forces in the structure. The lab test results of analogical structures have not been available yet. Calculations of fire resistance are very progressive, because they bring time and financial savings for investors and producers of prefabricated concrete components.

REFERENCES [1] ENV 1991-2-2, Eurocode 1: Basis of design and actions on structures, part 2-2: Action on structures – Action on structures exposed to fire, CEN, Brussels, 1997 [2] ENV 1992-1-2, Eurocode 2: Design of concrete structures, part 1-2: General rules – Structural fire design, CEN, Brussels, 1998 [3] ANsys Classic 7.1 users guide, html on line documentation [4] Cajka, R., Zidkova, P.: Analysis of heat exposure and structural response model of tunnel lining in case of fire. In proceedings of conference Concrete in extreme conditions, Prague, Czech Republic 2004 (in Czech)

289

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Performance based design of steel frames F. Dinu Romanian Academy, Timisoara Branch

D. Dubina, D. Grecea & A. Stratan “Politehnica” University of Timisoara, Romania

ABSTRACT: The paper presents a parametrical study on the seismic performance of steel frames. Both moment resisting frames and eccentrically braced frames were selected for the study. Seismic performance is investigated using three performance criteria or limit states. Acceptance criteria for each limit state are given by partial q-factors, which are directly related to the level of damage in the structure. For study, a series of incremental dynamic analysis are carried out, using the Drain-3dx computer program.

1 INTRODUCTION Steel structures are designed to comply with the demands set up in the serviceability and ultimate limit states. For steel structures designed to resist the seismic loading, the design criteria have to be extended, due to the fact that, during strong earthquakes, yielding is allowed to occur so that part of the energy input may be dissipated through inelastic deformations. The extension refers to the requirement for provision of ductility, which correspond to the capacity of certain parts of the structure to undergo large inelastic deformations under cyclic loading conditions, without considerable reduction in stiffness and strength. The criteria in respect to stiffness, strength and ductility may be complementary in design. It is therefore necessary to examine if one of these criteria is prevailing, under what conditions and to what extend (Vayas & Dinu, 2000). Actual seismic codes are based on force-controlled design, using the base shear approach. The parameter related to the degradation of the structure is the q-factor, based on the maximum capacity of the structure to dissipate energy in the ultimate limit state (ULS). The ULS corresponding ductility cannot be attained if higher levels of performance (i.e. less damage to the structure) are required. In this case, a reduced ductility corresponding to a partial q-factor characterises the behaviour of the structure. As an alternative to this approach, present methodology proposes a “three objectives” performance based design, with acceptance criteria for each limit state given by partial q-factors. This approach allows the designer to predict, with reasonable accuracy, the performance (level of damage) of a building for a specified level of ground motion intensity. Traditionally, two structural typologies have been used for multi-storey buildings in seismic regions: concentrically braced frames (CBF) and moment resisting frames (MRF). Unfortunately, neither of these traditional systems fulfils contemporary the requirements for the three limit states: serviceability, damageability and ultimate. The eccentrically braced frames (EBF) represent a more advanced typology, which provide a suitable compromise between the properties of the previous systems. In this paper, two structural typologies are investigated, namely MRF and EBF. For study, a series of incremental non-linear time-history analysis are carried out, using the Drain-3dx computer program. Finally, partial q-factor values for each limit state are proposed. 291

2 PARTIAL Q-FACTORS FOR PERFORMANCE BASED DESIGN OF STEEL FRAMES As mentioned before, steel structures have to comply with several requirements and associated design criteria such as stiffness, strength and ductility. For many years, the main goal of seismic design was the protection of people against serious injury or loss of life and the avoidance of building collapse under a major earthquake. The second goal was the reduction of the damage level in the buildings. After each earthquake, more attention was paid on the structure performance than on the behaviour of non-structural elements, which can produce more economic losses than the structural damages. This observation shown that to only fulfil the condition of life protection is economically unacceptable (Gioncu & Mazzolani, 2002). As an alternative, the methodology developed by the authors proposes a “three objectives” performance based design. In this methodology, the acceptance criteria for each limit state are given by partial q-factors. Methodology requires the following steps: 1. 2. 3. 4.

definition of performance levels definition of seismic hazard selection of analysis procedure determination of global ductility

1. Definition of performance levels (limit states). Structures designed against earthquakes have to comply with specific criteria such as stiffness, strength and ductility. In the present methodology, three limit states, which are referring to conditions of drift, residual drift and rotation capacity of elements, are introduced: – Serviceability limit state SLS (stiffness criterion): under frequent earthquakes, the building can be used without interruption and the structure remains in elastic range. Acceptance criteria – interstorey drift limit (depends on the structural typology). – Damageability limit state DLS (strength criterion): under rare earthquakes, the building presents important damages of non-structural elements and moderate damages of structural elements which may be repaired without high costs or technical difficulties. Structure is responding in elasto-plastic range and the determinant criterion is the resistance of member section. Acceptance criteria – permanent interstorey drift limit (depends on the structural typology). – Ultimate limit state ULS (ductility criterion) – under very rare earthquakes, the building presents major damages of both non-structural and structural elements but safety of people is guarantied. Damages are extended so that structure cannot be repaired and demolition is unavoidable. Structure is in the elasto-plastic range and the determinant criterion is the local ductility (rotation capacity of elements and connections). Acceptance criteria – plastic rotation capacity (depends on the structural typology). Most of the steel structures may belong to one of the basic structural typologies mentioned in the introduction, such as moment-resisting frames (MRF) or eccentrically braced frames (EBF). These structural typologies rely on different combinations of strength, stiffness and ductility and therefore different characteristic values are used for each typology (Table 1, Table 2). 2. Definition of seismic hazard. In the design process, performance levels must be translated into seismic actions, represented by magnitudes or accelerations. The level of ground motion acceleration for the three performance levels (limit states) introduced in the first step may be determined as a function of the return periods. For DLS and ULS, respectively, there are no contradictions concerning the return periods (475 and 970 years, respectively). Contrary to this, for SLS there are different proposals (ranging from 10 to 75 years), due to the difficulties in choosing a rational criterion for non-damage limit states. If the acceleration for DLS – ad is considered as a basic value for ground motion acceleration, the accelerations for SLS and ULS are determined with the equation (Gioncu & Mazzolani, 2002)(Figure 1):

292

Table 1. Characteristic values associated to the performance levels for MR frames. Performance level/limit state

Limit drift [%]

SLS DLS ULS

Limit residual drift [%]

Plastic rotations in joints [rad]

0,6 1,0 0,03*

* Plastic rotation capacity obtained from experimental tests.

Table 2. Characteristic values associated to the performance levels for EB frames. Performance level/limit state

Limit drift [%]

SLS DLS ULS

Limit residual drift [%]

Plastic rotations in link [rad]

0,6 0,5 0,1*

* Plastic rotation capacity obtained from experimental tests. a ad

475 (DLS)

970 (ULS)

0,5

20 (SLS)

1,0

500

1000

pr (years)

Figure 1. Characteristic of ground motion: acceleration vs. return period.

based on ATC 40 (1997) proposal. For the SLS and ULS the corresponding accelerations are:

3. Selection of analysis procedure. Seismic design in current seismic design provisions permits the use of several static and dynamic analysis procedures: the equivalent lateral force, modal analysis, capacity spectrum and nonlinear dynamic analysis. Among all these methods, the use of nonlinear dynamic time-history procedures has become increasingly popular in recent years due to the expanding availability of faster computers. Incremental Dynamic Analysis (IDA) is a parametric analysis method that has recently emerged in several different forms to estimate more thoroughly structural performance under seismic loads. It involves subjecting a structural model to several ground motion records, each scaled to multiple levels of intensity, thus producing curves of response parameterized versus intensity level (Vamvatsikos & Cornell, 2002). In the present approach, accelerations corresponding to the development of first plastic hinge and accelerations corresponding to each limit states are determined by performing IDAs in order to find out the corresponding q-factors. Structural response is strongly influenced by 293

Table 3. Design concepts, behaviour factors and structural ductility classes Eurocode 8 (1994). Design concept

Behaviour factor

Required ductility class

Concept (b)

Low dissipative structure

1,5–2

L (Low)

Concept (a)

Dissipative structure Dissipative structure

1,5 < q < 4 q≥4

M (Medium) H (High)

the characteristics of the ground motion, therefore several ground motion accelerograms are employed. 3. Determination of global ductility. The most suitable approach for seismic design based on performance is the deformation-controlled design while today codes are based on a force-controlled design, using the base shear concept (see Table 3). In the latter approach, the most important parameter is the behaviour q-factor, based on the maximum capacity of structure to dissipate energy during the plastic deformations corresponding to ULS. The ductility corresponding to ULS cannot be attained if higher levels of performance are required. In that case, a reduced ductility corresponding to a partial q-factor is attained by the structure. The use of partial q-factor gives the possibility to implement the multiple performance design in the actual code methodology. To determine the q-factor, the following equation is used:

where: λu – acceleration multiplier for a limit state; λe – acceleration multiplier for first yielding. 3 PARAMETRIC STUDY 3.1 Design of structures The methodology presented in the previous paragraph is applied on trial frames to determine the partial q-factor values. Both MRF and EBF structures will be considered in the analysis. Each frame is designed according to actual seismic codes, to fulfil the requirements set up in the serviceability and ultimate limit states. 3.1.1 Design of MRF structures The geometric and other properties of the frames under investigation are presented in Figure 2. The structures were designed according to Romanian seismic code P100/92, using the following seismic design parameters: 0,25 g peak ground acceleration, soil condition characterised by Tc = 1,5 sec and the behaviour factor q = 5,9. The steel grade for all the frames is S235. 3.1.2 Design of EBF structure The geometric and other properties of the structure are presented in Figure 3. In practical applications, standard EB frames (EBF) are seldom used alone but as hybrid framing systems. This is done by a combination of a standard EB frame and pinned beam to column joints in the outer bays. The structure was designed according to Eurocode 3 (1993) and Eurocode 8 (1994). Dimensions of the structural sections resulted from design are presented in Figure 3. 3.2 Selection of accelerograms For the non-linear incremental dynamic analysis, the general purpose DRAIN-3DX software package was used. As was previously mentioned, incremental dynamic analysis is sensitive to the ground 294

3 3 3 3 3 3

3 3

5

5

5

5

5

5

5

5

5 3

2

1 Frame type

L(m)

H(m)

Beams

Columns

Structure period (sec)

Frame 1 Frame 2 Frame 3

5 5 5

3 3 3

IPE330 IPE330 IPE360

HEB220 HEB320 HEB400

0,45 0,70 0,88

3,5

Figure 2. Geometric properties of the MR frames (dimensions in m).

0,4

Links & beams middle span

Beams outer spans

Columns

IPE 240 S235

IPE 330 S355

IPE 260 S355

Structure period (sec)

3,5

Structure

0,64

4

3,5

3,5

EBF

5

5

5

Figure 3. EBF configurations (dimensions in m).

motion characteristics, and therefore, two different sets of ground motion records were used (see Figure 4). The first group (A) was composed of two “standard” records, with many acceleration pulses of similar magnitude and the acceleration response spectrum similar to the EC8 spectrum for class A soil (characteristic period in the short period range 0,3–0,5 sec): – Kobe earthquake, January 17, 1995, NS component, Kobe JMA record. – Northridge earthquake, January 17, 1994, 360 deg component, Sylmar Conv. Station record. The second group (B) of earthquake motions was composed of three records with a long acceleration pulse and the characteristic period in the long period range (≈1,5 sec): – Vrancea earthquake, March 4 1977, NS component, Bucharest INCERC record. – Montenegro earthquake, April 9 1979, EW component, Ulcinj – Hotel Olimpic record. Ground motions will be scaled so that the mean spectral acceleration will match the corresponding design mean spectral acceleration. This scaling procedure ensured initial forces approximately equal to the design ones, and roughly the same seismic input into the structures. 295

0,8

group A

2,5 Sa [g]

0,7 0,6

KOBE 1995 NS Northridge 1994 360deg

2,0

Sa [g]

3,0

1,5 1,0 0,5 0,0 0,0

group B Vrancea 1977 NS Montenegro 1979 EW

0,5 0,4 0,3 0,2 0,1

1,0

2,0 T [sec]

3,0

4,0

0,0 0,0

1,0

2,0 T [sec]

3,0

4,0

Figure 4. Response spectra of the ground motions.

4 NONLINEAR ANALYSIS The frames were subjected to the seismic records belonging to group A and B. In order to study the seismic performance of the two frame typologies, the peak ground accelerations corresponding to the attainment of each limit state (serviceability, damageability and ultimate) were determined by appropriate scaling (as , ad and au ). It was also recorded the acceleration corresponding to the attainment of the first plastic hinge in the structure. For each limit state, the peak ground acceleration was divided by corresponding acceleration levels (as , ad and au ). The acceleration multipliers α obtained in this way represent the safety margin for each limit state. q-factors are then determined for each limit state, using eq. 3. According to the definition of the damage state associated to serviceability limit state, a value of 1,0 for the q-factor is imposed, considering the structure is in elastic range. 4.1 MRF structures In Table 4 the values of the acceleration multipliers for MRFs are presented. It may be observed that the performance levels are well-balanced (similar values of acceleration multipliers) for DLS and ULS but much lower values are recorded for the SLS (see Figure 5). The governing criterion is the interstorey drift (corresponding to the SLS), with a value of acceleration closed to the design one. This confirms MRFs are designed according to the serviceability requirements, while the strength and ductility reserves remain high. It may also be observed a strong influence of the soil condition on the values of limit accelerations. This scatter may be a result of scaling procedure, also. Recent research has demonstrated that scaling methods may result in an excessive scatter in the estimated seismic demands, compromising the accuracy of nonlinear dynamic analysis procedures (Shome & Cornell, 1998). In Table 5 the values of the q-factor are presented. To assess the influence of ground motion input, seismic performance of structures was investigated using the A and B sets of ground motions. Values of q-factor for group A (characteristic period in the short period range 0,3–0,5 sec) are almost two times higher than the values of q-factor for group B (characteristic period in the long period range ≈ 1,5 sec). Therefore, besides structural configuration, soil condition may have a great influence on the q-factor values. For group B, the value of q factor prescribed by the code (q = 6 for the ULS) is on the unsafe side. 4.2 EBF structures In Table 6 the values of the acceleration multipliers for MRFs are presented. The performance levels are well-balanced (similar values of acceleration multipliers) and the governing criterion is the local ductility in the link (ULS). This is in line with the real behaviour of such structures, characterised by a higher stiffness compared to MRFs, and the concentration of ductility demand 296

Table 4. Limit acceleration multiplier α. Limit state

Group A

Group B

Mean value

SLS DLS ULS

1,1 2,3 2,6

1,2 1,5 1,5

1,15 1,9 2,1

Acceleration multiplier α

5,0 4,0

group A group B

3,0 2,0 1,0 Unacceptable Unacceptableperformance performance 0,0 SLS

DLS Limit state

ULS

Figure 5. Limit acceleration multiplier α.

Table 5. q-factor values for MR frames. Limit state

Group A

Group B

Mean value

SLS DLS ULS

1,0 5,1 7,1

1,0 3,0 3,8

1,0 4,0 5,4

8,0 7,0

q factor

6,0

group A group B

5,0 4,0 3,0 2,0 1,0 0,0

SLS

DLS Limit state

ULS

Figure 6. q-factors for the three limit states vs. ground motion input (MR frames).

297

Table 6. q-factor values for MR frames. Limit state

Group A

Group B

Mean value

SLS DLS ULS

2,15 2,10 2,0

1,45 0,8 1,0

1,8 1,45 1,5

Acceleration multiplier α

5.0 4.0 3.0

group A group B mean values

2.0 1.0 Unacceptable performance 0.0 SLS

DLS Limit state

ULS

Figure 7. Limit acceleration multipliers α. Table 7. q-factor values for EB frames. Limit state

Group A

Group B

Mean value

SLS DLS ULS

1,0 8,0 9,4

1,0 3,6 4,5

1,0 5,8 7,0

10.0

q factor

8.0

group A group B

6.0 4.0 2.0 0.0

SLS

DLS Limit state

ULS

Figure 8. q-factors for the three limit states vs. ground motion input (EB frames).

in the link, only. An excessive scatter in the estimated seismic demands is also recorded for the two sets of accelerograms (see Figure 7). Table 7 shows the values of the q-factor for EB frames. To assess the influence of ground motion input, seismic performance of structures was investigated using the A and B sets of ground motions. 298

Values of q-factor for group A (characteristic period in the short period range 0,3 – 0,5 sec) are two times higher than the values of q-factor for group B (characteristic period in the long period range ≈1,5 sec). The value of q-factor prescribed by the code (q = 5 for the ULS) is on the safe side. 5 CONCLUSIONS A performance-based methodology for seismic performance evaluation of steel moment frames and eccentrically braced frames is presented. Acceptance criteria for each performance level (limit state) are given by partial q-factors. A parametric study using an incremental dynamic analysis procedure was performed using two sets of ground motion records. Different values of limit accelerations and q-factor for the two ground motion sets were recorded. This scatter shows a strong influence of the ground motion type. It may be a result of scaling procedure, also. Recent research has demonstrated that scaling methods may result in an excessive scatter in the estimated seismic demands. Results showed a very good performance of eccentrically braced frames, with q-factors much higher than those given by the codes. REFERENCES ATC 1997. Seismic evaluation and retrofit of concrete buildings. ATC-40, Applied Technology Council, Redwood City, Vol. 1. Eurocode 3 Part 1.1 1993. Design of steel structures. Brussels: CEN, European Committee for Standardisation. Eurocode 8 1994. Design provisions for earthquake resistance of structures. Brussels: CEN. Gioncu, V., Mazzolani, F.M. 2002. Ductility of Seismic-Resistant Steel Structures. London: SPON PRESS. P100-92 1992. Romanian Code for the seismic design of residential, social cultural, agricultural and industrial buildings. Shome, N., Cornell, C.A. 1998. Normalization and Scaling Accelerograms for Nonlinear Structural Analysis. Proc. of the 6th U.S. National Conference on Earthquake Engineering, Seattle, Earthquake Engineering Research Institute, Oakland (CD-ROM). Stratan, A., Dubina, D., Dinu, F. 2003. Control of global performance of seismic resistant EBF with removable link; Proc. intern. conf. STESSA 2003 – Behaviour of steel structures in seismic areas, Napoli, 9–12 June 2003. Vayas, I., Dinu, F. 2000. Evaluation of the seismic response of steel frames in respect to various performances, Proc. intern. conf. STESSA 2000 – Behaviour of steel structures in seismic areas, Montreal, 21–24 August 2000. Vamvatsikos, D., Cornell, C.A. 2002. The Incremental Dynamic Analysis and its application to PerformanceBased Earthquake Engineering. Proc. of the 12th European Conference on Earthquake Engineering, London, September 2002.

299

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Flexural cyclic behaviour and low-cycle fatigue of cold-formed steel members B. Calderoni, A. Formisano & A. De Martino University of Naples “Federico II”, Napoli, Italy

ABSTRACT: The use of cold-formed thin-walled steel profiles in seismic structures is nowaday strongly penalized by code prescriptions, which impose for this kind of members a q-factor equal to 1, so that the design for the action of a severe earthquake must be carried out practically in the elastic field. On the other hand, some theoretical studies on the seismic performance of such structural systems gave encouraging results. For these reasons, a new experimental and theoretical research program on the behaviour of cold-formed members loaded by cyclic loads has been started by the authors. In this paper a brief presentation of the results obtained from tests in bending, under monotonic and cyclic load, on a number of unlipped channel beams, is given, a theoretical simulation of the flexural cyclic degrading behaviour is assessed and the collapse behaviour with reference to low-cycle fatigue resistance is analysed too.

1 INTRODUCTION Steel structures, built up by light steel gauge members, are nowadays practically not used for seismic application, despite their not doubtable advantages as lightness, low-cost and possibility of obtaining several different shapes. In fact cold-form steel members are strongly affected by local buckling of compressed zone, which gives rise to a degrading monotonic and cyclic behaviour, both for stiffness and strength. Then actual seismic codes strongly penalize application of light steel gauge members on the base of their lack in dissipation capacity, which doesn’t allow a favourable behaviour in case of seismic event. Furthermore connections between members are not well codified and their behaviour under cyclic actions is not deeply analysed. For these reasons it seems of strong interest to deepen the behaviour of such kind of structures in order to demonstrate the possibility to use them in an economic and safe seismic design, by applying a reduction factor of elastic seismic action greater than one. A fundamental question for defining the seismic capacity of cold-formed steel structures is to establish failure criteria for members and connections, taking into account the degrading behaviour and the low-cycle plastic fatigue. Nevertheless, while the elastic behaviour of this type of structure has been deeply analysed, the post elastic behaviour of the members, the technology and the elastic and post elastic behaviour of connection deserve more studies. In the recent years, the scientific community was strongly concerned about establishing the allowable ductility of connections, members and structures, in order to estimate the post-elastic capacity of them, useful during a strong earth motion. Recent researches, based on simplified assumptions of the cyclic degrading behaviour as resulting from experimental data given by some authors, allowed the modelling of a simple portal frame behaviour under the action of several earthquakes, and showed the possibility of using light gauge member frames in low seismic areas with q-factor varying from 1 to 3 (Calderoni et al. 1997, Calderoni & De Martino 2001). In this paper, on the base of these studies, the behaviour of thin gauge members is deeply analysed by means of experimental testing on simple beams, taking into account the cyclic nature of the action and the degrading phenomena in terms of strength, stiffness and rotational capacity. Elaboration of 301

Figure 1. Specimen ready for testing.

Figure 2. Geometrical scheme of specimens.

the obtained results are made with reference the degradation in post-buckling field, in order to asses a theoretical law of the cyclic behaviour and the corresponding low-cycle fatigue resistance line. 2 THE EXPERIMENTAL TESTS The experimental investigation on light gauge steel members has been performed on five identical simply supported beams, made of double back-to-back coupled cold-formed unlipped channel profiles and loaded with a single mid-span concentrated force (three-point bending tests). A global view of a specimen is given in Fig. 1, while its scheme is depicted in Fig. 2. Considering the single channel profile, the geometrical slenderness (b/t) (wide to thickness ratio) for the flange and web elements are 10.5 and 61.7 respectively. Consequently, according to the Eurocode 3, the profile may be classified in class 4 for the flange and class 2 for the web, so that the whole section has to be classified as slender section (class 4). With reference to the material, the yield and ultimate stresses of the used steel were directly obtained from six tensile tests specifically performed. The resulting average yield stress was fy = 364 MPa, while the ultimate one was fu = 425 MPa. Two monotonic (T01, T02) and three cyclic (T03, T04, T05) tests have been carried out. All tests were carried out in displacement control. The vertical displacement at mid length, the horizontal displacement of the simple support and the rotation of the edge hinge have been acquired during the test by means of displacement transducers (LVDT). Deformation of the flanges in different transversal sections have also been measured by using strain gauges. In the monotonic test T01 the displacement has been progressively increased up to the complete collapse reached at 120 mm, while in test T02 the maximum displacement has been fixed at 20 mm, in order to investigate also the response in the unloading phase immediately after the reaching of the local buckling. Cyclic tests has been carried out using a fixed constant displacement amplitude. The number of cycles has not been initially defined: the test was ended when the failure occurred. The displacement amplitudes were 15 mm, 30 mm and 45 mm for test T03, T04 and T05, respectively. More detailed information about the beams, the loading equipment, the measurement instrumentations and the loading histories are given in (Calderoni et al. 2004, Formisano 2003). 3 TEST RESULTS The structural responses exhibited in the monotonic tests (T01 and T02) are drawn in Fig. 3 in term of acting force (F) and applied deflection (δ) at mid-span. Viewing to these F–δ curves, it can be 302

80

75 kN

80

T01

75 kN T02

70

60

60

50

50

F (kN)

F (kN)

70

40

40

30

30

20

20

10

10

0

0 0

10

20

30 40 50 60 70 80 90 100 110 120 130 δ (mm)

0

10

20

30 40 50 60 70

80 90 100 110 120 130

δ (mm)

(a) Test T01

(b) Test T02

Figure 3. Force (F)–displacement (δ) curves for monotonic tests.

80 60

F (kN)

T03

40 20 0 -20 -40 -60 -80 -50

Figure 4. Local buckling in test T01.

-40

-30

-20

-10

0

10

20

30

δ (mm) 40 50

Figure 5. F–δ curve for test T03.

observed that both the beams exhibited a quite linear behaviour nearly up to the attainment of the maximum force withstood by the beams. In the test T01 the little loss of response linearity just before the maximum load was due to the development of the first yielding in the section, which theoretically should correspond to an external force equal to 75 kN, as indicated in the graphs by the horizontal solid line. A little bit after, the critical load due to local buckling was reached, involving the right bottom compressed flange for both the specimens, as shown in Fig. 4. The values of the maximum loads recorded during the two tests were Fmax = 77.6 kN and Fmax = 72.4 kN, respectively for T01 and T02. The corresponding displacements were 16.2 mm in T01 and 11.7 mm in T02. Once these displacements were exceeded, the specimens experienced an unstable decreasing behaviour. Test T01 was leaded up to the collapse of the beam. The collapse displacement was considered 80 mm. In fact, at this value of deflection, the bottom buckled flanges of the profiles went in contact with the central stiffening plates of the specimen. For this reason, further increasing of the deflection led to fictitious growing of the reaction force and of the stiffness of the beam. The cyclic tests (T03, T04 and T05) presented a behaviour characterized by progressive deterioration of strength and stiffness, typical of steel members when affected by local buckling. The cyclic responses for the beams T03, T04 and T05 has drawn as F−δ curves in the Figs. 5, 6 and 7, respectively. In all these tests, local buckling started in right bottom flanges of both profiles. During the test, as the number of cycles (N ) increased, cracks always arose in the buckled zone of the flanges, at the flange-to-web connection. These cracks gradually developed toward the free edge of the flange. When the whole length of the flange was involved by the crack, the beam lost quite at all its load bearing capacity and the test has been ended. The phases corresponding to the crack beginning and to its full development are shown in Fig. 8. With reference to beam T03 (displacement amplitude  = 15 mm), a maximum force (Fmax ) equal to 72.3 kN was recorded for a deflection of 13.1 mm in the first cycle, when local buckling was already developed. The strength and stiffness degradation was significant in the first four cycles. First cracks arose at the 15th cycle and involved the whole flanges at the 47th cycle. 303

80

80

40

40

20

20 0

0

-20

-20

-40

-40 -60

-60 -80 -50

T05

60 F (kN)

T04

60 F (kN)

δ (mm) -40

-30

-20

-10

0

10

20

30

40

50

Figure 6. F–δ curve for test T04.

-80 -50

-40

-30

-20

-10

0

10

20

30

δ (mm) 40 50

Figure 7. F–δ curve for test T05.

40 35

Mlim

25 20 15 10

Moment (kN m)

30

5 0 0,00

Figure 8. Full development of the cracks.

Plastic rotation(rad) 0,01

0,02

0,03

0,04

T01 0,05

0,06

Figure 9. Plastic rotation of mid span section.

For beam T04 ( = 30 mm), the maximum force reached resulted 76.7 kN, at a deflection of 14.5 mm. Also in this test the first three-four cycles were the most important for beam degradation. The cracks started at the 7th cycles and reached the free edges of flanges in the 21st cycle. For beam T05 ( = 45 mm) the maximum load bearing capacity was 72.9 kN at a deflection of 13.5 mm. The strength and stiffness deterioration was very important in the first two cycles. The crack started during the 3rd cycle and was fully developed in the 6th cycle. It is worth to notice that all the performed tests (monotonic and cyclic) have shown a quite good uniformity in terms of global behaviour, load bearing capacity and displacement corresponding to the maximum load, making us confident in the reliability and the correctness of testing. All the tested beams experienced the local buckling phenomena, which happened for a value of the applied load practically coincident with the one producing the first yielding in the mid-span section, so confirming that the adopted profiles are on the boundary between the class 3 and class 4 of the EC3 classification. This means that, even if the yielding stress has been reached in the material, the beam is not able to exhibit a fully plastic behaviour connected to the development of a plastic hinge: local buckling in the compressed flange lead to a decrease of load bearing capacity when displacement increases up to the failure. Nevertheless the decreasing branch of the obtained F–δ curve (in the monotonic test) shows a not negligible ductility of the tested beam (more than 4 in displacement). Likewise is meaningful the rotation capacity of the mid-span section, reported in Fig. 9 as moment vs. plastic rotation curve for test T01. Looking at the figure it can be observed that the maximum plastic rotation at the failure is more than 5.5% rad, value analogous to those obtained for compact sections, even though the loss of bearing capacity is more than 50%. Note that the mid-span section of the tested beam corresponds (due to the adopted scheme) to a fully restrained base section of a cantilever beam and that the extension of the plastic (or, better, the buckled) hinge is equal to about half of the section height, while for a compact profile this extension is generally equal to the section height. This means that the same plastic rotation corresponds in a light gauge beams to a double value 304

Ei/E1 1,1

T04

1 0,9 0,8 0,7 0,6 0,5 0,4 0,3 0,2 0,1

21

19

17

15

13

9

11

7

5

3

1

0

N

Figure 10. Ratio Ei /E1 vs. cycle number (N ) for T04 beam.

of the section curvature and than produces larger local specific deformation, further increased by the local buckling deformation phenomena, so giving reason for the progressive crack in flange material observed during the tests. As far as the energy dissipation capacity is concerned, the cyclic tests have shown that the energy dissipated in each cycle decreases as increasing the number of cycles (N ) and that the more significant loss of dissipation capacity occurs in the initial two or three cycles, depending on the amplitude of the performed cycles. As an example, in Fig. 10 the ratio energy dissipated during the generic cycle (Ei ) above energy dissipated during the first cycle (E1 ), is drawn in function of the number of cycles with reference to test T04. It can be observed that the increasing of the displacement amplitude applied in the test produces a more significant reduction of structural dissipation as the number of cycles grows. Furthermore the dissipated energy becomes practically constant (equally to about 20% of that dissipated in the first cycle), when the number of cycles becomes high and the beam is close to the failure condition. Anyway the “shape” of each cycle in the response F–δ diagram can be considered satisfactory, due to the absence of negative phenomena as, for instance, pinching, independently of the amplitude of the cycle.

4 ASSESSMENT OF CYCLIC BEHAVIOUR In order to perform numerical simulation of the response of light gauge steel structures under seismic actions, it is necessary to assess the behaviour of the members in theoretical way. The monotonic F–δ curve must be described numerically on the base of few parameters obtained from the experimental results. In addition, simple rules must be defined for taking into account stiffness and strength cyclic degradation. At this aim the experimental F–δ curve has been represented by means of a simplified multilinear function, having two increasing branches (corresponding to the initial elastic behaviour up to the maximum bearing capacity), two decreasing branches (corresponding to the post buckling behaviour) and a final horizontal constant branch representing the residual bearing capacity. In Fig. 11 this simplified representation of the monotonic experimental behaviour (for test T01) is drawn, and the couples F–δ used for defining the curve are reported too. According to the results of cyclic tests, simple deterioration cyclic rules have also been defined, based on how much the applied deflection (δ) exceeds the displacement (δlim ) corresponding to the maximum bearable load (Flim ) in each cycle. In particular: • if δ is smaller than δlim both in positive or negative field, the reverse behaviour (unloading and loading in reverse field) remains elastic with the same stiffness of the loading branch, even if there is a residual deformation (in case of δ > δel ); • if δ is greater than the positive (or negative) value of δlim , in the following unloading condition the negative (or positive) value of Fel is reduced and the corresponding elastic displacement is increased; the ratios between Flim and Fel remain constant while the ratio between δlim and δel is 305

80

F (kN) B (Flim, δlim)

70 60 A (Fel, δel)

50

C (Fc,δc)

40 30

D (Fd, δd)

20 10 δ (mm) 0 0

10

20

30

40

50

60

70

80

90

100

Figure 11. Multi-linear representation of the experimental F–δ curve (test T01).

F (kN)

100

100

80

F (kN) 80

60

60

40

40

20 -40

-30

-20

-10

0 0 -20 -40

20 10

20

30

40

δ (mm)

-40

-80 -100

-20

-10

0 0 -20 -40

-60

a)

-30

Experimental Numerical

b)

-60 -80

10

20

30

40

δ (mm) Experimental Numerical

Figure 12. Comparison between experimental and theoretical F–δ diagram for test T03 (a) and T04 (b).

increased at each cycle. So we get a reduction both in strength and stiffness, as greater as the “plastic” excursion is, by means of a linear correlation. More in detail: – in the current i cycle:

where: F (i−1) is the value of the load reached in the previous (i – 1) cycle (i) (i) (i) (i) (i) (i) The new values Fc , Fd and δc , δd are calculated in the same way of Flim and δlim :

(i)

– only for the second cycle (i = 2), that is to say the first application of the procedure, Flim and (i) δlim are further increased by the factor I (i) :

Applying these rules and starting from the multi linear F–δ curve represented in Fig. 11 (obtained from the monotonic test T01), the experimental cyclic tests, T03 and T04, have been simulated. In Fig. 12 the so obtained theoretical F–δ cyclic responses are compared with the corresponding experimental ones. 306

It can be noted a quite satisfactory agreement between the theoretical simulation and the experimental results, even if some discrepancies in term of energy dissipation are still visible. Anyway more comparisons with experimental results, also for different type of beam sections, have to be done before to declare the possibility to use this model in dynamic analysis of structures made by light-gauge steel members.

5 CONSIDERATIONS ON LOW-CYCLE FATIGUE FOR COLD-FORMED MEMBERS On the base of the performed constant amplitude displacement cyclic tests, some preliminary considerations can be made also on the fatigue behaviour for a small number of cycles (low-cycle fatigue). In fact structural behaviour under seismic actions is usually characterised by a small number of cycles often involving significant excursions in the plastic range, so that many times a premature collapse of members or connections is related to a plastic fatigue failure. This kind of failure is achieved as results of the interaction of many different effects, as ductility, cyclic deterioration both in stiffness and in strength, local buckling, energy dissipation, development of cracks, and so on (Castiglioni 1999). It is very difficult to share the “responsibility” of collapse among each of these different effects, while it is more convenient to refer to a whole parameter, like the low-cycle fatigue endurance. It is now consolidated (Ballio & Castiglioni 1995) that, also in the case of low-cycle fatigue, the failure prediction function can be represented by means of the S–N curve approach (or Woehler law), which leads to the equation:

where N is the number of cycles to failure at the constant stress (or strain) range S. The nondimensional constant m and the dimensional parameter K depend on both the typology and the mechanical properties of the steel components. The above equation is usually used in the form:

which, in a Log–Log domain, represents the fatigue resistance line, having slope equal to −1/m. Differently by the traditional approach to fatigue, it is more convenient to adopt global quantities (e.g. displacements or internal forces) instead of local ones (like stress or strain) as parameter S, because they are more clearly connected to the global seismic behaviour of the structure and more easily predictable in the design phase. In this paper we refer to mid-span displacement (i.e. the amplitude of the cycles) as key parameter S to be used in the S–N relationship (Bernuzzi et al. 1997, Castiglioni et al. 1997). With reference to the number of cycles to failure (Nf ), it can be derived by adopting a specific failure criterion. In this paper three different criteria have been considered for determining Nf : (a) Nf is the number of the cycle in which the cracks have developed in the whole length of flanges; (b) Nf is the number of the cycle in which the reaction force became less than 20% of the maximum reached one (residual bearing capacity – RBC = 20%); (c) Nf is the number of the cycle in which the dissipated energy Ei (area enclosed below the curve in the plane F–δ) is equal to the 50% of that dissipated in the first cycle E1 . This criterion has been defined by Calado and other authors (Castiglioni & Calado 1996), who recommended to assume equal to 0.5 the parameter αf = Ecf /Eco at the collapse, where Ecf is the ratio between the absorbed energy at the last cycle before collapse above the energy that might be absorbed in the same cycle in case of elastic perfectly plastic behaviour, and Eco is the same ratio referred to the first cycle in the plastic range. 307

Log (S)

Log (S) 3,2

TEST

2,8

Nf (number of cycles) S [mm] Citerion (a)

T03 T04 T05

2,4 2

15 30 45

47 21 6

3,2 2,8 2,4 2

1,6

1,6

1,2

1,2

0,8

0,8

0,4

0,4

0

TEST

S [mm]

T03 T04 T05

15 30 45

Nf (number of cycles) Citerion (b)

Citerion (c) 17 5 3

18 6 3

0 0

0,4

0,8

1,2

1,6

2

2,4

2,8

3,2

Log(N) = 4.04 - 1.94 Log(S) a - Experimental results according to (a) criterion

3,6

4

4,4

Log (N)

0

0,4

0,8

1,2

1,6

2

2,4

2,8

3,2

3,6

4

4,4

Log (N) Log(N) = 3.24 - 1.69 Log(S) b - Experimental results according to (b) and (c) criterion

Figure 13. Evaluation of S–N fatigue resistance lines according to different failure criteria.

According to these criteria, the experimental tests have given the results given in the tables included in Fig. 13. The values of Nf corresponding to the (c) criterion has been obtained as shown in Fig. 10 for test T04, where the relationship αi versus N is drawn. It can be observed that the criteria (b) and (c) practically give the same results. In Fig. 13, in the Log(S)–Log(N ) diagram, the full dots represents the fatigue failure results obtained from tests applying both the criterion (a) (Fig. 13a) and the criterion (c) (or (b)) (Fig. 13b). On the same Figures the corresponding fatigue resistance lines, obtained as the best linear fitting of the experimental dots, is depicted. According to criterion (a) the line is characterized by a slope (−1/m) equal to −0.52 (i.e. m = 1.94) and a Log(K) = 4.04, while according to criterion (b) and (c) we obtain m = 1.69 and Log(K) = 3.24. It is worth to note that the slope of the lines in both cases is significantly higher than the ones given by Eurocode 3 (m = 3 or m = 5) for (high-cycle) fatigue design of normal (not cold formed or light gauge) steel structures details. Values of m ranging from 1.23 to 3.46 have been reported in (Calado 2000), where the results of several experimental cyclic tests, performed in Lisbon in the last years on fully welded and top and seat with web angle connections, have been elaborated with reference to low cycle plastic fatigue. Note that the tests reported in this paper are practically representative of fully welded connections of cold formed members, considered that the mid-span section of the beam can be considered as a fully restrained beam edge. The results obtained, while interpreting the tests from the point of view of fatigue behaviour, appear to be in line with expectations. In fact, whatever the adopted criterion for defining the failure, the slope of the fatigue resistance line which fit the experimental data have resulted ever higher than the ones corresponding to compact steel profile details: higher the slope, quicker is the degradation of the structure when the number of cycles increases. It means high sensibility of cold formed profiles to cyclic actions in plastic field, due to local buckling, which affects the compressed zones, giving rise in them to very concentrated yielding. This fact has to be considered with attention in evaluating the adoptable q-factor to be used in design of light gauge structures in seismic zones.

6 CONCLUSION The monotonic and cyclic tests, carried out on unlipped channel beams made with cold-formed light gauge sections, have confirmed progressive and significant loss of both load bearing capacity and flexural stiffness, after attainment of local buckling in the compressed flanges. 308

Nevertheless the experimental results have shown not negligible available ductility and energy dissipation capacity, which could be considered in a “not elastic” seismic design of light gauge steel frame, at now not allowed by the current seismic codes. On the other hand the progressive deterioration of the mechanical response affects significantly the low-cycle fatigue life of the members so that particular attention in defining the collapse condition under the earthquake actions has to be paid. At this aim theoretical law of the cyclic behaviour, which takes into account the progressive deterioration of the mechanical response by means of simple rules based on only few experimental parameters, has been defined and low-cycle fatigue resistance lines, characterized by higher slope than for the steel compact sections, has been obtained. Future developments of the research have already been defined: cyclic test with variable displacement amplitude will be carried out and the effects of axial load on the monotonic and cyclic behaviour will be also analysed. REFERENCES Ballio, G. & Castiglioni, C.A. 1995. A unified approach for the design of steel structures under low and high cycle fatigue. Journal of Constructional Steel Research, Vol. 34. Bernuzzi, C., Calado, L. & Castiglioni, C.A. 1997. Ductility and Load carrying Capacity Prediction of Steel Beam-to-Column Connections under Cyclic Reversal Loading. Journal of Earthquake Engineering, Vol. 1, no. 2, 401–432. Calado, L. 2000. Fatigue Resistance Design, In F.M. Mazzolani & V. Gioncu (eds), Seismic Resistant Steel Structures: Springer Wien New York, ch. 4, 381–399. Calderoni, B., De Martino, A., Ghersi, A. & Landolfo, R. 1997. Influence of local buckling on the global seismic performance of light gauge portal frames, F.M. Mazzolani & H. Akijama (eds), Proceedings of STESSA ’97, Kyoto, 29–40, Salerno: Edizioni 10/17. Calderoni, B. & De Martino, A. 2001. Sulla progettazione di strutture intelaiate in parete sottile in zona sismica, Proceedings of the XVI C.T.A. Congress, Venice. Castiglioni, C.A. 1999. Failure criteria and cumulative damage models for steel components under lowcycle fatigue (Criteri di collasso e modelli di danno per componenti strutturali in acciaio soggetti a fatica oligociclica), Proceedings of the C.T.A. Congress, Napoli. Castiglioni, C.A., Bernuzzi, C. & Agatino, M.R. 1997. Progettare a fatica oligociclica (Low-cycle fatigue: a design approach), Proceedings of the C.T.A. Congress, Ancona. Castiglioni, C.A. & Calado, L. 1996. Seismic damage assessment and failure criteria for steel members and connections, Proceedings of the International Conference on Advances in Steel Structures, Hong Kong, 1996. Formisano, A. 2003. Analisi teorico-sperimentale della fatica oligociclica di travi di acciaio in parete sottile, Graduation Thesis, University of Naples “Federico II”, Faculty of Engineering.

309

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Resistance and ductility of stainless steel bolted connections A. Bouchair LGC-CUST, Blaise Pascal University, Clermont-Ferrand, France

ABSTRACT: The non linear and ductile behaviour of stainless steel, with a high ratio of the ultimate to yield strength, guarantee an excellent global behaviour of the connection. The EC31.4 requirements for stainless steel connection design are close to that for carbon steel. However, some modifications were introduced to avoid excessive deformations at the SLS. In this paper, the experimental results for the cover-plate connections and bolts, are compared to the predictions of EC3. The tests show that the various ultimate state connection resistance are safely predicted. A presentation of the major differences between carbon and stainless steel is given. That concern the Ultimate and Serviceability Limit States criterion applied on bearing, net and gross sections. The stainless steel connection show a large deformation before failure. The study is now being complimented by the use of finite element modelling in order to extend the results to the design of other geometrical configurations and other types of connections.

1 INTRODUCTION The use of stainless steel structural elements is not very common but various new potential uses continue to be identified. In construction, stainless steel is commonly used for secondary elements. The structural design is covered by ENV 1993-1.4. The austenitic steel is the most widely used in construction. The stress–strain curve shows a non-linear and ductile behaviour with a high ratio of the ultimate to yield strengths. The large ductility will give some advantages for the energy dissipation in case of cyclic loading and a large capacity of loads redistribution before failure. It has an excellent resistance to corrosion, a good aesthetic aspect, a good ductility, a better resistance in fire (Mäkeläine & Outilane, 1998) and large possibilities of strain hardening. Its use in the train structures showed that it offers, in addition to the reduced life cost, a capacity of important deformations which allow to absorb an important part of the energy during a shock. Several recent studies of the mechanical behaviour of the stainless steel structural elements were conducted (Burgan et al., 2000; Kouhi et al., 2000; Van Den Berg, 2000). They seem to follow the evolution of the world production and consumption. The plate-to-plate bolted connections are easy to use and their load transmission is made by plates in bearing and bolts in shear. As part of an ECSC investigation, a study on stainless steel bolted cover-plate connections was carried out at LGC (Bouchair & Muzeau, 1999; Ryan, 2000; SCI, 2000). Plates in several types of stainless steel (austenitic, ferritic and duplex) connected by three bolt diameters, in different arrangements were examined. The results of the austenitic steel (304L), which represents the biggest market share in the field of the construction and is characterized by its high ductility, are presented. The experimental results are analysed to assess the resistance formulae for the cover-plate bolted connections and to obtain an estimation of the bearing deformation (Bouchair & Baptista, 2002). Attention could be given to the resistance capacity of the connection compared to that of the connected element. Thus, for bolted connections, a design for the ULS loads may not be sufficient to avoid unacceptable plastic deformations in the SLS. The available design codes are focused on structural carbon steel. The formulae and design approaches are be modified to take into account the non-linear behaviour of stainless steel. The bolts pretension is less used because of the low 311

Table 1. Mechanical characteristics of the austenitic steel used in tests. fy (0.2%)

fu

Thickness (mm)

Test (MPa)

Nominal (MPa)

Test (MPa)

Nominal (MPa)

Test

Nominal

5 10

271 288

200 200

577 581

520 520

2.13 2.02

2.6 2.6

fu /fy

friction coefficient between the plates and the difficulty to obtain a suitable pretension. However, an experimental study (Scot & Trautner, 1998) showed that the stainless steel fasteners can perform well in resisting slip. The application concerns at first the connections submitted to direct tension load but it could be used to other kinds of cover plate connections. Beam-to-beam connection could be a realistic example. In this case, the analysis could take into account the connection deformation at different level of loading to allow the assessment of the global behaviour of the structure.

2 STAINLESS STEEL MECHANICAL BEHAVIOUR The stainless steel stress–strain curve is different from that of carbon steel. Indeed, the carbon steel presents an elastic linear behaviour followed by a plastic deformation before entering in the field of strain-hardening, while the stainless steel has a curve without a well marked elastic limit and it is nonlinear even for low levels of load. A conventional limit at 0.2% of plastic strain is used. Also, this steel has a non symmetrical behaviour in tension and compression (ESDEP, 1997; EuroInox, 2002) and they have a general tendency to have a non linear behaviour more marked in tension than in compression. So, an anisotropy exists and it is implicitly recognized in the codes which recommend to proceed to the characterization on test coupons taken in the transverse direction of rolling. However, the aspect which has potentially most influence for structural applications is the non linearity. So, as for certain metal alloys, in opposition to carbon steels, stainless steels are subject to significant creep at ambient temperatures. This incites to use stainless steels in a controlled way if a high level of stress must be maintained during a long period. It is suggested to limit the stress due to the long-term actions to 60% of the conventional elastic limit (Euroinox, 1994). The materials characteristics used for the connections were defined in the transverse direction of rolling and the connection specimens were also taken in this direction. Material characteristics supplied by the manufacturers of steel are given in the Table 1. The tested steel (304L) of European designation (X2CrNi19-11 or 1.4306) is in the S220 resistance class, according to EC3-1.4. There is a high ratio between the ultimate and elastic limits (Table 1). The measured elongation is close to 63% while the design (nominal) one is equal to 45%. A one parameter Ramberg–Osgood’s relation is often used to represent the non-linear behaviour of the stainless steel. It could be improved for the part above the elastic limit (Rasmussen, 2003). However, if the n parameter is well chosen the relation can describe accurately the stress–strain relation (Fig. 1). This relation, used in EC3-part 1.4, can be written as follows:

This relation has three independent constants which define the shape of the stress–strain curve, where σ0.2 = elastic limit and n = a coefficient which gives the degree of the material strain hardening. The n value can be obtained experimentally or in EC3-part 1.4 (CEN, 1996; EuroInox, 1994). An example of the real stress–strain curve is given in Figure 1. 312

Figure 1. Example of the Ramberg–Osgood curve for the austenitic steel used in tests (304L).

3 DESIGN APPROACH Design methods are often based on those for carbon steel with the nominal characteristics of the stainless steel defined in standards with a particular consideration of the relatively important deformation. At present, European draft standard EC3-1.4 covers the use of austenitic or duplex stainless steel for structural applications. The proposed formulas cover also beams and columns. The part dedicated to connections concerns mainly cover-plate types. This could be explained by the fact that the most use of stainless steel is for thin elements. The few documents dealing with stainless steel connections mainly concern either welded or bolted cover-plate connections of thin elements (ASCE, 1990). All the types of bolted connections used in the carbon steel structures can be used with stainless steel. The design of carbon steel joints is usually made to the ULS, while the deformation criteria at the SLS is considered to be implicitly satisfied. This assumption is justified by the linear behaviour of carbon steel up to its elastic limit and by the low ratio between ultimate and yield strengths, which is generally comprised between 1.1 and 1.5. In the case of stainless steel, the stress–strain curve is non-linear and the ratio between the ultimate strength and the elastic limit may exceed 2. So, a design to the ULS may not guarantee that no excessive deformations will occur at the SLS, since the ratio between the ULS and SLS loads is usually comprised between 1.35 and 1.5. A limitation is introduced for the net section resistance to limit the deformation at the SLS. Another possible verification should concern the hole deformation in bearing, which is not easy to predict. However, the EC3 considers that the ULS verification is sufficient if a reduced bearing ultimate limit is used instead of the real ultimate limit. In this study, the analysis is based on the simple failure mechanisms proposed by the EC3 taking account of the ductile deformation of stainless steel. Test results show that the various ultimate state joint strengths are safely predicted with a very large ductility of all the connections under static loads. This will be a good guarantee of safety under cyclic or dynamic loads. Also, the evaluation of a beam deflections and a plastic loads redistribution need to be more analysed. For the members, the secant modulus could be used in substitution to the elastic modulus (Van Den Berg, 2000). Also, attention has to be made to ensure that calculated plastic moment transmitted to the connections do not exceed their own resistance. 3.1 Bolts in tension and shear (ULS) Stainless steel bolts and nuts are covered by the ISO 3506 standard according to their metallurgical structure. For the austenitic bolts, letter A is used with a number according to their resistance classes 313

to corrosion (A1, A2, A3, A4 and A5). The nuances A2 and A4 are the most used in the structural applications and they can be obtained in 3 classes of strength 50, 70 or 80 (tenth of the ultimate strength). A bolt A4-70 has good resistance to corrosion with an ultimate limit of 700 MPa. It will be noticed that the class 70 is the most common and the most available on the market for steel structures. 3.1.1 Bolts in shear The shear strength of carbon steel bolts, for one shear plane, depends on the position of the shear plane with regard to the threading and the ductility of bolts. The design resistance formulae given below, with a coefficient of 0,6, corresponds to ductile bolts (class: 4.6, 5.6 and 8.8) whatever is the position of the shear plane (A or As) or bolts less ductile (4.8, 5.8, 6.8 and 10.9) if the shear plane is situated in the smooth part of the shank (A). For these latter types of bolt, if the shear plane passes through the threaded part (As), it is necessary to replace 0.6 by 0.5. The bolt shear resistance is given by the following formulae: Fv.Rd = (0.5 or 0.6) · fub · As/γMb : Threaded plane Fv.Rd = 0.6 · fub · A/γMb : Non threaded plane For stainless steel bolts, the relations given above are used considering that their behaviour is close to that of ductile carbon steel bolts. Thus, only coefficient 0.6 is used independently of the shear plane position, in the threaded part or not of the shank (As or A). 3.1.2 Bolts in tension The design formula is similar to that for carbon steels. It can be written: Ft·Rd = 0.9 · fub · As /γMb . Resistance is therefore based on the ultimate strength of the bolt material. However, bolts with a class of resistance equal to 50 have a high ratio between the ultimate and elastic limits (about 2.4). This would load them beyond their elastic limit under serviceability loads. To avoid long term effects, it is suggested to use a reduced tensile strength of 1.9 fyb , where fyb is the conventional 0,2% elastic limit (EuroInox, 1994). 3.2 Cover-plate bolted connections (ULS) The design approach is based on that of EC3-1.1(6) for carbon steel connections (CEN, 1992). One of fundamental difference concerns net cross-section and bearing resistances. Safety factor to be used in the design formulae is γMb = 1.25. The proposed formulae are defined on the basis of various associated failures modes : net section, gross section and bearing. 3.2.1 Net cross-section resistance The net cross-section resistance in a plate with holes for bolts, in tension, is given by the following formula: Nu.Rd = 0.9 · kr · Anet · fu /γM 2 where: kr = (1 + 3 · r · (d0 /u − 0.3)), and kr ≤ 1 r = ratio of the number of bolts at the cross section to the total number of bolts in the connection; u = 2 · e2 , with u ≤ p2 ; Anet = the net cross-sectional area; d0 = the nominal diameter of the bolt hole; e2 = the edge distance from the center of the bolt hole to the adjacent edge, in the direction perpendicular to the direction of the load transfer; and p2 = the spacing between the bolt holes, in the direction perpendicular to the direction of load transfer. Reduction factor kr is the new element which does not appear in the formula proposed for carbon steels. It is based on a statistical evaluation of experimental results and used for bolted connections in thin elements covered by EC3-1.3 (CEN, 1996; EuroInox, 1994). It is supposed to cover the stress concentrations and the excentricity due to the bolt load transfer to the hole, in comparison to the case of the non loaded hole in a plate in tension. So, it is also based on connections with only a single shear plane for which the tension load transferred to the bolt affects the tension resistance of the net section and in particular for the thin elements. For carbon steels, the reduction factors kr and 0.9 are used in a exclusive way (kr for thin elements and 0.9 for the others). For stainless steel, these coefficients are used simultaneously but the coefficient 0.9 is suppressed in the new EuroInox document (EuroInox, 2002) because of the 314

ductile nature of stainless steel. American standard (ASCE, 1990), proposes a similar expression for the coefficient kr but making difference between connections with one or two shear planes. Also, it proposes a limit criterion on stress to limit the deformation at the net section. 3.2.2 Gross section resistance Design formulae are similar to those for carbon steel and the design check consists in limiting the stress in the section to its elastic limit. The plastic resistance of the gross-section is given by: Npl,Rd = A · fy /γM 0 3.2.3 Bearing resistance Bearing resistance is given by a similar formula as used for carbon steel. It may be written as following: Fb.Rd = 2.5 · α · fu · d · t/γMb where: α = min{e1 /(3do ); p1 /(3do −1/4); fub /fu ; 1.0} According to EC3-1.4, this resistance verification is not sufficient because of the bearing resistance could be governed, for stainless steel, by the need to limit the ovalisation of the hole under serviceability loads. This ovalisation is not easy to calculate. However, check at the SLS can be avoided if the real ultimate limit strength fu is replaced by a reduced one fur given by the combination of the elastic and the ultimate limits as follows: fur = 0.5 · fy + 0.6 · fu . 3.3 Serviceability limit state (SLS) A new verification, in the net cross-section and in the bolts at the SLS, is introduced. The stress there should not exceed the elastic limit under serviceability loads. The aim is to limit SLS deformations. Another possible verification would concern the hole deformation in bearing, but it is difficult to find a predictical approach for doing this. However, the Eurocode allows not making this check at the SLS, but at the ULS, if a reduced bearing ultimate limit is used instead of the real ultimate limit (CEN, 1996b). 4 TESTS CONDUCTED AT LGC The experimental study of cover-plate connections, was conducted at the Civil Engineering Laboratory in Clermont-Ferrand and the results were compared to the predictions of EC3-1.4 (Bouchair et al., 2001). One of the principal objectives of the experimental study was to verify, and propose improvements to the EC3-1.4. The geometrical configurations of connections tested at the LGC Laboratory were chosen so as to cover all the failure modes foreseen for cover-plate connections. In the comparison between the EC3 formulae and the experimental results, the ultimate strengths (bolts and plates) and the nominal dimensions given by the manufacturer were used. 4.1 Bolts in tension and shear The experimental programme concerned three bolt diameters (12, 16 and 20 mm). For each diameter, three test types were realized with four specimen for every type to verify the uniformity of results. The three test types are : direct tension, shear in tension and shear in compression. The tested bolts (A4L-80) were provided by Ugivis with a shank thread on all their lengths. Tests are realized on complete bolts (shank and nut). The stresses at failure were obtained by dividing the failure load by the nominal bolt section areas. The test values of resistance in direct tension are higher than guaranteed values (Table 2). Elongations to failure are also higher than the minimal values given by standards. Bolt shear tests were carried out in two different ways, with plates either in tension (τt ) or in compression (τc ). The ratio between the shear resistance in tension to that of direct tension gives mean values going from 0.69 to 0.62 for diameters from 12 mm to 20 mm. The comparison is based on the mean values of half resistance for each diameter with double shear planes (twice the net cross-section 315

Table 2. Test and nominal bolt resistances in tension compared to that in shear by tension or compression. Shear in tension τt and compression τc

Tension (fu ) Bolt diameter

Test (MPa)

SD (MPa)

Nominal (MPa)

Test/ Nominal

τt (MPa)

SD (MPa)

τc (MPa)

SD (MPa)

Nominal (MPa)

M12 M16 M20

846 928 845

5.3 11.4 15.0

800 800 800

1.06 1.16 1.06

586 610 525

15.4 26.2 2.7

602 640 564

0.7 5.5 1.9

480 480 480

Mean value and SD: standard deviation.

Table 3. Geometrical characteristics of the tested connections (bearing parameter α). Bolt diameter

e1 (mm)

p1 (mm)

e2 (mm)

p2 (mm)

(e1 /3d0 )

(p1 /3d0 – 0.25)

M12 M16 M20

22.5 27.5 35

45 55 70

22.5 27.5 35

45 55 70

0.54 0.51 0.53

0.82 0.77 0.81

in shear). This ratio has a high value for the small diameter while direct tension gives comparable ultimate test resistances between diameters 12 and 20 mm. So, the greatest diameter (20 mm), which has a ratio equal to 0.62, is the most close to that considered in the design formulae of the EC3-1.4. For shear in compression, the results dispersion is lower than that for shear in tension for the three diameters, and the average resistances are higher. In reality, the shear test in tension which gives lower mean values of resistance is more secure in the use. Also, bolts are really used in this configuration. This tendency is comparable to that of carbon steel bolts, where tests of shear in compression give resistances higher than those of tests by shear in tension by 6 to 13% (Kulak et al., 1987). One of possible explanations would be the fact that bolts in shear by tension are subjected to the direct tension in addition to shear. The measured bolts elongations are higher than the guaranteed values. For the three diameters the elongation value at failure is between 18 and 21% when the minimum required values are between 9 and 12%. The standard deviation for the measured elongation is less than 1.6%.

4.2 Cover-plate connections Tests conducted at LGC (Bouchair & Muzeau, 1999) concern twelve different configurations for three diameters (12, 16 and 20 mm) and four geometrical configurations of bolts (A2L, A2T, A3 and A4) (Table 3). Figure 2 shows a schematic of test loading and bolts for the tested connections. The thickness of the two lateral plates and central plate are respectively 5 and 10 mm for all the tests. An example of the global force-displacement curve is given in Figure 3, for the connection A2T-20. It shows the large ductility of the connection behaviour. Also, cycles of loading–unloading (AA , BB , CC ) conducted at regular intervals showed that the slopes of the load–displacement curve remain constant during the test (Bouchair & Muzeau, 1999) what gives evidence of a good stability of the connection stiffness in spite of the important elongation which it undergoes. 316

F

U - Test

A4

A3

A2T

A2L

F

Figure 2. Schematic of tested cover-plate connections and different bolts configurations.

500

A2T-20

F (kN)

Fu-t (test)

400 C 300

Fu-c(EC3)

B Fs-c(EC3)

200 A 100

B' A' Us-c

0 0

10

C' Uu-c 20

30

40 50 ∆ U (mm)

60

Uu-t 70

80

Figure 3. Example of the global load–displacement curve (A2T-20 connection).

4.3 Comparison of experimental and calculation results for connections The force–displacement curves (Bouchair & Muzeau, 1999), representing distinct failure modes (bearing, net-section and bolt shear), are analysed for different connections and the results are given in Table 4 for the resistance and the deformation. They show the high ductility of the connections, and the large difference between the test ultimate loads (Fu-t) and the design values calculated according to the EC3 rules with a partial safety factor taken equal to 1 (Fu-c). The major failure modes predicted by calculation according to the EC3 formulae are associated to the plastic limit resistance of the gross section. This limit is reached before the net section ultimate state because the ratio between the areas of the net and gross cross-sections is close to 0.7, while the ratio between the steel elastic limit and its ultimate limit is smaller than 0.5. The tests where failure occurred in the net cross-section show that, even when the 0.9 and kr reducing factors are dropped, the real values of the connection resistance are always higher than those predicted by calculations. It was found that the ratio between the experimental and calculated resistances varied between 1.05 and 1.16 (Bouchair & Baptista, 2002). This fact is probably due to “notch” effects and it is confirmed by other experimental and numerical studies, using finite 317

Table 4. Resistances and displacements from tests and calculations using actual material characteristics (austenitic 304L). Connection

Fu-t (kN)

Uu-t (mm)

Fu-c (kN)

Uu-c (mm)

Fs-c (kN)

Us-c (mm)

Uu-t/Us-c

Fu-t/Fs-c

Failure mode

A2T-12 A2T-16 A2T-20 A2L-12 A2L-16 A2L-20 A3-12 A3-16 A3-20 A4-12 A4-16 A4-20

179.4 341.9 444.5 173.9 234.4 297.1 269.6 496 583.6 345.6 475.8 580.9

10.9 44.2 69.4 40.8 65.1 70.6 14.5 90.2 76.4 34.2 69.6 72.9

174.6 236.2 306.9 125.8 153.7 195.7 251.6 307.5 391.3 251.6 307.5 391.3

9.4 10.7 17.5 7.2 8.4 9.2 9.8 8.7 12.0 6.6 7.8 11.1

129.3 175.0 227.3 93.2 113.9 144.9 186.3 227.7 289.9 186.3 227.7 289.9

3.9 4.1 6.8 3.0 4.1 3.6 4.0 3.2 5.1 3.2 3.6 4.9

2.8 10.8 10.2 13.6 15.9 19.6 3.6 28.2 15.0 10.7 19.3 14.9

1.39 1.95 1.96 1.87 2.06 2.05 1.45 2.18 2.01 1.85 2.09 2.00

Bolt shear Bearing Bearing Bolt shear Net section Net section Bolt shear Net section Net section Bolt shear Net section Net section

Figure 4. Net-section and bearing failure modes (A3-20 and A2T-20 connections).

element models to analyse the elastic–plastic behaviour of a single plate with a hole, subjected to a tension load (bi-directional state of stress and using the Von Mises criterion). Test specimens were able to support high deformations, due to the high ductility of austenitic steel. At the ULS, the hole ovalisation seems to result more from the reduction of net cross-section than from bearing (Fig. 4), except for 2T connections.

4.4 Estimation of the local deformations (bearing and net section elongations) To make an estimation of the elongation from each component of the joint, the global displacement measured on the length (Fig. 2) is divided into three components, corresponding to the elongation in the gross section components of the plates, in the net section components and in the holes bearing the loaded bolts (Fig. 5). This method allows an analytical estimation of the contribution of each component, in order to isolate the one corresponding to the bearing load (Bouchair & Baptista, 2002). The real stress–strain curve based on the Romberg-Osgood formula is used to calculate the deformation at the net section for a symmetrical connection. Experimental study with different cover-plates configurations, with the estimation of the net section elongation, allow the bolt hole elongation due to bearing and shear bolt to be deducted. The main idea is to define an analytical method to evaluate the real deformation around the bolt hole which will avoid the conservative limits. During tests, the deformations corresponding to the maximum load for connections in bearing reach relatively high values. However, at both the ULS and the SLS loading, the calculated deformations still relatively low, either in the case of bearing or net section failure modes. 318

F

F

Main Plate

do

S0/2

S0/2

do

Cover Plate

F

F/2

2.e2 or e2 + p2/2

F/2 U-Test

F

do

S1/2

F

S1/2

do

2.e2 or e2 + p2/2

F

Figure 5. Basic components (A4 connection) used to calculate the elongation and definition of S0 and S1.

Net section elongation (S0) - (mm)

9

M12-(d0=14) M16-(d0=18) M20-(d0=22)

8 7 6 5 4 3 2 1 0 1

1,2

1,4

1,6

1,8

2

σ / σ 0,2

Figure 6. Net-section elongation depending on the mean applied stress (section S0).

The calculated bearing deformations at the SLS remain relatively small (lower than 1.6 mm) with a bearing resistance based on the ultimate limit (fu). It can be easily shown that, if the mean stress at the net section is about 1.5 times greater than the elastic limit, the net section elongation is still lower than 1 mm (Fig. 6), for the three hole diameters used in the tested connections. The new condition introduced by the EC3-1.4, limiting the stress in the net section at the SLS to the yield strength has to be discussed. The estimated bearing deformation at the hole is not very high and the net section elongation is even lower (Table 5). Actually, the length concerned by the net section elongation is small (the hole diameter as a maximum) and even if a plastic deformation occurs, its effect on the hole elongation is always small except at failure, when necking takes place at the net section. In the case of failures in bearing, which presented the most complex behaviour, the ultimate loads were 1.5 to 1.7 times greater than their calculated values, but failure occurred together with holes ovalisation. The main difficulty lies on making a realistic calculation of this ovalisation, which results not only from the bearing effects of the bolt but also from the elongation of the net section. 319

Table 5. Bearing and net-section elongations in different components of the connections. U net

U bear

U net

U bear

Connection

Fu-c (EC3) (kN)

S0 (mm)

S1 (mm)

S0 (mm)

S1 (mm)

Fs-c (EC3) (kN)

S0 (mm)

S1 (mm)

S0 (mm)

S1 (mm)

A2T-12 A2T-16 A2T-20 A2L-12 A2L-16 A2L-20

174.6 236.2 306.9 125.8 153.7 195.7

0.05 0.13 0.16 0.49 0.74 0.79

0.03 0.07 0.09 0.25 0.37 0.41

2.20 2.42 4.11 0.98 1.11 1.31

2.22 2.47 4.17 1.25 1.49 1.69

129.3 175.0 227.3 93.2 113.9 144.9

0.02 0.03 0.04 0.07 0.10 0.11

0.01 0.02 0.03 0.04 0.06 0.07

0.88 0.92 1.59 0.57 0.82 0.70

0.88 0.93 1.60 0.60 0.87 0.74

5 CONCLUSIONS The failure modes in bearing or net cross-section occur with a large deformation. This effect will have some advantages because loads will always be able to be redistributed in a suitable way amongst the bolts. For the resistance calculation of the net cross-section, the stress limitation to the conventional elastic value have to be discussed because the high ratio between the ultimate and the elastic limits. This is principally true for connections in double shear where the symmetry favors a good distribution of loads. However, it is necessary to develop analytical approaches which allow to calculate in a precise way the elongation of the net cross-section and the deformations due to the bearing and to identify the cases for which the excessive deformations of holes have very important effects (inversion of load, etc.). So, it may be desirable to make a distinction between single and double shear connections. The study is now being complimented by the use of finite element modelling in order to better understand connection behaviour. It should also help to extend the study to the design of other geometrical configurations and other types of connections (beam-to-column for example). REFERENCES ASCE. 1990. Specification for the design of cold-formed stainless steel structural members, ASCE Standards, 114 pages. Bouchair, A. and Muzeau J.P. 1999. Comportement des boulons et des assemblages de type couvre-joint en aciers inoxydables. Clermont-Ferrand, research report (french), LGC-CUST, Blaise Pascal University, Clermont-Ferrand, 26 pages. Bouchair, A., Ryan, I., Kaitila, O. and Muzeau, J. P. 2001. Some aspects of the analysis of bearing-type stainless steel bolted joints. 9th Nordic Steel Const. Conf., Helsinki, June 2001, pp. 755–762. Bouchair, A. and Baptista, A. 2002. Strength and deformation of stainless steel bolted joints with reference to Eurocode 3. 3rd Eurosteel Conference, Coimbra, Portugal, september 2002, Vol. 2, pp. 879–888. Bouchair, A. 2004. Assemblages boulonnés de structures en acier inoxydable – Approaches expérimentale et réglementaire. Séminaire sur les méthodes de calcul en vissage et boulonnage, Cetim, Saint-Etienne (France), 21 September 2004, 31 pages. Burgan, B. A. et al. 2000. Structural design of stainless steel members – comparison between Eurocode 3, Part 1.4 and test results. Journal of Constructional Steel Research, Vol. 54, pp. 51–73. CEN. 1992. Eurocode 3 – Calcul des structures en acier, Partie 1.1 : Règles générales et règles pour les bâtiments, ENV 1993-1-1. CEN. 1996a. Eurocode 3 – ENV 1993-1.3 General rules – Supplementary rules for cold formed thin gauge members and sheeting supplémentaires pour les aciers inoxydables. CEN. 1996b. Eurocode 3 – ENV 1993-1.4 Règles générales – Règles supplémentaires pour les aciers inoxydables. EuroInox. 1994. Design Manual for Structural Stainless Steel, Document published by NiDl (Nickel Development Institute).

320

EuroInox. 2002. Structures en acier inoxydable – Guide de conception (version 2). janvier 2002, Published by Euroinox, CTICM, SCI. ESDEP. 1997. Comportement structural et dimensionnement des aciers inoxydables, Leçons 18, Cahiers de l’APK, No. 16, pp. 126–150. Kaitila, O., Bouchair, A. & Muzeau, J.P. 1999. Rigidité des assemblages de type couvre-joint en aciers inoxydables et dimensionnement selon l’Eurocode 3. Rapport de recherche, LGC-CUST, Blaise Pascal University, Clermont-Ferrand, 24 pages. Kim, H. J. &Yura, J. A. 1999. The effect of ultimate-to-yield ratio on the bearing strength of bolted connections. Journal of Constructional Steel Research, Vol. 49, pp. 255–269. Kouhi, J. et al. 2000. Current R&D work on the use of stainless steel in construction in Finland. Journal of Constructional Steel Research, Vol. 54, pp. 31–50. Kulak, G. L. et al. 1987. Guide to design criteria for bolted and riveted joints, John Wiley & Sons. Mäkeläine, P. & Outinen, J. 1998. Mechanical properties of an austenitic stainless steel at elevated temperatures, Journal of Constructional Steel Research, Vol. 46, No. 1–3, paper No. 101. Rasmussen, K. J. R. 2003. Full-range stress–strain curves for stainless steel alloys Journal of Constructional Steel Research, Vol. 59, pp. 47–61. Ryan, I. 2000. Bolted connections (WP4.2), ECSC Project – 7210-SA/327, Final Report for partner UgineCTCIM. SCI. 2000. Development of the use of stainless steel in construction, Contract 7210-SA/842, 903, 904, 327, 134, 425, Final Report to ECSC. Scot, J. G. & Trautner, J. J. 1998. Behavior of stainless steel fasteners under cyclic load. Journal of Constructional Steel Research, Vol. 46, No. 1–3, paper No. 245. Van Den Berg, G. J. 2000. The effect of the non-linear stress–strain behaviour of stainless steels on member capacity. Journal of Constructional Steel Research, Vol. 54, pp. 135–160.

321

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

About the configuration of long links in eccentrically braced frames H. Köber & B. Stef ¸ a˘ nescu Technical University of Civil Engineering, Bucharest, Romania

ABSTRACT: The present paper is intended to illustrate the advantages of a new solution for eccentrically braced frames with long links. The lateral stiffness of eccentrically braced frames with long links was analyzed and static non-linear analyses were performed to observe the history of the formation of plastic hinges.

1 INTRODUCTION In a well-designed seismic resistant eccentrically braced frame, the greatest part of the energy induced in the structure by strong earthquakes is dissipated through plastic deformations concentrated in the links. To provide this, the links are sized for code specific lateral loads and all the other structural members of the frames (columns, braces, beam segments outside the links) are designed for the forces generated by the fully yielded and strain hardened links. Making the link to be the weakest element of the frame, the designer can force the yielding to occur in the ductile link elements while preventing inelastic deformations in members with a non-ductile behavior. Generally this design concept conducts to smaller cross-sections for the links than for the beam segments near them. Especially in eccentrically braced frames with long links the smaller cross-sections of the links leads to smaller values for the lateral stiffness of structure. On the other hand if the same cross-section is used for the links and for the beam segments outside the links the lateral stiffness of the frame increases but the plastic deformations may occur in other members than links too (a favorable general failure mechanism would be difficult to obtain). As the plastic deformations in a long link are concentrated at the ends of the member, a configuration with reduced beam flanges sections at the member end is proposed. This paper tries to outline the advantages of this solution on a designed structure. 2 DESCRIPTION OF THE ANALYZED STRUCTURE A five-story structure was considered, having the eccentrically braced frames placed as shown in figure 1. The structure has two spans and four bays of 8.0 m. The story height is 3.5 m, the length of the links is 3.4 m. The long links were placed in the central part of the frame girders, between two braces. The eccentrically braced frames were configured according to European codes [2], [4]. It was considered that only the eccentrically braced frames carry the whole horizontal seismic force. The contribution of the unbraced frames was neglected. The configuration of an eccentrically braced frame is shown in figure 2. 3 DESCRIPTION OF THE DIFFERENT STEPS OF THE DESIGN Each of the four eccentrically braced frames was designed to carry a lateral seismic load of Scode = 1030 kN. This value of the seismic load was calculated according to the Romanian code P100-92. 323

Figure 1. Placement of the eccentrically braced frames with long links.

Figure 2. Configuration of an eccentrically braced frame with long links.

Table 1. Values of different i coefficients for the designed link cross-sections. Story

1

2

3

4

5

i

1.899

1.525

1.531

1.599

1.540

According to [3] members not containing seismic links (columns, braces and beam segments outside the links) were dimensioned considering an increased value of the lateral seismic force equal to:

Where  = min (i ) = 1.525; i = Mp,link,i /MEd,i Mp,link,i is the bending plastic design resistance of link “i” and MEd,i is the design value of the bending moment in link “i” in the seismic design combination of actions. The values of the i coefficients determined after the design of the link cross-sections are showed in table 1. According to [3] to achieve a global dissipative behavior of the structure, the cross-sections of dissipative zones were sized so that the values of the ratios i do not differ from the minimum value  by more than 25%. 324

Table 2. Cross-sections of the structural members of frame nr. Story 1

Story 2

Story 3

Story 4

Story 5

Element

Web

Flange

Web

Flange

Web

Flange

Web

Flange

Web

Flange

Link

400 × 10 400 × 10 400 × 15 400 × 15 300 × 10

150 × 20 300 × 20 400 × 25 400 × 30 250 × 15

400 × 10 400 × 10 400 × 12 400 × 12 300 × 10

200 × 20 350 × 20 400 × 20 400 × 25 250 × 15

400 × 10 400 × 10 400 × 12 400 × 12 300 × 10

180 × 20 300 × 20 300 × 15 300 × 15 250 × 15

400 × 8 400 × 8 400 × 12 400 × 12 250 × 8

150 × 20 300 × 20 300 × 15 300 × 15 220 × 12

350 × 8 350 × 8 400 × 12 400 × 12 250 × 8

130 × 15 250 × 15 300 × 15 300 × 15 200 × 12

Beam segment Central column Marginal column Brace

The dimensions in the table are given in mm.

Figure 3. Different constructive solutions for long links in eccentrically braced frames (the solution A-A was used for frame nr.2; the solution B-B was used for frame nr.1 and frame nr.3; the solution C-C was used for frame nr.4).

The cross-sections shown in Table 2 were obtained after the design for the different structural members of the eccentrically braced frames. The frame obtained after this dimensioning was named frame nr.1. It can be observed that after the design the links had smaller flanges than the adjacent beam segments. The smaller cross-sections of the long links conduct to an insufficient lateral stiffness of the frame (see table 6 and 7). The simplest method to assure a greater stiffness of the frame under lateral loads is to consider greater cross-sections for the links (section A-A in figure 3). A frame where the links had the same cross-sections like the adjacent beam segments was considered (frame nr.2). The main disadvantage of this frame is that a favorable general failure mechanism could not be achieved (see figure 4). The plastic hinges do not appear mainly in the seismic links. The next picture shows the different analyzed structural solutions for long links. To obtain a favorable general failure mechanism another structural solution was considered: the small cross-sections were maintained along the whole length of the links; the lateral stiffness of 325

Figure 4. Some steps of the static nonlinear analysis (biographic analysis) of frame nr.2 (many plastic hinges appeared in the braces and beam segments outside the links).

the frame was increased by considering greater cross-sections for the braces, columns and beam segments outside the links. To avoid an excessive material consume the values of the moment of inertia for the different members cross-sections were increased mainly by considering greater heights for the webs. Table 3 contains the dimensions of the cross-sections for the different structural members of this new frame named frame nr.3. This frame had sufficient lateral stiffness and a favorable general plastic failure mechanism, but the greater cross-sections for the braces, columns and beam segments outside the links conducted to a greater material consumption (see table 4). To avoid this, another structural solution was considered (frame nr.4): the members not containing seismic links had the smaller cross-section obtained after the first design (those used for frame nr.1); to assure a sufficient lateral stiffness the links had mainly the same cross-sections like the adjacent beam segments, only near the links ends a reduced width of the links flanges was used. Some steps from the static nonlinear analysis of frame nr.4 are shown in figure 5. 326

Table 3. Cross-sections of the structural members of frame nr.3. Story 1

Story 2

Story 3

Story 4

Story 5

Element

Web

Flange

Web

Flange

Web

Flange

Web

Flange

Web

Flange

Link

400 × 10 400 × 10 500 × 15 500 × 15 400 × 12

150 × 20 300 × 20 400 × 25 400 × 30 300 × 16

400 × 10 450 × 10 450 × 12 450 × 12 400 × 12

200 × 20 350 × 20 400 × 20 400 × 25 300 × 16

400 × 10 450 × 10 450 × 12 450 × 12 400 × 12

180 × 20 300 × 20 300 × 15 300 × 15 300 × 16

400 × 8 450 × 10 450 × 12 450 × 12 300 × 10

150 × 20 300 × 20 300 × 15 300 × 15 250 × 15

350 × 8 400 × 8 450 × 12 450 × 12 250 × 8

130 × 15 250 × 15 300 × 15 300 × 15 220 × 12

Beam segment Central column Marginal column Brace

The dimensions in the table are given in mm.

Table 4. Material consumption. Elements

Frame nr.1

Frame nr.2

Links Beam segments Central columns Marginal columns Braces Total consume

2621 kg 5453 kg 2423 kg 5066 kg 6812 kg 22375 kg

4030 kg 5453 kg 2423 kg 5066 kg 6812 kg 23784 kg

Frame nr.3

Frame nr.4

2621 kg 5648 kg 2530 kg 5500 kg 8019 kg 24318 kg

3774 kg 5453 kg 2423 kg 5066 kg 6812 kg 23528 kg

Table 4 shows the different values of the material consumption for the four analyzed frames. Frame nr.1 is the frame with the cross-sections of the structural members obtained after the first dimensioning. This frame has not sufficient lateral stiffness. The widths of the links flanges are smaller than those of the adjacent beam segments at the whole length of the links. A favorable global plastic failure mechanism is achieved. Frame nr.2 is the frame where the links have the same large flanges like the beam segments near them. This frame has great stiffness under lateral loads but a favorable general failure mechanism cannot be achieved. Frame nr.3 is the frame where members not containing links (columns, braces and beam segments outside the links) have greater cross-sections comparative to those obtained after the first dimensioning. For the links the small cross-sections sized after the first design were maintained. This frame has sufficient lateral stiffness and a favorable global plastic failure mechanism can be achieved, but the material consume is greater. Frame nr.4 is the frame with a reduced width of the links flanges only near the member ends. The other structural elements of the frame have the same cross-sections like frame nr.1. This frame has sufficient lateral stiffness and a favorable global plastic failure mechanism is achieved. The material consume is smaller. The next three tables (Tables 5, 6 & 7) contain the maximal lateral displacements, the relative lateral displacements and the drift for the four analyzed frames in the seismic design situation (seismic design loads combination).

327

Figure 5. Plastic hinges distributions at different steps of the static nonlinear analysis (biographic analysis) of frame nr.2.

Table 5. Maximal lateral floor displacements obtained in the seismic design combination of actions (loads). max (mm) Floor

Frame nr.1

Frame nr.2

Frame nr.3

Frame nr.4

1 2 3 4 5

3.87 10.34 16.77 23.14 29.45

3.27 8.53 13.73 18.73 23.63

3.21 9.01 14.82 20.48 26.30

3.54 9.35 15.09 20.70 26.23

328

Table 6. Maximal relative lateral floor displacements obtained in the seismic design combination of actions. rel (mm) Floor

Frame nr.1

Frame nr.2

Frame nr.3

Frame nr.4

0÷1 1÷2 2÷3 3÷4 4÷5

3.87 6.47 6.43 6.37 6.31

3.27 5.26 5.20 5.00 4.90

3.21 5.80 5.81 5.66 5.82

3.54 5.81 5.74 5.61 5.53

Table 7. Maximal story drift obtained in the seismic design combination of actions. rel (mm) Story

Frame nr.1

Frame nr.2

Frame nr.3

Frame nr.4

1 2 3 4 5

0.00552 0.00924 0.00918 0.00910 0.00901

0.00467 0.00751 0.00742 0.00714 0.00700

0.00458 0.00828 0.00830 0.00808 0.00831

0.00505 0.00830 0.00820 0.00801 0.00790

Where: – rel = lateral floor displacements obtained in the seismic design combination of actions – ψ = coefficient of reduction of seismic forces, takes into account the structural ductility, the stresses redistribution capacity, the capacity of the structure to dissipate energy through inelastic deformations and the vibration damping effects; according to the Romanian norm P100-92, ψ = 0.2 for eccentrically braced frames – h = the floor height; h = 3500 mm for the analyzed structure According to reference [1] the maximal admitted value for the drift was considered:

4 CONCLUSIONS Numerical tests showed a good behaviour of the structure with reduced beam flanges at the ends of long links. These structures have sufficient stiffnes under lateral loads and they develop a favourable global plastic failure mechanism. They are also efficient from the point of view of material consumption. Compared analyses led to the conclusion that the increase of labour cost due to the supplementary cuttings of the flanges at the link ends is fully compensated by the reduction of overall material cost.

REFERENCES 1. Code for aseismic design of residential buildings, agrozootehnical and industrial structures P100-92, Romania, Ministry of Public Works and Territory Planing, 1992 2. Eurocode 3, Bemessung und Konstruktion von Stahlbauten, 1993

329

3. Eurocode 8, Auslegung von Bauwerken gegen Erdbeben, 1998 4. Eurocode 8, Design of Structures for Earthquake resistance, Revised Final Draft May 2002 5. Behavior of long links in eccentrically braced frames, E.Popov; M.D.Engelhardt, Earthquake Engineering Research Center of California, Berkeley 1989 6. Experimental performance of long links in eccentrically braced frames, E.Popov; M.D.Engelhardt, Journal of Structural Engineering, ASCE 11/1992

330

Session 6: Exceptional actions

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

A capacity approach to the design of buildings to resist terrorist attack M.P. Byfield School of Civil Engineering, University of Southampton, Southampton, SO17 1BJ, UK

ABSTRACT: Tall buildings are an important part of modern cities and it is important that developers have the confidence to continue to build high. Prior to 9/11 building designers tended to concentrate on gravity, wind and seismic loads; however, the ability to resist terrorist attack is now a serious consideration. The precise mode of attack by terrorists is impossible to predict. This is identical to automotive design, in which robustness is achieved by targeting failure to energy absorbing large flexural deformations, whilst attempting to limit the sudden losses of strength due to connection failures (known as crumple zone design). In contrast, the present limit state design approach does not target failure. This paper demonstrates that the loads transferable from typical composite beams can be much higher than those predicted from ultimate limit state design. Furthermore, industry standard nominally pinned beam to column connections are shown to lack ductility. The combined effects of over-strength beams and low ductility connections could result in sudden connection failures due to bolt rupture if beams are subjected to the large sagging deflections associated with blast or impact loading. In response to this problem a capacity design approach is advocated. This targets failure through flexural failure in the beams. This is achieved by determining the upper-bound load capacity of the flexural members, thereafter designing all components further down the load path to resist these upper bound loads, rather than the conventional ultimate limit state design loads. The result is frames with relatively strong connections and weak beams. This will improve the ability to absorb energy from blast or impact through beam flexure and will help prevent sudden connection failures that can lead to progressive collapses.

1 INTRODUCTION If a structure is subjected to an extreme loading event, for example, the blast loading from a terrorist attack, collapse can be prevented if the structure is capable of absorbing the immediate energy from the blast, and the subsequent impact from falling debris. It is feasible to design routine structures to survive such events, since composite beams are generally very ductile and are therefore capable of absorbing large amounts of energy. The prime factor to ensure robustness is that connections should be capable of resisting the loading they receive from the beams. One of the first national standards governing the use of steel in buildings (BS449) was issued in 1932 (BSI, 1932). Design was primarily elastic, with no guidance provided for utilizing the substantial post-elastic reserve of strength present in most beams. This seemingly safe approach has an undesirable side effect on robustness. Connections are designed to resist only the elastic loads received from the beams. Since the ultimate flexural strength can be much higher, connections could become overloaded in the event of an extreme load. Thus failure could be via a sudden, catastrophic mode, rather than via a ductile one. This situation has improved somewhat with the introduction of limit state design. Although, as this paper will demonstrate, a mismatch between the strength of beams and that of connections and columns may still exist, preventing structures from realizing their full potential for surviving extreme events. The approach is also wasteful of construction materials, leading to unnecessarily heavy frames. The UK has experienced several terrorist attacks using large truck bombs, such as the 1992/3 attacks on St. Mary Axe (Figure 1) and Bishopsgate in London; the Manchester bomb in 1996 and 333

Figure 1. 1992 St. Mary’s Axe Bomb, London, (Arup, 2003).

Figure 2. Progressive collapse at Ronan Point (UK) in 1968.

the London Docklands bomb, again in 1996. None of these attacks resulted in progressive collapses, with damage mainly restricted to glazing and cladding. Following the Ronan Point disaster in 1968 (Figure 2), in which part of a 22-storey block of flats collapsed as a result of a gas explosion, the UK introduced relatively stringent requirements to avoid similar events. The resulting requirements contained in BS5950 (BSI, 1995) lead to structures with an inherent ability to resist blast loading. 334

Figure 3. Collapse of the Murrah Federal Building in Oklahoma City in 1995, (Sania, 2004).

This is partly achieved by ensuring that all structural members are adequately tied together. Further to this, structural elements vital to overall stability, known as key elements, must be designed to resist abnormal (blast) loads (specified as 34 kN/m2 (DOE, 1991)). The benefits of this approach were revealed in 1995, when the Murrah Federal Building in Oklahoma City partially collapsed with the loss of 168 lives (Figure 3). The collapse occurred as a result of the failure of key elements and the failure to adequately tie members together. It can therefore be speculated that the consequences of the attack by Timothy McVeigh would have been less severe had the philosophy underpinning UK practice been applied during the buildings design.

2 THE UPPER BOUND STRENGTH OF COMPOSITE BEAMS IN FLEXURE There are four main factors that can cause conventional beam design to underestimate strength (Byfield, 2004). The most important is strain hardening. Secondly, steel manufactures routinely supply over strength steel. Thirdly, simple connections can generate very high bending moments when subjected to large rotations; and lastly, high strain rates increase the strength of normal grades of structural steel. 2.1 Strength increases due to strain hardening Conventional predictions of the plastic bending strength of composite beams are based on the rigidplastic stress–strain relationship. This is convenient from a design perspective because it creates simple design formulae and provides good lower bound approximations to strength. In reality however, hot-rolled sections strain-harden shortly after yielding. This is illustrated in Figure 4, which shows the characteristics of modern hot rolled sections based on data from 25 separate coupon tests. During the formation of a plastic hinge considerable straining of the lower flange will occur. This can increase strength over and above that predicted by the rigid-plastic theory. 2.2 Strength increases due to the routine supply of over strength steel It is generally accepted that the nominal yield stress quoted by manufacturers is a characteristic value, i.e. that it corresponds to the lower 95% confidence limit. It follows that the average yield stress will be higher than the nominal value, see Figure 4. The US Military manual TM5-1300 (US Army, 1991) accounts for this over strength by recommending a 10% increase to yield stress when 335

500

Stress (N/mm2)

400 300 200 100 0 0.0

1.0

2.0 Strain (%)

3.0

4.0

Figure 4. Stress–strain relationship for S355 hot rolled steel sections (mean values in solid line, upper and lower 95% confidence limits dashed).

Figure 5. Composite frame test.

designing against the effects of accidental explosions. Furthermore, a survey of over 7000 mill tests (Byfield and Nethercot, 1997) collected from two leading EU producers revealed an average strength increase factor of 16%, where material thickness is greater than 10 mm. For thinner material the average yield stress was found to exceed the nominal value by 37%. The survey also showed that over strength steel is particularly common in S275 steel, because sections failing the S355 grade are often downgraded in classification. In short, it is common for material strength to significantly exceed design values. 2.3 Strength increases due to poor connection ductility Nominally pinned connections such as the partial depth end plate connection have been shown to transform into moment connections when subjected to significant beam end rotations (Dhanalakshmi et al, 2002). This is because a couple is generated between the bolts and the lower 336

Moment (kN.m)

-400 -300 -200

Prying Action

-100 0

0.0

2.0 4.0 Rotation (degrees)

Figure 6. Prying action.

beam flange given sufficient rotation. The test frame shown in Figure 5 represented a bay of a multi-storey building. The moments generated in the partial depth end-plate connection in this frame represented 40% of the nominal plastic moment capacity of the connected beam in sagging, see Figure 6, substantially increasing load capacity. Furthermore, the prying action in the connection was shown to lead to the failure of the upper rows of bolts, increasing the likelihood of a brittle connection failure. 2.4 Strength increases due to high strain rates Blast loads result in very high rates of strain. The dynamic yield stress can be as much as 70% higher than the static yield stress assumed in design, although such an enhancement would be associated with a very extreme rate of strain. The average increase in yield stress assumed by the US military for blast resistant design is 10%, and this contribution to strength enhancement would be a realistic outcome from a terrorist attack. 2.5 Overall strength increases A realistic measure of the true strength of a typical composite beam with typical connections can be gained by consideration of the frame test discussed earlier (Figure 5). The design moment capacity of the section, Mc equaled approximately 1000 kN · m, although the maximum observed flexural strength was 1500 kN · m (due mainly to strain hardening and over-strength steel). The test also demonstrated that the flexible end-plate connection could generate a moment of 400 kN · m even though it is a nominally pinned connection. Thus the beam and connection system was found to be capable of resisting almost double the ULS loading. This can be regarded as a conservative measure of the true flexural strength since the test was aborted after only 300 mm of mid-span deflection due to concern regarding the rupturing of connection bolts.

3 A CAPACITY DESIGN APPROACH Experimental tests (Jolly et al, 1987) have shown that composite beams incorporating over-strength steel sections can exhibit lower bending strengths than otherwise identical beams manufactured from lower strength steel. This reduction in strength illustrates a negative outcome from a seemingly over-safe beam, as the mode of failure was observed to transfer from flexure to the rapid “unzipping” of the shear studs. The large tensile strains that develop during the flexural failure of composite beams are ideal for absorbing energy from short duration but highly intense loads, i.e. from blast or impact. Unfortunately, limit state design does not direct failure. Instead it attempts to prevent it by ensuring the 337

Step 1: Serviceability limit state design of beams for flexure Objective – support working loads elastically

Step 2.2: Estimate moment capacity of end connections subjected to large rotations

Step 2.1: Determine upper-bound moment capacity of beam

Step 2.4: Check upper-bound load capacity > ultimate limit state load, i.e. 1.4 dead + 1.6 imposed

Step 2.3: Calculate upper-bound load capacity of beam

Step 3: Complete beam design Objective – ensure a flexural failure, e.g. design shear studs etc to support upper-bound loads from step 2.3

Step 4: Connection design Objective – to resist upper-bound end moments (step 2.2) and reactions (step 2.3)

Figure 7. The alternative design approach.

strength is higher that the worst-case load a structure is likely to receive. The approach can lead to potentially brittle structures that will under perform if subjected to extreme dynamic loading. Transferring to a capacity design approach would be necessary to ensure that structural failure occurs through flexure. The initial objective for the engineer should be to achieve fitness for purpose under working loads. Beyond that, no attempt should be made to predict the maximum loading a structure would receive. Therefore the lower-bound flexural strength of beams would not be required to resist the ultimate limit state (ULS) requirement. Instead, the upper-bound flexural strength of beams would be established. All other beam failure modes and components further down the load path would be designed to receive these upper-bound loads, not the conventional ULS loads. The approach can be described as a systems approach because the interaction between components throughout the load path is considered. This should improve robustness by targeting flexural strength of beams to be the weakest mode of failure. It should also produce economies in beam design, since the serviceability limit state loads (SLS) would drive beam sizing. The form of the approach is outlined below for the design of composite beams, connections and supporting columns. The approach is also sketched in Figure 7.

3.1 Beam design The initial design of beams would be identical to conventional SLS design. Floor beams would be required to resist serviceability loads elastically. Since they would not be designed specifically for the conventional ULS loads, it may be necessary to apply a small safety factor to dead loads to account for the extra weight due to ponding of concrete and or screed that may occur during unpropped construction. Other than that, beams would be required to resist only the unfactored imposed load. Subsequently the upper-bound flexural strength would be established, inclusive of strain hardening effects, over-strength steel and over-strength but nominally pinned connections. The strength formulae could be calibrated to restrict probability of the upper-bound strength being exceeded to a suitably small target probability. The remaining lower-bound strength checks associated with beam design would then be completed, e.g. shear strength. In contrast with the normal requirement for resisting the ULS loads, these checks would be aimed at resisting the often more onerous upper-bound loads that would 338

result from a flexural failure. In order to comply with existing practice, a check to ensure that the upper-bound flexural strength exceeded the ULS requirement could be included. 3.2 Connection design The connection design would be on the basis of traditional lower-bound strength checks, whereas the applied reaction and moment would be consistent with the upper-bound loads from a beam flexural failure. The design of nominally pinned connections to resist high bending moments may affect connection detailing. The couple between the bolts and the bottom flange in a partial depth end-plate connection would place considerable tensile stresses in the bolts. It may therefore be more economical to amend connection details, to accommodate increased rotations without incurring high moments. This could be achieved by notching the lower flange of beams, preventing the reaction between beam flange and column face, Figure 6. 3.3 Column design The columns would be designed to resist the upper bound reactions and moments received from the beams. Clearly, advantage can be gained by detailing connections such that they develop low moments at high rotations, since it will reduce both the axial load and the bending moments imposed in the columns. Consideration may also need to be given to the lateral movement of perimeter columns that can occur due to the reaction between the beam and column in response to large rotations, Figure 6. During the composite frame test shown in Figure 3 this reaction resulted in 26 mm of horizontal movement in the perimeter column. This movement exceeded the elastic limit and caused a plastic hinge to form in the column.

4 DISCUSSION There are similarities between the design of building structures and automotive structures. In both areas it is impossible to predict the maximum load received during a structures life. Automotive engineers recognise this uncertainty by targeting failure to a defined “crumple zone”. This more intelligent approach has led to a reduction in frame weight. More importantly however, safety has been improved because far more energy can be absorbed during impacts. Modern limit state design codes assume that the maximum load a structure will receive can be predicted. Loads are modelled using the lognormal probability distribution, thereby enabling safety factors to be suitably tailored to restrict failure to a suitably small level. Since the probability of the design load being exceeded is small, no implicit effort is made to ensure failure is ductile. September 11 has however demonstrated that it is not possible to predict the ultimate load a structure will receive, given that buildings can stand for hundreds of years, during which time the environment will undoubtedly change. It could be argued, since structural failures are extremely rare (except in cases of negligence), that the current design approach is adequate. This may be because the existing approach produces unnecessarily heavy structures. But, it is also true that the level of robustness of modern frames is largely unproven, since extreme loading events have fortunately remained rare since the end of World War II. Because of the impossibility of predicting the ultimate loads, it would seem appropriate to target failure, rather than prevent it. Since it is feasible to predict working loads, structural design should ensure frames behave elastically during a working life. Thereafter, sudden failures should be avoided. This can be achieved by ensuring the lower-bound strength of connections and columns exceeds the upper-bound flexural strength of supported beams. In the same way that cars have become lighter and safer with the advent of crumple zone design, a move towards this approach would improve design safety and economy. The upper-bound strength of composite beams can be far higher than the ultimate limit state requirement. Therefore, the SLS design requirements would 339

be likely to govern beam design. This would lead to design economies, whilst improving the ability to survive extreme loading events in a ductile and relatively safe manner. Since safety is the paramount design objective, the opportunity to reduce the cost of construction materials whilst improving robustness would be substantial reward for the effort required to adjust design procedures.

5 CONCLUSIONS The failure mode is more important than absolute strength when designing structures to survive extreme loads such as blast. The optimum failure mode for absorbing energy is through beam flexure. Unfortunately, this paper shows that a flexural failure may be less likely than a connection failure because composite beams possess a “hidden” reserve of strength. A modification to the limit state design approach is advocated that targets failure through beam flexure. The approach would utilise the full flexural strength of the beams in design and would therefore generate substantial savings in the cost of steelwork. More importantly, the ability to survive extreme loads would be improved, because the connections would be designed to withstand the maximum load transferable from the beams.

ACKNOWLEDGEMENTS The composite frame test reported was carried out in collaboration with the Steel Construction Institute the UK and the Building Research Establishment using funding from the EPSRC and DETR. This paper expands on aspects of work described in (Byfield, 2004). REFERENCES Arup. 2003. Terrorism – why buildings collapse. Report by Arup Security Consulting, London. BCSA and SCI 1991. Joints in simple construction, Vol. 1: design methods. The Steel Construction Institute, Publication No. 205. BSI 1932. British Standard Specification No. 449 for the use of structural steel in building. British Standards Institution, London. BSI 1985. BS5950: Structural use of steel in building. British Standards Institution, London. Byfield, M.P. 2004. The design of steel framed buildings at risk from terrorist attack. submitted for publication in The Structural Engineer. Byfield, M.P. and Dhanalakshmi, M. 2002. Derivation of strain hardening factor from mill tests. International Conference on Advances in Steel Structures (ICASS 02), Hong Kong. Byfield, M.P. and Nethercot, D.A. 1997. Material and geometric properties of structural steel for use in design. The Structural Engineer, Vol. 75, No. 21, Nov., 363–373. Dhanalakshmi, M., Byfield, M.P. and Couchman, G.H. 2002. Composite connections at perimeter locations in unpropped composite floors. International Conference on Advances in Steel Structures (ICASS 02), Hong Kong. DOE 1991. Manual to the Building Regulations. Department of the Environment, HMSO, London. Jolly, C.K., Moy, S.S.J., and El-Shihy, A.M. 1987. Flexural Tests on Composite Beams with Unwelded Shear Connectors. Structural Assessment – The Use of Full and Large Scale Testing. Garston, Watford, Building Research Establishment, 90–96. US Army 1991. Technical Manual TM5-1300: Structures to resist the effects of accidental explosions, Chapter 5: Structural Steel Design. US Department of Commerce, National Technical Information Service, Washington, DC. Sania 2004. www.sandia.gov/Surety/Surehome.htm. Accessed 26/8/4.

340

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

The collapse of WTC twin towers: general aspects and considerations on the stability under exceptional loading of columns with partial-strength connections A. De Luca, E. Mele, A. Giordano & E. Grande Dep. of Structural Analysis and Design, University of Naples Federico II

ABSTRACT: The collapse of WTC started a wide scientific debate in the engineering community about tall buildings’ design and safety measurements. It has been proven that a main role on the sudden failure of the twin towers has been played by connections. The structural system featured partial strength continuity connections at columns distributed along the height and width of the building with different configurations, in so-called staggered patterns. This was mainly due to economical reasons, since small bending moment values in such zones do not always require full resistance, and to space considerations regarding the overall dimensions of the joints. As a result, vertical members were designed mainly against vertical loads, while under exceptional loadings they may possibly undergo lateral deflection, as dramatically occurred on the September 11th. In this paper, a brief description of the partial strength connections of WTC perimeter columns is presented. Then, the effect of partial strength column splices and of their positioning along the height of the column on the ultimate behaviour of the member is evaluated through parametric nonlinear analyses. The column is represented by two elastic elements connected by a rigid-plastic spring. P-δ effects are taken into account, while residual stress and out of camber are implemented using the concept of equivalent geometrical imperfection.

1 INTRODUCTION The World Trade Center (WTC) towers were highly redundant structures and had significant overstrength as a result of structural typology (framed tube with top outrigger trusses), of horizontal and vertical stiffness requirements, and of some design load conditions which exceeded the minimum code provisions (De Luca et al., 2003). However, it seems that the huge reserve of strength of the structure could not be completely utilized in the exceptional load conditions of September 11th, mainly due to fire, but also due to some structural weakness. In fact, from the observations made in the aftermath of the WTC towers global collapse (FEMA, 2002; NIST, 2004; Astaneh, 2003), a number of local failure modes, involving single structural components can be recognised. In particular the local failures mainly concern the different connection types present in the framed tube, suggesting that connections might represent a weak point in the structural system and may have played a major role in the global collapse. In fact, both column and spandrel bolted splices, which connect each other the prefabricated panels of the exterior tube, were partial strength (PS) connections. In addition, the floor to exterior column connections, simply made up of a bearing seat with single bolt, are also thought to be somewhat responsible for the collapse, since their failure eliminated columns transversal restraint, thus affecting their buckling behaviour. In this paper, the effect of partial strength column splices on the ultimate behaviour of plane frames is evaluated through parametric nonlinear analyses. In this aim, first the failure modes of the bolted column splices of the WTC tower perimeter structure are examined, and the relevant resistance are evaluated; then the ratio of connection to column bending moment capacities is evaluated along the height of the tower for a representative column alignment of the exterior tube; 341

finally, the ultimate vertical load behaviour of column characterised by different PS column splice is evaluated by means of nonlinear analyses. The beam-column model used in the nonlinear analyses is described and its accuracy is proved by reproducing Eurocode 3 (EC3, CEN 1992) stability curves. 2 ANALYSIS OF CONNECTIONS IN THE WTC STRUCTURAL SYSTEM The WTC towers were the first high-rise buildings to implement the concept of tube as lateral load resistant system. The tube, working as a 3D system, utilized the entire building perimeter to resist wind loads, while the central core structure only carried gravity loads. The tube behaviour was achieved by arranging closely spaced columns (at 1016 mm distance, for a total of 59 columns on each façade) connected by deep spandrel beams, so that the four exterior walls, each acting as a huge Vierendeel truss, formed a cantilever beam (Framed Tube, FT) with square box section, internally-braced by the floor system. The perimeter columns had built-in sections made of 4 shop-welded plates, with thickness ranging from 6.35 mm to 101.6 mm along the height. Twelve grades of steel were used for these columns, with yield strength ranging from 290 MPa to 690 MPa (FEMA, 2002). The high spandrel plates, which linked adjacent perimeter columns at each floor level, were 1321 mm deep and typically made by the same steel grade of the columns. Prefabricated “tree” panels of the façade, usually three story tall and three columns wide, were site assembled through high strength bolted connections. Therefore, while shop-welded connections were used for the assemblage of the sub-components of the tree panels (plates of built-up column section, spandrel-to-column connections), bolted connections were used for assembling the prefabricated panels of the façades. Three main type of bolted connections were present in the FT structure: end plate connections at column splices, inside and outside plate connections at spandrel splices, and floor-to-column/spandrel connections. Due to page limitations, in this paper only the end plate connections at column splices is examined. 2.1 Bolted column splices (end plate connections) The prefabricated panels of the exterior wall were connected each other through bolted splices, made of a butt plate with four bolts in the upper stories and six bolts in the lower stories (figure 1). The bolts, according to (NIST, 2004), were generally high-strength bolts ASTM A 325 (or ASTM A 490) type 1, with Fy = 120 ksi (equal to 827.4 Mpa) for diameters up to 1 in (25.4 mm), and Fy = 105 ksi (equal to 723.94 Mpa) for larger diameters. All column end plate bolted connections appeared to fail from unanticipated out-of-plane bending (figure 2). The bolts either exhibited classical tensile fracture or bent in the shank, and no plastic

Figure 1. Column connection detail.

Figure 2. Failure of column connection.

342

deformation was observed in the end plates (FEMA, 2002). This failure modes are a consequence of the unanticipated out-of-plane deformations of columns: in fact, while bending moments at the bolted splices, located at mid interstorey height, were small in the design scenarios, in the out-of-plane deformation mode they increased up to exceed the ultimate moment capacity of the connection and resulted in unstable deflection of columns. According to appendix B of FEMA report (Fisher & Iwankiw, 2002), the moment capacity of connections can be computed following AISC specifications of 1963. Namely, the ultimate moment capacity of connection is equal to moment capacity of bolt group:

where a is the distance of the inner bolts group by inner edge of column; b is the distance between bolt groups; Fn = Ct Fu is the nominal tension strength of bolt; Ct = 0.75, is the ratio of the stress area to the nominal shank area; Fu is the bolt tensile strength; Ab is the nominal shank area of bolt; 1.18 is a factor accounting for overstrength of A325 bolts (according to studies carried out by Kulak in the 1960s and 1970s, as reported in (Fisher & Iwankiw, 2002)). Based on this expression, the ultimate moment capacity of the connections of the centre column of North WTC perimeter side, column “130A”, has been calculated and the values are reported in table 1, while table 2 reports plastic moments, partial strength ratios and non dimensional slenderness with reference to upper and lower columns. In figure 4. the ratio of the flexural moment capacity of connections to the plastic moment capacity of column 130A is reported as a function of story level. The ratio ranges from 0.22 to 0.40, with the lowest values at the lowest storeys. With reference to the same rigid end-plate connection mechanism (fig. 3 right), and assuming elastic-perfectly plastic axial behaviour of bolts, the initial stiffness, the yield moment (corresponding to yielding of a single line of bolts), and the post-yield stiffness have been calculated as follows:

Table 1. 130A column connections properties.

Table 2. 130A column properties.

Story

Ab (mm2 )

Type of bolts mat.

Mu conn (kNm)

10–50 60 70–80 90–100

642.42 642.42 506.71 394.08

A490 (Fu = 1500 Mpa) A325 (Fu = 827 Mpa) A325 (Fu = 827 Mpa) A325 (Fu = 827 Mpa)

343 274 216 168

Story

Mu cl (kNm)

PS ratio

λ

10 100

1559 419

0,22 0,40

0.36 0.30

F2,δ2 F1,δ1 a

b

a

b

Figure 3. Collapse behaviour scheme (left); rigid end plate mechanism behaviour (right).

343

400 OUT-PLANE BENDING MOMENTS (kNm)

Mconnection / Mpl,column

1,20 1,00 0,80 0,60 0,40 0,20 storeys

0,00

CONNECTIONS BEHAVIOR

350 300 250 200 150

STOREYS 10-50 STORY 60 STOREYS 70-80 STOREYS 90-100 ROTATIONS (rad)

100 50 0

10

20

30

40

50

60

70

80

90 100

Figure 4. Connection/column capacity ratio.

0

0,005

0,01

0,015

0,02

0,025

Figure 5. Moment–rotation law for connections.

V0

Moment

ELASTIC ELEMENT

h

RIGID-PLASTIC SPRING

ELASTIC ELEMENT Rotation

Figure 6. Column element and moment–rotation behaviour of the hinge.

In figure 5, the simplified tri-linear moment-rotation curves are provided for the splice bolted connections of column 130A at different examined storeys. Note that while elastic stiffness is comparable for all stories, the yielding and ultimate moments are larger at lower stories, the same as for plastic moment of columns. 3 THE COLUMN MODEL FOR BUCKLING ANALYSIS The model adopted for column elements should account for residual stresses and out of camber effects typical of production members. Nevertheless, it has to correctly predict the failure load at any given value of the slenderness, so that the column buckling curves are matched as closely as possible. As far as the out of camber and residual stresses effects are concerned, the concept of equivalent geometrical imperfection is used, which has shown to well represent both phenomena. According to this concept, a concentrated plasticity model can be used. The above described concept can be easily implemented by modelling the column with two linear elastic elements, linked by a rigid-plastic spring. The geometrical imperfection is given by transversely displacing the mid node where the hinge is positioned (fig. 6). 344

1.0

N / Ny

λ = 0.6

0.8

λ = 1,0 λ = 1,6

0.5

λ = 2,0 λ = 2,6

0.3

0.0

0

0.02

0.04 0.06 Nondimensional displacements

0.08

0.1

Figure 7. Non-dimensional force-displacement curves for single columns.

In previous works (De Luca et al., 1993) this approach has shown terrific accuracy in reproducing the code column stability curves in the entire range of non-dimensional slenderness, varying from 0.26 up to 3, regardless of the equivalent geometrical imperfection. In this paper, similar column element has been implemented in the computer code ABAQUS, using B31 type elements for the elastic component and CONN3D2 non linear connector element for the plastic hinge. Figure 7 shows the results of numerical non linear analyses, performed on columns with different values of non-dimensional slenderness and with 1/250 initial geometrical imperfection. In such curves the actual load is normalised to the squash load, while the displacement measured transversely at hinge position is normalised to the nominal length of the element. These plots are used to derive the buckling curves by extracting each maximum value. This is shown in figure 8, where the numerically derived results are represented by bullet points, while the continuous curves are EC3 buckling curves. The maximum errors are obtained for extreme slenderness values, which are thought to be less significant.

4 EFFECT OF PS CONNECTION ON THE COLUMN BEHAVIOUR Once the reliability of the model has been assessed, the effect of partial strength connection on the buckling behaviour has been studied. To do this, the plateau level of the hinge element has been varied in order to model reduced connection plastic capacity with respect to the one of the connected elements. Partial strength ratios of 0.3, 0.5, 0.7, 0.8 have been selected. The results of the numerical analyses performed with such values are shown in figure 9. Quite obviously, the weaker the connection, the smaller the carried load. More interesting is the reduction ratio as a function of the non dimensional slenderness, which is shown in figure 10. All curves show similar shape. Actually, the initial reduction percentage rapidly increases to achieve the maximum value at λ = 1, with a subsequent degrading path characterized by a lower rate. In case of small slenderness values, the element failure is mainly governed by the squash load, so that the reduced strength of the connection has little influence on the carried load. Similarly, for high slenderness (non-dimensional slenderness greater than 2), the column may undergo elastic buckling, with again very small role played by the partial strength of the connection. On the other hand, for medium slenderness values, interaction between buckling and plasticity phenomena takes place, with stronger influence of the connection strength on the column behaviour. Another aspect that might possibly turn out interesting from an applicative point of view is the position of the hinge along the height of the element. To investigate its influence, the same column model has been used for the analyses, this time positioning the hinge at different stepwise locations, identified by their non-dimensional abscissas z/l respectively equal to 1/2, 1/3, 1/4, 1/5, 1/6. 345

1.0 EUROCODE 3 curve a

0.8

Nu / Ny

NUMERICAL

0.6 0.4 0.2 λ

0.0 0.0

0.4

0.8

λ

1.2 0.6 0.890 0.884 0.67

0.3 0.978 0.956 2.20

EC3. a Num. error %

Nu/Ny

1.6 0.8 0.796 0.760 4.49

1.0 0.666 0.640 3.85

2.0 1.3 0.470 0.469 0.28

2.4 1.6 0.333 0.331 0.66

2.8 1.8 0.270 0.252 6.67

3.2 2.0 0.223 0.220 1.30

2.3 0.172 0.170 1.05

2.6 0.136 0.137 0.59

3.0 0.104 0.099 4.44

2.3 0.162 0.152 6.43

2.6 0.130 0.120 7.60

3.0 0.099 0.091 8.30

1.0 EUROCODE 3 curve b

0.8 Nu / Ny

NUMERICAL

0.6 0.4 0.2 λ

0.0 0.0

0.4

λ EC3. b Num. error %

Nu/Ny

0.8 0.3 0.964 0.959 0.50

1.2

0.6 0.837 0.879 4.95

0.8 0.725 0.760 4.84

1.6

2.0

1.0 0.597 0.623 4.41

1.3 0.427 0.431 0.94

2.4 1.6 0.308 0.300 2.44

2.8 1.8 0.252 0.241 4.30

3.2 2.0 0.210 0.198 5.37

1.0 EUROCODE 3 curve c

0.8 Nu / Ny

NUMERICAL

0.6 0.4 0.2 λ

0.0 0.0

0.4

λ EC3. c Num. error %

Nu/Ny

0.8 0.3 0.949 0.910 4.12

1.2 0.6 0.785 0.765 2.60

0.8 0.662 0.646 0.72

1.6 1.0 0.540 0.51 5.54

2.0 1.3 0.389 0.390 0.31

2.4 1.6 0.2842 0.281 0.60

2.8 1.8 0.235 0.238 1.49

3.2 2.0 0.196 0.199 1.43

2.3 0.154 0.151 1.76

2.6 0.123 0.124 0.49

3.0 0.095 0.090 5.36

Figure 8. Comparison between cumputed and EC3 buckling curves.

The value of the non-dimensional slenderness chosen is equal 1, in accordance to what discussed here above. As easily seen from figure 11, the weakest behaviour is obtained with the hinge positioned at mid-way along the element, what appears to be trivial given the geometrical configuration. Nevertheless, almost 50% capacity can be gained, though it must be said that, when the column is part of a structure, the bending moment from external actions is very little at mid height and can increase in other sections. For this reason, it can be advised to position the connection at different locations only if the member is basically simply compressed. 346

Figure 9. Buckling curves for different partial strength ratios and imperfection values (from top: 1/500, 1/250, 1/200).

5 APPLICATION TO WTC COLUMN 130A Column 130A of the towers had variable non-dimensional slenderness and plastic moments depending on the section and material at different floors. The main properties of the column are summarized in table 2. At lower floors the slenderness was = 0.36 and the partial strength connection factor PS = 0.22. For these values, according to what reported in this paper, the reduction of buckling capacity of the column is around 8%. This consideration suggests that PS ratio of the connection might have been a secondary issue in the collapse of the columns. More extensive and complicated models which take into account the entire panel with the staggered connections and the out of plane possible buckling instability of the entire panel also taking into account the collapse of the floor to column connections could provide more precise information about the possible contributions to the collapse. 347

35 Partial strength ratio = 0.3 Partial strength ratio = 0.5 Partial strength ratio = 0.7 Partial strength ratio = 0.8

Percentual reduction

30 25 20

imperfection = 1/500

15 10 5

λ

0 0.0 35

0.5

1.0

1.5

2.0

Percentual reduction

30

2.5

3.0

3.5

Partial strength ratio = 0.3 Partial strength ratio = 0.5 Partial strength ratio = 0.7 Partial strength ratio = 0.8

25 20

imperfection = 1/250

15 10 5

λ

0 0.0 35

0.5

1.0

1.5

2.0

3.0

3.5

Partial strength ratio = 0.3 Partial strength ratio = 0.5 Partial strength ratio = 0.7 Partial strength ratio = 0.8

Percentual reduction

30

2.5

25 20

imperfection = 1/200

15 10 5

λ

0 0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

Figure 10. Percentual reduction as a funtion of the slenderness and partial strength ratio (from top: 1/500, 1/250, 1/200). 0.8 N/Ny

imperfection = 1/250

z/l = 0.5 z/l = 0.33

0.6

z/l = 0.25 z/l = 0.2

0.4

z/l = 0.166 0.2 Non-dimensional tranverse displacement ( d/length) 0 0

0.005

0.01

0.015

0.02

0.025

Figure 11. Influence of the hinge position on the buckling behaviour of the column.

6 CONCLUSIONS In this paper, a brief description of the partial strength connections of WTC perimeter columns is presented. It is shown that the connections are capable of providing only 20% of the ultimate bending capacity of columns and therefore the buckling capacity of columns is undermined. A model is 348

then presented to evaluate the buckling capacity of columns with partial strength connections. The model allows to state that the reduction of buckling capacity of columns at bottom stories in WTC is around 8%. The model has demonstrated the importance of defining ad hoc buckling curves for columns with partial strength connections. The reduction of buckling capacity can be in fact up to 30%. These partial strength connection buckling curves have been presented in section 4 of this paper. This reduction leads to minor safety of the column when subjected to exceptional loadings which determine unpredicted out of plane behaviour. The case of WTC towers needs, in this specific field of buckling of columns, more refined investigation and modeling taking into account buckling of the entire panel and connections in a staggered configuration.

ACKNOWLEDGEMENTS The authors would like to thank professors A. Astaneh-Asl and R. Testa for the useful discussion exchanged on the subject. REFERENCES Astaneh-Asl, A. 2003. WTC collapse, field investigations and analyses. Emerging Technologies in Structural Engineering – Proc. of the 9 th Arab Structural Engineering Conf. Dec., Abu Dhabi, UAE CEN. 1992. ENV 1993-1-1: Eurocode 3. Design of steel structures. Part 1-1: General rules and rules for buildings. CEN/TC250. De Luca A., Faella C., Mele E. 1993. Advanced inelastic analysis: numerical results and design guidelines for rigid and semirigid sway frames. in: “Plastic Hinge Based Methods for Advanced Analysis and Design of Steel Frames. An Assessment of the State-of-the-Art”, SSRC, March 1993. De Luca, A., De Stefano, M & Mazzolani, F.M. 1995. Frame imperfections for advanced inelastic analysis of sway frames. Proceedings, American Society of Civil Engineering, Mech. Division, Boulder. De Luca, A., Di Fiore, F., Mele, E., Romano, A. 2003. The collapse of the WTC twin towers: preliminary analysis of the original design approach. Proc. 4th Int. Conf. STESSA. June, Naples, Italy. FEMA. 2002. World Trade Center Building Performance Study: Data Collection, Preliminary Observations, and recommendations. FEMA403. May. New York: ASCE. Fisher, J & Iwankiw, N. 2002. Structural Steel and Steel Connections. FEMA 403 – Appendix B. NIST. 2004. June 2004 – Progress Report on the Federal Building and Fire Safety Investigation of the World Trade Center. NIST-SP1000-5

349

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Collapse analysis of timber and steel sway frames under increasing fire temperatures B.W.J. van Rensburg University of Pretoria, Pretoria, South Africa

ABSTRACT: It is often necessary to determine the behaviour and time to collapse of a frame subjected to the action of a fire. Given the time – temperature relationship for a plane timber or steel frame, the paper shows how a collapse analysis of the frame could be conducted. The effect of temperature with time on the timber and steel properties and thus on the members is considered. The effect of charring is also considered. Iterative non-linear analyses are done for selected temperature distribution increments. The geometric non-linear (or second-order) effects are considered and the changes in material and member properties are taken into account to compute action effects. The member (out-of-plane) resistances are also computed. Given a factored imposed load case, the collapse mode and the time to collapse can be determined. The procedure is described with the help of a flow diagram.

1 INTRODUCTION Fire resistance and protection are important issues when designing steel and timber buildings and these considerations influence the economics and feasibility of the structure (Östman, 2004). The scientific and engineering modelling of the fire performance of such structures is a rapidly advancing research field. There is a move away from prescriptive rules for fire insulation towards performance-based codes (Buchanan, 2001). This development allows room for innovation. It is more realistic to determine the fire resistance of a redundant (statically indeterminate) frame based on the behaviour of the complete structure than only basing the resistance on the behaviour of individual components. The fire resistance tests on the Cardington Building (Newman et al., 2000) indicate that standard fire test results are not a reliable measure of real building performance in fire. Fire engineering principles allow the designer to determine a temperature–time relationship for a fire in a compartment according to the fire load, its distribution, ventilation conditions and sizes and positions of openings. Given the nature and thickness of the insulation and ceiling (if any), the properties of the materials and the shape and properties of the structural members, the temperature in the structural members can be computed (Walton et al., 1995). The changes in the strength and stiffness properties of the members can subsequently be determined. A structural analysis for each time step of the fire can be done given the changing structural properties. The structural analysis yields the action effects. The resistance in combined bending and compression can also be computed for each time step. When an action effect exceeds the corresponding resistance, the structure can be considered to have failed. It is consequently possible to predict the behaviour of the structure during a fire, the collapse mode and the time to failure if this is to occur. This paper describes a computer program that analyses a primary plane frame of the structure consisting of beams and columns with semi-rigid connections. The expansion and bending of members due to temperature increases are considered and a second-order analysis is done. 351

@

@@

@

@

@

@

Figure 1. Plane frame with semi-rigid connections.

2 STRUCTURAL ANALYSIS OPTIONS Efficient commercial structural analysis packages are available. For frames several such packages (Prokon, 2000) can model the semi-rigidity of joints by using spring elements. Second-order analyses (moment-magnification) are also possible. Changes in temperatures can be specified but very often only as uniform throughout the member. The material and member cross-sectional properties are constant for each analysis. That implies that the user has to change the properties after each time step (i.e. manual intervention). Certain finite element packages (Hibbit et al., 1998) can model both material and geometrical non-linearities. (By a finite element package is meant a program that contains, amongst others, membrane and block elements.) The properties can be varied through the depth of structural members (by dividing member sections into a number of elements. Thus frame members (beams and columns) are made up of three-dimensional elements. Modelling semi-rigid connections becomes quite intricate. The resulting input for a general-purpose finite element packages becomes somewhat involved. These general-purpose finite element packages normally do not compute the resistances of members (beam-columns) according to the design codes. Finite element packages (modelling both material and geometrical non-linearities) have an important role in forensic work or as a research tool. The process of analysis proposed in this paper extends the normal frame analysis options available to a structural designer to include semi-rigid connections, temperature effects and resistance computations (See Figure 1). The process involves the determination of the structural frame behaviour for each time step increase in temperature). The resistance of the structural members is also computed for the same time step. The action effects (moments and forces) are compared to the member resistances. The structural behaviour under increasing temperature and the time to failure (if collapse is to occur) can then be predicted. 3 THE EUROCODES The Eurocodes provide a framework for the consistent design of structures. Partial factors and combination factors to be applied to the actions are specified. Global analysis methods, including imperfection definitions, stability criteria, second-order analysis techniques are outlined. On the other side of the equation, criteria to establish column, lateral-torsional beam and beam-column resistance are codified. Fire design rules (König, 2004) are being refined. The Eurocode methods are being implemented into the outlined program as they are finalized. 4 THE COMPUTATION PROCESS 4.1 The matrix stiffness method The matrix stiffness method for linear-elastic (first-order) analysis is well documented (Coates et al., 1988). The displacements and rotations of the nodes (or joints) are considered as the unknowns. 352

Compatibility and equilibrium are evaluated and a set of equations in terms of the unknowns is compiled and then solved. The main steps of the method are: – – – – – – –

determination of the stiffness matrices for the elements; transformation from element to structural axes; compilation of the structural stiffness matrix; introduction of boundary (support) conditions; solving for the unknowns; transformation from structure to element axes; and determining the forces and moments in members.

4.2 Second-order analyses In a first-order analysis the effect of axial forces on the stiffness matrix of a member is disregarded and the equilibrium equations are based on the undeformed structure. The matrix stiffness method can be adjusted to do second-order (geometric non-linear) analyses (Coates et al., 1988), (Allen et al., 1980). The effect of axial forces on the stiffness matrix of a member is taken into account and the equilibrium equations are based on the deformed structure. A step-wise iterative process could be followed (Coates et al., 1988): At a given load-factor the axial forces in members are taken as zero (or estimated) for the first iteration. After an analysis the axial forces and displacements are known and the effect of the forces and displacements can be included. Iterations are carried out until a desired degree of convergence has been obtained. At a next, higher, load-factor the process is repeated. When a zero or negative determinant for the structural stiffness matrix, with the increasing load factor, is found, the structure has become unstable and the calculations are terminated. 4.3 Second-order elastic-plastic analyses With an elastic-plastic stress–strain relationship for the material (of the members or connections), the structure could reach a material based collapse mode (as a plastic mechanism) before the secondorder elastic limit load (the elastic instability load) is reached (Allen et al., 1980). Second-order elastic-plastic analysis procedures for steel frames have been described (Coates et al., 1988). A step-wise iterative process is again followed. During an iteration the actions are checked against the member resistances. If the moment exceeds the plastic moment of resistance a plastic hinge is inserted in the structure and the stiffness matrix is modified for subsequent iterations. 4.4 Semi-rigid connections It is appreciated that it is more realistic to model many connections in steel or timber frames as semi-rigid (Madsen, 2000). The traditional matrix stiffness method for frames could also modified to handle semi-rigid connections (Kruger et al., 1993), (Al-Jabri, et al., 2001), (Liu, 1998). The analysis of steel frames combining second-order elastic-plastic analysis with semi-rigid connections has been demonstrated (van Rensburg et al., 1994), (Kruger et al., 1995). Many different connection types for timber frames containing steel have been proposed; amongst others, complete steel joints, connections with exposed steel plates, connections with hidden steel plates and connections with steel dowels (Madsen, 2000), (Goetz et al., 1989), (Racher, 1995), (Buchanan et al., 2001), (Leijten, 1999). 4.5 Rising temperature The second-order elastic-plastic analysis procedure, as outlined above, could be modified that, instead of analysing for an increasing load-factor, an analysis is done at a constant load-factor but for step-wise increments in temperature. 353

When the temperature rise significantly under constant imposed load the action effects change. The expansion and bending (due to temperature gradients) will cause additional (compressive) forces and bending moments. In addition, the deflections (and relative rotations in semi-rigid connections) will increase and moment redistribution due to yielding could take place in ductile (steel) members and connections. The increasing deflections increase the second-order moments (the stability effects or P--effects). The deformations thus not only redistribute moments but also increase the moments. The resistance of the structure decreases with significant rise in temperature. The yield stress of steel components lowers and timber cross-sections reduce due to loss of material (AFPA, 1999), (Hartl, 1995a), (Hartl, 1995b). These changes lead to reductions in both the resistance based on the strength of the material and the resistance based on stability (column buckling or beam lateral-torsional buckling or beam-column instability) (Allen et al., 1980). For each time step (temperature level) the properties of the steel (the modulus of elasticity and yield stress) are recalculated. This results, on the one hand, in increasing in-plane deflections and forces and moments due to decreasing stiffness and, on the other hand, lowering resistance due to weakening material strengths and (also out-of-plane) stability effects. 5 INPUT AND COMPONENT CONSIDERATIONS 5.1 Load-factors A fire resistance analysis is a consideration of an ultimate limit state. Partial factors have to be applied to the permanent and imposed loads together with the load combination factors applicable to a fire limit state (EN 1991-1-2:2002). 5.2 Insulation properties Information on the heat insulation properties of, for example, various new spray-on products for exposed steel connection elements, are being generated. It is thus advantageous to develop the structural analysis system in a transparent modular programming environment (such as MATLAB Chapman, 1999). 5.3 Material properties The structural material properties will influence both the structural actions and resistances. Data on the properties of structural steel and timber and timber charring rates at extreme temperatures are given in the Eurocodes and have also been published elsewhere (Buchanan, 2001), (AFPA, 1999), (Hartl, 1995b). When the information have been finalized this should also be incorporated into the program. 5.4 Structural and member properties The higher conductivity and expansion coefficients of steel will have a significant influence on steel component behaviour. Steel members will expand and flex subjected to a temperature gradient. This will induce forces and moments in a statically indeterminate frame. The charring of timber in the heat-affected zones will reduce cross-sections influencing both the stiffness and resistance of members. The changing stiffnesses due to yielding or changing cross-sections will influence the redistribution of forces and moments in statically indeterminate frames. 5.5 Connections Many connections have been proposed for timber frames. The behaviour of these connections in static and even dynamic situations has been documented. Further information is necessary for modelling some of these connections in a fire situation. The semi-rigidity of connections could be 354

modelled by using internal rotational springs (Madsen, 2000). With the matrix stiffness method, internal springs could be are handled by two alternative methods. With the first method an additional degree of freedom (a rotation) is added to the vector of unknown displacements/rotations for each member connected with a semi-rigid joint. With the second method the element stiffness matrix is modified (Kruger et al., 1993) to incorporate the additional flexibility (static condensation). To improve the general application of the program it was decided to incorporate a connection element in the program so that not only rotational stiffness is considered but also the two linear stiffnesses. 5.6 Geometric non-linear analysis The use of the stability functions in the analysis of frames and the iterative solution process has been documented (Coates et al., 1988) and is effective in doing a second-order analysis. The possible pitfalls when using the stability functions (when the axial force in a member is small) have been described elsewhere (Kruger et al., 1995). 5.7 Member resistances The resistances of compression and flexural members and beam-columns are formulated to different degrees of complexity. The advantage of a computer program is that the resistance can be programmed with the aid of transparent (even if it implies using ‘complicated’) formulae Allen et al., 1980) instead of resorting to simplified (obscure) design equations. The Eurocodes provide the necessary formulae. 6 PROGRAMMING LANGUAGE It will be shown how this process of analysis could be computerised. Research results (on fire computations and high temperature resistance of members and frame connections) are continuously being updated. It is thus desirable not to develop the computer program as a ‘black box’ but as a transparent modular system that could be gradually enhanced as new knowledge becomes available. In the past solutions to scientific and engineering problems have been programmed in computer languages such as Fortran and C. A higher-level programming environment has been created by MATLAB® (Chapman, 1999). The matrix stiffness method is appropriate for solving statically indeterminate structures. The solution process involves the solution of simultaneous equations. MATLAB contains a comprehensive library of routines for, amongst others, matrix operations that simplifies programming. Input and output are flexible and compatible with Microsoft Excel. Easy-to-use graphics facilities are also available in MATLAB. It was thus decided to develop the program in MATLAB. 7 THE STRUCTURAL ANALYSIS SYSTEM The flow diagram for the computer program is shown in Figure 2. The various steps in the system are now described. Step 1: The following are required: Models to describe the relationship between temperature and the strength and stiffness properties of steel and timber; Charring rates; Formulae to compute the changing cross-sectional properties for variations in dimensions; Formulae to model different semirigid connections; and Resistance models for tension, compression, flexure and beam-columns. Step 2: Similar to other frame analysis programs the following data are required: Nodal coordinates; Member connectivity; Joint types; Member properties (including insulation); Material properties; Support data; and Nodal and member loads. Step 3: An initial temperature distribution is required. The information could be generated by another fire engineering module. Step 4: With data on temperature, insulation, conductivity and member shape the material and member properties are computed. 355

1. Material, member & resistance models

Start

2. Structural description & Load description

3. Initial temperature distribution 15. Increased temperature distribution

4. Determine material & member properties

14. Next iteration

5. First iteration

6. Calculate member & structural stiffness matrices

7. Introduce support conditions & solve the system of equations No 8. Stable?

9. Critical temperature

Stop

Yes

10. Determine displacements, forces and moments

11. Recalculate stiffness properties

No

Yes

12. Convergence?

13. Output

Figure 2. Flow diagram for the structural analysis system.

Step 5: For a given temperature situation a material and geometrical non-linear analysis is done. In the initial stages, at low temperatures, there should be a small difference between the results of a linear or non-linear analysis. But as the deflections increase and the steel begins to yield, larger differences could be expected. The non-linear analysis is done in iterative manner and the process has to be initialised. Step 6: The member stiffness matrices are computed. The s- and c-stability functions are used. For the first iteration the axial forces are taken as zero. For subsequent iterations the axial forces of the previous iteration are employed. The member stiffness matrices are transformed from the 356

member to the structural axes and added into the structural stiffness matrix. The nodal forces due to member loads and member temperature distributions are computed. Step 7: The support conditions are introduced and the system of equations is solved. In the matrix stiffness method the displacements and rotations are the unknowns and are thus solved. Step 8: The determinant of the structural stiffness matrix is computed. If the determinant is negative the structure has become unstable and no solution is possible. The resistances are also checked against the action effects. If the action effect exceeds the (out-of-plane) resistance failure is also considered to have occurred. Step 9: The structure cannot sustain the loads at the specified temperature distribution and the analysis is terminated. Step 10: If the structure is stable the computation process can be continued. The displacements are transformed from the global to the member axes and with the use of the stiffness relationships for each element, the member end forces and moments are computed. Step 11: Given the forces and moments the amount of material or connection non-linearity is established. A recalculation of stiffnesses is then done. Step 12: The results of this iteration are compared to the previous iteration. If the solution process has not converged to within an acceptable degree, a next iteration is required. Initially convergence could be achieved within two or three iterations, but as the structure is getting closer to collapse many more iterations might be required. Step 13: If the solution process has converged to within an acceptable degree, any desired output could be provided. Step 14: If convergence has not been achieved a next iteration is initialized. Step 15: The analysis process is again initialized for a new increase temperature distribution started. 8 CONCLUSION A frame analysis program for a fire engineering design has been outlined. The complexity of the input data is reasonable and user intervention in the analysis process is limited. Techniques and rules outlined in the Eurocodes have been incorporated as far as possible. It has been shown how the linear frame analysis procedures can be modified to do second-order elastic-plastic analyses for steel frames with an increasing load-factor. The matrix stiffness solution process has been modified for semi-rigid connections. This procedure has been adjusted to do second-order material non-linear analyses for steel and timber frames with semi-rigid joints subjected to an increasing temperature. More complex finite element analyses are necessary for fire engineering investigations of structures that are not amenable to be treated as plane frames. Specialized programs are necessary for determining fire resistance design data for composite structural panels or components. However, a large number of conventional structures could be treated as frames. The program proposed could then provide a framework as a useful fire engineering design aid. REFERENCES Al-Jabri, K.S., Burgess, I.W. & Plank, R.J. 2001. The influence of connection characteristics on the behaviour of beams in fire. In A. Zingoni (ed.) Proc. of structural engineering, mechanics and computation 2001, Cape Town, April 2001: 1087–1094. Amsterdam: Elsevier. American Forest & Paper Association 1999. Technical Report 10: Calculating the Fire Resistance of Exposed Wood Members. Washington, DC. Allen, H.G. & Bulson, P.S. 1980. Background to Buckling. Maidenhead: McGraw-Hill. Buchanan, A.H. 2001. Structural design for fire safety. Chichester: Wiley. Chapman, S.J. 1999. MATLAB Programming for Engineers. Hampshire: Thomsons Learning. Coates, R.C., Coutie, M.G. & Kong, F.K. 1988. Structural Analysis. London: Van Nostrand Reinhold.

357

El-Rimawi, J.A., Burgess, I.W. & Plank, R.J. 1999. Studies of the behaviour of steel subframes with semi-rigid connections in fire. J. of Constructional Steel Research 49: 83–98. EN 1990:2002. Eurocode – Basis of structural design. EN 1991-1-2:2002. Eurocode 1: Actions on structures – Part 1-2: General actions – Actions on structures exposed to fire. EN 1993-1-1:2004. Eurocode 3: Design of steel structures – Part 1-1: General rules and rules for buildings. EN 1995-1-1:2004. Eurocode 5: Design of timber structures – Part 1-1: General – General rules and rules for buildings. EN 1995-1-2:2004. Eurocode 5: Design of timber structures – Part 1-2: General – Structural fire design. Fornather, J., Bergmeister, K. & Hartl, H. 2001. Fire behaviour of steel fasteners in wood composites – Experimental analyses. In S.Aicher & H.W. Reinhardt (eds), Joints in timber structures, Stuttgart, September 2001. 619–628. Rilem Publications Proceedings PRO 22. Goetz, K.-H., Hoor, D., Moehler, K. and Natterer, J. 1989. Timber Design & Construction Sourcebook. New York: McGraw-Hill. Hartl, H. 1995a. Behaviour of timber and wood-based materials in fire. (Lecture A13). In STEP Timber Engineering. The Netherlands: Centrum Hout. Hartl, H. 1995b. Fire resistance of timber members (Lecture B17) In STEP Timber Engineering. The Netherlands: Centrum Hout. Hartl, H. 1995c. Fire resistance of joints (Lecture C19) In STEP Timber Engineering. The Netherlands: Centrum Hout. Hibbit, Karlsson & Sorensen, Inc. 1998. ABACUS Finite Element Computer Program. König, J. & Fontana, M. 2001. The performance of timber connections fire – Test results and rules of Eurocode 5. In S. Aicher & H.W. Reinhardt (eds), Joints in timber structures, Stuttgart, September 2001. 619–628. Rilem Publications Proceedings PRO 22. König, J. & Winter, S. 2004. The Eurocode 5 fire part – EN 1995-1-2. In Proceedings of the 8th world conference on timber engineering, Lahti, June 2004. Kruger, T.S. & Van Rensburg, B.W.J. 1993. The matrix stiffness method adapted to incorporate semi-rigid connections. J. of the S A Institution of Civil Engineers, 35(4). Kruger, T.S., Van Rensburg, B.W.J. & Du Plessis, G.M. 1995. Non-linear analysis of structural steel frames. J. of Construction Steel Research, 34: 285–306. Leijten, A.J.M. 1999. An effective timber joint developed at TU-Delft. In Proc. of Pacific Timber Engineering Conference, Rotorua, March 1999. Forest Research Bulletin 212 (3): 253–258. Liu, T.C.H. 1998. Effect of connection flexibility on fire resistance of steel beams. J. of Constructional Steel Research, 45(1): 99–118. Madsen, B. 2000. Behaviour of Timber Connections. Vancouver: Timber Engineering Ltd. Moss, P.J. & Clifton, G.C. 2001. The effect of fire on multi-storey frames. In A. Zingoni (ed.) Proceedings of structural engineering, mechanics and computation 2001, Cape Town, April 2001: 1063–1070. Amsterdam: Elsevier. Newman, G.M., Robinson, J.T. & Bailey, C.G. 2000. Fire safe design: A new approach to multi-storey steel-framed buildings. SCI Publication P288, Ascot: Steel Construction Institute. Östman, B. 2004. National fire regulations limit the use of wood in buildings. In Proceedings of the 8th world conference on timber engineering, Lahti, June 2004. Papadopoulou, A.K., Papaioannou, K. & Papadopoulos, P.G. 2004. Steel structure and fire: Analysis of a steel portal frame. In A. Zingoni (ed.) Proceedings of progress in structural engineering, mechanics and computation 2004, Cape Town, July 2004: 1445–1448. London: Taylor & Francis. Prokon. 2000. Structural Analysis & Design – User’s Guide. London: Prokon Software Consultants. Racher, P. 1995. Moment resisting connections (Lecture C16). In STEP Timber Engineering. The Netherlands: Centrum Hout. Van Rensburg, B.W.J. & Kruger, T.S. 1994. Analysis of steel gable frame structures. In Proc. of the 3rd International Kerensky Conference on Global Trends in Structural Engineering, Singapore, July 1994. Van Rensburg, B.W.J. 2001. Collapse analysis of timber frames with steel joints under increasing temperature. In S. Aicher & H.W. Reinhardt (eds), Joints in timber structures, Stuttgart, September 2001. 629–638. Rilem Publications Proceedings PRO 22. Walton, W.D. & Thomas, P.H. 1995. Estimating temperatures in compartment fires. Chapter 6 in SFPE handbook of fire protection engineering 3-134–3-147. Boston: Society of Fire Protection Engineers.

358

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

On the structural effects of fire following earthquake G. Della Corte, B. Faggiano & F.M. Mazzolani Dept. of Structural Analysis and Design, University “Federico II” of Naples, Naples, Italy

ABSTRACT: Fires following earthquakes are one major threat in seismic regions. This is testified by several historical great earthquakes, which were characterised by the development of large fires: San Francisco 1906 and 1989, Tokyo 1923, Kobe 1995 are representative examples. The obstruction of access roads caused by rubble and collapsed buildings in the streets, the possible difficulties in water supply soon after the earthquakes, as well as the concomitance of multiple fires are some possible factors contributing to such risk. Besides, depending on the extent of earthquakeinduced building damage, the fire resistance rating of the structure could be significantly reduced. To this concern, paper summarises the results of some preliminary numerical investigation, dealing with steel moment-resisting frames. Multiple levels of earthquake intensities, hence earthquakeinduced damage, are considered by scaling acceleration records to increasing values of the elastic pseudo-acceleration. Moreover, two building structural systems are considered, one characterized by perimeter moment-resisting frames only, the other characterized by moment connections in all the bays.

1 INTRODUCTION Risks coming from fires following strong earthquakes are considered very high in the main seismic prone Countries. In fact, losses resulting from fires developing soon after the earthquake may be comparable to those resulting from the shaking (Buchanan, 2001). The risk is certainly increased by the hampering of fire response due to the extreme traffic congestion, collapsed houses and buildings, rubble in the streets, the concomitance of multiple fires and the possible difficulties in water supply soon after the earthquake (EQE, 2000). In addition, another aspect deserves investigation when evaluating the post-earthquake fire safety of buildings: the reduction of fire resistance due to earthquake-induced building damage. In fact, according to current seismic design standards, most buildings are designed allowing some plastic deformation to be experienced by the structure during strong earthquakes. Consequently, fires following earthquakes may find different, more vulnerable, structures, whose fire resistance may be seriously reduced. Hence, the concomitance of the increase in the time needed to firemen for reaching the place of the fire and the decrease of the collapse time of the structure may result in high risk. Moreover, according to the modern performance-based philosophy, the effects of earthquakeinduced damage on fire resistance must be evaluated even if no fire develops immediately after the earthquake. In fact, in case of a structure surviving the earthquake with a certain amount of damage, the overall future building performance must be assessed, in order to consider also the need for a post-earthquake fireproofing intervention. The trend of fire engineering towards a performance-based format (Buchanan, 1994) should consider fire protection of structures located in seismic areas as a fundamental issue. For example, the possibility to either protect only columns of buildings with composite slab/steel beam flooring system (Robinson, 1998) or even to completely unprotect steel buildings by realizing concretefilled steel tubular columns (Wang and Kodur, 2000) should be carefully evaluated in case of buildings located in seismic areas. 359

2 ANALYSIS ASSUMPTIONS 2.1 Structural and damage modelling Modelling the behaviour of buildings subject to fires following earthquakes is a challenging but very difficult task for a structural engineer. In fact, not only knowledge about the mechanical response of the structure to the external action, but also dominance of several interdisciplinary issues, like modelling of seismic and fire actions is required. Grossly, the following general modelling aspects could be identified: a) modelling of the seismic action; b) modelling of the structural response during the earthquake; c) modelling of the fire action; d) modelling of the thermo-mechanical behaviour of the structure subject to fire. The problem cannot be simply solved by facing each aspect separately, because of possible interactions of the different modelling issues. For example, the fire-response of the building must be modelled taking into account the effects of the previous seismic action, which could have produced some plastic deformations in the structure, thus modifying the “initial conditions” for the subsequent fire action. A recent attempt to face the problem in a rational, but simple, way is given in Della Corte et al. (2003), where the following idealisation of the earthquake-induced structural damage is proposed: 1) geometrical damage, which is represented by the residual deformations of the structure after the earthquake; 2) mechanical damage, which is represented by the strength and stiffness reduction produced in those parts of the structure engaged in the plastic range of deformation during the earthquake. Figure 1 efficiently synthesises this scheme, evidencing that the structure after the earthquake could be subjected to significant residual P-Delta effects, which, together with the reduced lateral strength of the frame, could induce an important reduction of the frame fire resistance. The level of geometrical damage is strictly dependent on the level of the mechanical one: it seems reasonable that the strongest the strength degradation is the largest the geometrical damage is expected to be. However, this interrelation is complex and then hard to be predicted a-priori: in both the case of strength and stiffness degradations, the interrelation may be largely dependent on the type of earthquake ground motion for a given structure. At present, the prediction of both strength and stiffness degradation for plastic hinges in MR steel frames is not yet a simple task. Existing models are empirical ones, whose applicability out of the range of model calibration is often unreliable. The advantage of the proposed subdivision between geometrical and mechanical damages lies in the fact that the effects of strength and stiffness degradation can be considered negligible in some cases. For example, in case of well-engineered steel structures, the assumption of non-degrading structural components is realistic in the range of plastic deformation induced by earthquakes at the G + Q ri / hi = i-th inter-storey drift angle

r3 G + Q r2 G + Q r1

h3

h2

h1

Figure 1. Residual P-Delta effects and local plastic deformation due to earthquake.

360

design performance level (i.e. in case of earthquakes having a 10% probability of being exceeded in 50 years, according to modern design codes). However, for structures not adequately designed against earthquakes and/or in case of rare earthquakes (i.e. earthquakes having a 2% probability of being exceeded in 50 years, according to current research trends (FEMA 2001)), the effect of strength and stiffness degradation should be taken into account too (Krawinkler 2000). Obviously, if the structure is sufficiently strong, plastic deformation demand could be relatively small also for the maximum credible earthquake the structure would be subjected to. As it will be shown later, for MR steel frames designed according to Eurocode 8, strength degradation often becomes significant only for very large values of the elastic spectral acceleration, which permits neglecting it in a wide range of practically useful seismic intensities. The geometrical type of damage has been measured using the maximum along the building height of the residual inter-storey drift angles, defined as shown in Figure 1. The software platform OpenSEES (Mazzoni et al. 2003) has been used for modeling the steel frames under earthquake ground motions. To this purpose, the fiber beam-column element has been used, allowing for considering P-Delta effects. According to previous discussion, the effect of mechanical damage has been neglected, assuming the elastic-perfectly plastic hysteresis model. 2.2 Fire modelling After defining the post-earthquake Equilibrium State of the structure, the problems to be faced are the selection of the fire action and the modeling of the fire response of structural components. The fire scenario has been assumed as the most dangerous for the earthquake-damaged structure, which results in applying the fire action at the first level of frames. Temperature has been assumed uniform in the compartment and varying according to the ISO 834 time-temperature curve, which is also provided by Eurocode 1 (ENV 1991-2-2, CEN 1995a). Thermal and mechanical properties of steel have been modeled according to Eurocode 3 (ENV 1993-1-2, CEN 1995b). The numerical code SAFIR (Franssen et al. 2002) has been used for computing fire resistance ratings of structures. Within the assumptions of this finite-elements program, the temperature field is preliminarily determined on the cross-section of each beam-column element, neglecting heat transfer along the element length. Then, the mechanical analysis phase is started considering (large) displacements, non-linearly increasing with temperature (i.e. time) under constant external gravity-related loads. The structure collapse time has been defined as the time at which an instability phenomenon occurs. In this computation of the fire resistance rating, the effect of creep deformations of steel has been considered indirectly only through the use of the conventional stress–strain–temperature relationships suggested by ENV 1993-1-2. Thermal elongation of fibers has been taken into account. On the contrary, the effects of both residual stresses and initial geometrical imperfections have been neglected.

3 SINGLE-STOREY MOMENT RESISTING STEEL FRAMES 3.1 General The influence of the geometrical type of seismic damage on the fire resistance ratings of simple portal frames has been firstly investigated by varying a number of parameters. The parameters were chosen in such a way to fully characterise the mechanical behaviour of the chosen structural system. They are: – – – – –

L/H, which is the ratio between the span of the beam and the height of the storey; Ib /Ic , which is the ratio between the beam and column moment of inertia; Mpb /Mpc , which is the ratio between the beam and column section flexural plastic strength; N/Ncr , which is the ratio between the vertical load and its elastic critical value; δ/H, which is the level of geometrical damage, measured by the storey drift angle. 361

Table 1. Portal frames for parametrical analyses. Structural scheme (a) N/2

N/2

L/H

Ib /Ic

Mpb /Mpc

N/Ncr

δ/H

1

0.63

0.54

0.05–0.30

0.0–0.385

4.50

2.22

0.05–0.30

0.0–0.355

0.53

0.47

0.05–0.30

0.0–0.365

4.72

2.27

0.025-0.30

0.0–0.620

H L (b)

N/2

3

N/2

H L

The structural scheme of the examined frames and the numerical range of the considered parameters is listed in Table 1.

3.2 The effect of geometrical damage Symbols have the meaning explained hereafter: – tf /tf ,0 is the ratio between the fire resistance rating of the structure before the earthquake (i.e. the un-damaged structure) and the fire resistance rating of the structure after the earthquake (i.e. the damaged structure, for a certain level of seismic intensity); – N/Ncr is the rate of vertical loading, given by the ratio between the acting vertical loading and its (elastic) critical value; – (δ/H) is the residual inter-storey drift angle; – (δ/H)max is the maximum value of (δ/H) at room temperature, i.e. the residual drift angle for which the structure is unstable under the given rate of vertical loading, at room temperature. In Figures 2a through 2d the tf /tf ,0 ratio is plotted versus the (δ/H)/(δ/H)max ratio. As it can be expected, by increasing the level of geometrical damage, i.e. the ratio (δ/H)/(δ/H)max , the reduction of fire resistance rating increases, i.e. the ratio tf /tf ,0 decreases. Figure 2 highlights that the residual storey-drift angle very well correlates with the fire resistance rating reduction. Graphs, such as those shown in Figure 2, could allow a ready estimation of the post-earthquake fire resistance rating, once the fire resistance of the initial structure (tf ,0 ) and the level of residual drift angle are known. It is worth noting one main difference of behaviour between the frame in Figure 2a and the frame in Figure 2d. The collapse vertical load of the frame in Figure 2a, at room temperature and without seismic damage, is very close to the Euler lateral-buckling critical load ((Nu /Ncr )0 = 0.99). On the contrary, the collapse vertical load of the frame shown in Figure 2d is appreciably lower than the lateral-buckling critical load ((Nu /Ncr )0 = 0.48). In fact, the frame in Figure 2a is significantly slender than the frame in Figure 2d, as testified by the beam and column sizes. The case in Figure 2d is closer to practical applications than the one in Figure 2d. Therefore it is possible to conclude that the influence of the level of vertical loading on the fire resistance ratings reduction is relatively small. 362

1.4

1.4

t f /t f,0

1.0

(Nu/Ncr)0=0.99

N

N/Ncr=0.05

1.2

N/Ncr=0.10

N IPE140

HEB120

N/Ncr=0.20

0.6

L=H

N/Ncr=0.25 N/Ncr=0.30

0.4

0.2

0.4

0.6 0.8 (d/H)/(d/H)max

1.0

1.2

N/Ncr=0.15 N/Ncr=0.20

HEB120 L=H

0.6

N/Ncr=0.25 N/Ncr=0.30

L/H=3 Ib/Ic=0.53 Mpb/Mpc=0.47 r/r0=1

0.8

1.4

N/Ncr=0.10

N

N

0.4

0.6 0.8 (d/H)/(d/H)max

IPE180

1.2

N/Ncr=0.15 N/Ncr=0.25

0.6

N

(Nu/Ncr)0=0.48

N IPE330

HEB160

0.8

1.2

1.4

N/Ncr=0.05 N/Ncr=0.10 N/Ncr=0.15 N/Ncr=0.20

L=3H

0.6

N/Ncr=0.30

0.4

L/H=1 Ib/Ic=4.72 Mpb/Mpc=2.27 r/r0=1

1.0

N/Ncr=0.20

L=3H

1.0

1.6

N/Ncr=0.05

t f /t f,0

(Nu/Ncr)0=0.92

0.2

(b)

HEB160

t f /t f,0

N/Ncr=0.10

IPE240

0.8

0.0 0.0

1.4

1.4

N/Ncr=0.25 N/Ncr=0.30

0.4

0.2 0.0 0.0

N/Ncr=0.05

N

0.2

(a)

1.0

N

0.4

0.2

1.2

(Nu/Ncr)0=0.76

N/Ncr=0.15

0.8

0.0 0.0

L/H=1 Ib/Ic=4.50 Mpb/Mpc=2.22 r/r0=1

1.0 t f /t f,0

1.2

L/H=1 Ib/Ic=0.63 Mpb/Mpc=0.54 r/r0=1

0.2 0.2

0.4

(c)

0.6 0.8 (d/H)/(d/H)max

1.0

1.2

0.0 0.0

1.4

0.2

0.4

(d)

0.6 0.8 1.0 (d/H)/(d/H)max

1.2

1.4

1.6

Figure 2. Fire resistance ratings reduction for steel portal frames subject to geometrical damage.

(a)

3 x 6m

Earthquake direction

5 x 6m

3 x 6m

Earthquake direction

5 x 6m

(b)

Figure 3. Plan layouts of study cases.

4 MULTI-STOREY MOMENT RESISTING STEEL FRAMES 4.1 Study cases Two different plan layouts have been considered in the study (Figure 3): a perimeter frame system (Figure 3a) and a spatial frame system (Figure 3b). Ten-storey frames, which were extracted from both the perimeter and the spatial frame systems, have been analysed. Two different design strategies have been used for sizing the selected frames: 1) according only to the ultimate limit state (ULS) requirement of ENV 1998 (CEN 2000); 2) according to both the serviceability (SLS) and the ultimate limit state requirements of the same code. Figure 4 illustrates the elevation of the examined frames, with reference to the perimeter and the spatial system, respectively. In particular, Figure 4a and 4c show the elevation of frames designed considering only the ultimate limit state requirements, while Figure 4b and 4d refer to frames designed considering also the serviceability requirements. As it can be seen in the 363

IPE300

IPE400

IPE550

IPE300

IPE400

IPE750x147

IPE360

IPE500

3x6m

HE340M

(b)

4.0 m

3x6m

IPE750x173

IPE400

4.0 m

HE800M

IPE750x173

IPE360 IPE400

IPE400

(c)

3x6m

IPE500 IPE550 IPE550

IPE400

HE450M

IPE750x173 IPE750x173

IPE550 IPE550

IPE750x147

4.0 m

HE800B

IPE500 IPE550

HE450B

IPE500

IPE500

9 x 3.5 m

IPE360

HE340B

IPE750x147 9 x 3.5 m

IPE500

9 x 3.5 m

IPE400

HE450A

IPE550

HE340A

IPE400

IPE400

HE800A

IPE300

9 x 3.5 m

HE450A HE450B HE450M

S-SLS frame

IPE550

IPE550

(a)

S-ULS frame

IPE400

IPE550

(d)

3x6m

IPE550 4.0 m

P-SLS frame

P-ULS frame

Figure 4. Analysed frames.

Figures, the following acronyms are adopted: P-ULS and P-SLS indicate either the perimeter (P) frames designed considering only the ultimate limit state (ULS) or adding also satisfaction of the serviceability limit state (SLS), respectively; S-ULS and S-SLS indicate the frames extracted from the spatial (S) system with meanings analogous to the previous ones for the remaining part of the acronyms. Selected earthquake ground motions have been scaled up to increasing values of the peak ground acceleration (PGA) in order to increase damage to the frame by increasing the first-mode 5%damped elastic spectral acceleration (Sa,e ). The relevant analysis methodology (FEMA 2001) is quite well established. After getting the residual values of inter-storey drift angles, the structure has been subjected to the standard ISO 834 (see Section 1.2) fire for computing the fire resistance rating. 4.2 Analysis of results The main aim of this Section is to analyse the post-earthquake fire resistance of multi-storey moment-resisting (MR) steel frames. Differently from the analysis of the single-storey frames, where residual inter-storey drifts were generated artificially in order to conduct a parametric analysis, the post-earthquake geometry of the multi-storey frames has been generated using real ground acceleration records. This allows the relationship between the level of seismic intensity and the level of fire resistance rating reduction to be directly assessed. Symbols have the meaning explained hereafter: – Sa,e is the first-mode 5%-damped elastic spectral pseudo-acceleration; – Sa,e,d is the design value of Sa,e , corresponding to a 475 years return period and computed according to Eurocode 8. Post-earthquake fire resistance ratings have been normalized by means of the pre-earthquake values (tf /tf ,0 , see Section 2.2), while spectral accelerations have been normalized by means of their design values (Sa,e /Sa,e,d ). Figures 5a through 5d illustrate this normalized fire resistance rating reduction obtained for the four examined cases of MR steel frames, as a function of the normalized spectral acceleration, for a number of acceleration records. As expected, the fire resistance reduction is larger for the perimeter systems than for the spatial ones. Moreover, the effect of the design criterion appears to be quite important, with a stronger reduction for the ULS frames than for the SLS ones. For example, at Sa,e /Sa,e,d = 1 the minimum value of the ratio tf /tf ,0 , deduced from Figures 5a through 5d, is 0.90, 0.97, 0.98, 0.99 for the P-ULS, P-SLS, S-ULS and S-SLS frames, respectively. At Sa,e /Sa,e,d = 2 the analogous minimum values are 0.80, 0.94, 0.96 and 0.97. Assuming that a 10% reduction of fire resistance rating is negligible, it can be concluded that in all the examined cases the effect of 364

(a) (P-ULS)

(b) (P-SLS)

(c) (S-ULS)

(d ) (S-SLS)

Figure 5. Fire resistance rating reductions of study MR steel frames subjected to earthquakes.

earthquake-induced structural damage on fire resistance ratings can be neglected under the design earthquake intensity (Sa,e /Sa,e,d = 1). For rare earthquakes (Sa,e /Sa,e,d > 1) the effect of earthquakeinduced damages could be significant, depending on the design criterion and the structural system layout. In fact, the reduction still appears to be relatively small in case of spatial frame systems, but it becomes significant in case of perimeter frames designed neglecting the serviceability requirement.

5 CONCLUDING REMARKS In the current paper some numerical analyses devoted to investigate the effects of structural earthquake-induced damage on the fire resistance of MR steel frames have been presented. A key aspect of the study has been the interpretation of the earthquake-induced damage, which has been done by means of a simple modelling scheme. In particular, structural damage has been schematised as the combination of two damage types: a “geometrical damage”, which consists of the residual deformation of the structure, and a “mechanical damage”, which consists of the reduction of the main mechanical properties of the structural components (stiffness and strength degradation). This schematisation allows for a rational evaluation of the mechanical state of the structure after the earthquake and of its mechanical behaviour under external actions succeeding the earthquake. In addition, it is a modelling very useful approach for parametrical analyses. The study has been subdivided into two phases: a parametrical analysis of simple portal frames and the evaluation of the fire resistance rating reduction of moment-resisting multi-storey frames extracted from civil ordinary buildings. The analysis of the portal frames should be considered a pilot study. It allows the parameters potentially affecting the problem to be focused on and the range of residual inter-storey drift angles 365

for which the problem is significant to be determined. The main outcome of the study is that the effect of the geometrical type of damage is quite insensitive to the level of vertical loads, when the latter corresponds to normal values. Simple abaci could be developed, which relate the fire resistance rating reduction to the residual inter-storey drift angle. Interesting conclusions can also be drawn from the investigation on the multi-storey frames. As expected, the seismic design strategy demonstrates to significantly affect the frame post-earthquake fire performance. In fact, for the frame designed considering only the ultimate limit state requirement (ULS), the fire resistance reduction has been found to be important, so that a high fire risk derives from earthquake-induced structural damage. On the contrary, in case of structures designed considering also the serviceability seismic design requirement (SLS), the fire resistance reduction is relatively smaller. For such a frame, at the design seismic intensity level, the fire resistance reduction is usually less than 10% of the initial value. However, the fire resistance rating reduction usually becomes non-negligible for very rare earthquakes, i.e. earthquakes having a mean return period larger than 475 years. In view of the development of a comprehensive methodology of performance-based design of buildings, the fire resistance performance should be taken into account considering also the earthquake-induced damage for those buildings located in seismic areas. This consideration leads to the conclusion that the fire-safety codes should distinguish between structures located in seismic and non-seismic areas, by requiring more stringent fire resistance provisions for those buildings potentially subjected to seismic actions. Clearly, the development of a quantitative proposal for such a distinction requires the development of a more comprehensive numerical simulation than that presented in the current paper. Moreover, the authors are aware of the need to overcome some conceptual and numerical limitations of the simplified models used in this paper, but at the same time they are confident in the proposed approach. Future work could be carried out in order to both refine modelling and increase the number of simulated cases. In particular, further investigations will be developed by using the finite element program ABAQUS, in order to have a more accurate representation of seismic damage within the structure. REFERENCES Anderberg Y. (1988). “Modelling Steel Behaviour”. Fire Safety Journal, 13(1), 17–26. Buchanan A.H. (2001). Structural Design for Fire Safety. John Wiley & Sons, Chichester, England. Buchanan A.H. (1994). “Fire engineering for a performance based code”. Fire Safety Journal, 23(1), 1–16. CEN (1994). ENV 1998. Eurocode 8: Design provisions for earthquake resistance of structures. CEN (1995). ENV 1991-2-2. Eurocode 1: Basis of design and actions on structures – Part 2-2: Actions CEN (1995). ENV 1993-1-2. Eurocode 3: Design of steel structures – Part 1.2: General rules – Structural fire design. Della Corte G., Landolfo R. (2001). “Post-Earthquake Fire Resistance of Steel Structures”. In Zio E., Demichela M., Piccinini N. (eds), Safety and Reliability, Towards a Safer World (Proceedings of the European Conference on Safety and Reliability – ESREL 2001), Torino: Politecnico di Torino, 2001, 3: 1739–1746. Della Corte G., Mazzolani F.M. (2002). “Seismic Stability of Steel Frames”. In Ivanyi M. (ed.), Proceedings of the International Colloqium on Stability and Ductility of Steel Structures, September, Budapest. EQE International (1995). The January 17, 1995 Kobe Earthquake – An EQE Summary Report. April. Robinson J. (1998). “Fire – A Technical Challenge and a Market Opportunity”. Journal of Constructional Steel Research, 46: 1–3, Paper no. 179. Fajfar P. (2002). “Structural Analysis for Earthquake Engineering – A Breakthrough of Simplified Non-Linear Methods”. In Proceedings of the 12th European Conference on Earthquake Engineering, Keynote Lecture, London, Elsevier Science Ltd, Oxford, UK, CD-ROM. Feasey R., Buchanan A. (2002). “Post-Flashover Fires for Structural Design”. Fire Safety Journal, 37, 83–105. FEMA 350 (2001). Seismic Design Criteria for New Moment-Resisting Steel Frame Construction. Federal Emergency Management Agency. Franssen J.M., Kodur V.K.R., Mason J. (2002). User’s manual for SAFIR 2001 free – A computer program for analysis of structures submitted to the fire. University of Liege, Department: Structures du Génie Civil, Service: Ponts et Charpentes.

366

Hamburger R.O., Foutch D.A., Cornell C.A. (2000). “Performance Basis of Guidelines for Evaluation, Upgrade and Design of Moment-Resisting Steel Frames” In Proceedings of the 12th World Conference on Earthquake Engineering, Auckland, New Zealand, Paper No. 2543, CD-ROM. Krawinkler H. (2000). “System Performance of Steel Moment Resisting Frame Structures”, In Proceedings of the 12thWorld Conference on Earthquake Engineering, Auckland, New Zealand, Paper No. 2545, CD-ROM. Mazzoni S., McKenna F., Scott M., Fenves G.L., Jeremic B. (2003). Command Language Manual – Open System for Earthquake Engineering Simulation (OpenSees). Trifunac M.D., Brady A.G. (1975). “A Study on the Duration of Earthquake Ground Motion”. Bull. seism. soc. Am., 65, 581–626. Wang Y.C., Kodur V.K.R. (2000). “Research Toward Use of Unprotected Steel Structures”. ASCE, Journal of Structural Engineering, Vol. 126, No.12, December, 1442–1450. Wilson E.L. (1998). Three Dimensional Static and Dynamic Analysis of Structures. Computers & Structures Inc., Berkeley, California, USA.

367

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Soil–structure interaction in case of exceptional mining and flood actions ˇ Radim Cajka Faculty of Civil Engineering, VSB-Technical University of Ostrava, Czech Republic

ABSTRACT: This paper provides an example of a conceptual proposal and static solution to reinforced concrete frame construction with cellular raft, numerical modelling of interaction between the carrying structure and subsoil with mining influence, calculation of parameters of friction of soil environment under deformation, and solution to the mining influence or flooding on a reinforced concrete carrying structure by means of beam and wall/plate (shell) FEM elements. 1 INTRODUCTION Mining influences represent extreme distortion load for building structures. The nature of the mining influence is considerably different from typical designing situations. This paper gives general characteristics of parameters of the mining influence and restricts accuracy of calculation ˇ on the basis of simplified analytical methods pursuant to the standard CSN 73 0039 Designing of Constructions on Undermined Territories [3], and of possibilities of numerical solutions to spatial state of stress and distortion by Finite Element Method (“FEM”), and particularities of entry of input data (resulting from the distortion load and undermined subsoil) in contrast with typical designing situations. When rheological sliding joints are used, the friction between the foundation and subsoil under deformation can considerably decrease, reducing in turn tension forces and quantity of reinforcement. FEM provides further possibilities for interaction tasks dealing with spatial foundations and subsoil with the mining influence. 2 PARAMETERS OF DEFORMATION OF LANDSCAPE IN CONSEQUENCE OF UNDERMINING When sufficiently wide working (working face) is used to work out a seam, the geostatic or tectonic, state of stress changes in the neighbouring solid rock mass. The changes result in deformation and shift of rocks from the hanging layer down to the stopped-out working [9], [10]. In case of long mine workings, such as side drifts or drift ways, effects on the surface of land will not be, from a not-so-big depth, actually evident – because of the arching of rocks. In case of a continuous area mine working, a subsidence trough will appear after certain time – depending on the depth of working, geological structure of the hanging layer, seam thickness, and method of mining. The depth and floor dimensions of the subsidence trough depend especially on the depth (h), seam thickness (m), and open mining angle of draw (µ) being about 65 degrees in the Ostrava–Karviná Collieries. The volume of the subsistence trough depends also on the method of mining which is expressed by the mine working coefficient (a) (in case of caving, the ranges between 0.8 and 0.9). The depth (s) of the subsistence trough is the bigger the thicker is the thickness of works under mining and the thinner is the thickness of the seam under the surface. In the mining industry, the intensity of deformation of the landscape in edges of the subsidence trough is expressed in following geometrical quantities [3]: – subsidence – s [mm], – horizontal shift – v [mm], 369

Table 1. Groups of sites on undermined territories. Landscape deformation parameter

Line

Group of sites

Horizontal relative deformation ε

Radius of curve (R) in km

Declination (i) in rad

1 2 3 4 5

I II III IV V

ε > 7 · 10−3 7 · 10−3 ≥ ε > 5 · 10−3 5 · 10−3 ≥ ε > 3 · 10−3 3 · 10−3 ≥ ε > 10−3 10−3 and less

R<3 3≤R<7 7 ≤ R < 12 12 ≤ R < 20 20 and more

i > 10 · 10−3 10 · 10−3 ≥ i > 8 · 10−3 8 · 10−3 ≥ i > 5 · 10−3 5 · 10−3 ≥ i > 2 · 10−3 2 · 10−3 and less

– declination – i [rad], – radius of curve – R [km] – horizontal relative deformation – ε [-]. The declination (i) and horizontal shift (v) reach the maximum values in the inflection point of the sagline (s) above the edge of the working. The horizontal relative deformation (ε) and the curvature of landscape (ρ = 1/R) reach the maximum values at the distance of about +0.4 r from the edge of the working. Those horizontal relation deformations of the landscape (ε), which are most dangerous for items above the surface, have positive values above the inflection point of the sagline (extension of landscape) and negative values under the inflection point of the sagline (compression of landscape). Depending on expected intensity of deformation of the landscape caused by deep mine working, ˇ sites are divided into groups I. through V, pursuant to Table 1 in CSN 73 0039 [3]. 3 DEFORMATION LOAD Since it would be rather complicated to enter forced deformation of the subsoil into the calculation model, it is advisable to use an opposite task. The subsoil remains non-deformed, and the forced deformation corresponding to effects of the undermining is entered as the load of the reinforced concrete structure. Then, suitable software running on the basis of FEM [11] can be used to investigate into the strain and deformation of the structure. For solution of interaction, the surface model of subsoil, at least, is employed in the software. It is however essential to keep in mind that the resulting forces and deformations have reverse signs. The key issue is suitable expression of friction forces between the subsoil under deformation and foundation structure [15], [16]. 3.1 Landscape declination (i) For typical structures, the declination of the structure has generally no influence on the state of stress. In most case, it is not necessary to take the declination into account. In case of tall and lean structures, the load can be divided into vertical and horizontal components. The declination of the landscape can be also of certain influence on stability of the structure. 3.2 Curvature of landscape (R) The curvature of the landscape, this means the curvilinear distortion (1/R), can be entered as deformation load. The curvature of the landscape can (and often is) different for x and y axes. This however cannot be solved by most FEM applications.

370

Figure 1. Distribution of shearing stress and normal forces.

In case of foundation slabs, it is necessary to enter resistance of ground environment (C1z ) in order to describe internal forces. This parameter can be constant or, preferably variable [11], [13], [16]. Then, several iteration steps are usually needed to calculate the result. 3.3 Relative deformation of landscape (ε) Influence of the horizontal relative deformation of the landscape can be directly entered as the deformation load by means of the relative deformation of the structure. Figure 1 shows the distribution of shearing stress and normal forces. The friction forces between the foundation structure and subsoil subject to the horizontal deformation can be characterised by friction parameters: C1x , and C1y [11], [15]. 4 INFLUENCE OF CONCEPT OF FOUNDATION STRUCTURE ON STATE OF STRESS OF FOUNDATION STRUCTURE Composition of subsoil layers and, in turn, the distribution of friction forces between the foundation and subsoil is also influenced by the method of foundation, fixing of the slab in the foundation base and water-proofing. In order to eliminate possible future influence of mining activities or floods on the building work under study, it is essential to increase spatial rigidity of that work. Pursuant to frequent ˇ recommendations and comments to CSN 73 0039 [3], it appears that designing of a solid structural system is especially suitable for: – building works which are sensitive to changes of the geometrical shape of the structure in consequence of curvature and horizontal relative deformation, and such changes cannot be suitable rectified later on or step by step. – building works the structure of which can better transfer forces than deformation resulting from the undermining where “release” of such structures would be more complicated that dimensioning to resist such influence. Typical examples are massive cast-in-place foundation structures, slab structures, masonry construction and some tower structures, cast-in-place or prefabricated carrying structures for big live loads, or closed underground structures of sump, channels, or ducts. 371

4.1 Slab cast on subsoil When water construction concrete is used to cast a slab onto treated subsoil without proofing or if insulation lacks viscous properties or if the foundation slab is insulated with mass causing the concrete to re-crystallise, the friction forces appear directly between the bottom face of the slab (or concrete sub-floor) and subsoil. Details about calculation and distribution of shearing tension in the foundation bases in ˇ dependence on the influence of the relative deformation of the landscape are given in CSN 73 0039 [3]. 4.2 Rheological sliding joint The original purpose of use of viscous creeping of asphalt masses in the sliding joint was to eliminate friction between the foundation and subsoil subject to deformation for buildings on undermined territories [8]. Because of similar nature of the deformation, this method has been successfully applied to other structures subject to concrete shrinkage and high temperature gradients. With low temperatures, resistance of bitumen against shearing strain is big. With lower temperatures and higher humidity, the concrete shrinkage becomes less intensive. And with higher temperatures, the humidity of environment goes down, increasing in turn the concrete shrinkage [4], [5]. The increase in the concrete shrinkage and arising shearing force in the foundation base is eliminated with the higher temperature and lower resistance of the sliding joint [8]. The maximum average shearing force τxz [kPa] in the sliding joint consisting of melted on asphaltinsulation bands (NAIP) can be determined (in line with measurements and recommendations [7], [9] ) from the following linear relation:

where T [◦ C] is deviation from the reference temperature

and vux is speed of viscous shift [m · s−1 ] in the particular point of the contact joint in the distance x from the centre of gravity of horizontal shearing forces. Reduction of the friction forces by means of the rheological sliding joints is possible thanks to relatively slow rate of undermining deformation of the landscape. If the foundation slab has a flat bottom face, it is possible to use a sliding joint from melted on asphalt insulation bands (NAIP). Rheological properties of these bands restrict considerably transfer of undermining deformation on the carrying structure. 4.3 Rehabilitation and reconstruction of constructions In reinforced concrete structures which have been reconstructed or rehabilitated the torsion load of the structure caused by uneven setting down is frequently so adverse that it is impossible to employ typical rehabilitation means to achieve the required rigidity. During the twisting, the walls are subject to shear. Result is typical vertical cracking along the wall thickness in regular distances. Resistance of existing exterior walls to shear force effects can be increased by additional pre-tension of walls and foundations, and grouting of the cracking walls [12], [14]. Additional horizontal reinforcement of the masonry or reinforced concrete structure is a typical method used to protect the structure against effects of the horizontal deformation or landscape curvature. For this purpose, it is possible to use closed reinforced concrete bracing, iron ties or prestressing cables and strands [9], [10]. 372

Figure 2. FEM calculation model consisting of planar wall/plate and beam elements.

5 FEM SPATIAL MODEL The most efficient foundation for basements, which restricts undermining effects on the upper structure, seems to be closed cellular raft with reinforcing ribs. Because it is required to have a plane foundation base, it is essential to place the ribs in the upper part of the slab. Consequently, the foundation base is plane and it is possible to coat it with a sliding layer with asphalt insulation bands. Three different approaches can be used when modelling the state of strain and deformation of the foundation structure: – solution to the foundation slab and, possibly, wall footing – solution to the foundation slab, incl. basement walls and floor slabs above the basement – solution to the foundation space structure, incl. the carrying structure of the upper building, see Figure 2 It is evident that if the foundation slab only is modelled, the model will not show the undermining effects on the upper building. The easiest interaction solution for the deformation subsoil and ˇ foundation can be carried out pursuant to CSN 73 0039 [3] without employing FEM software. If other methods are used (this means, if the interaction of the foundation space structure, including the whole upper building, if any, is solved) the result is more detailed distribution of internal forces caused by the undermining effects on the whole carrying structure. The behaviour and distribution of tensile forces in the slab is given in Figure 4. For this solution, however, FEM and interaction task between the foundation and subsoil (including the undermining effects) are necessary. The spatial model of the structure consists of FEM wall/plate and beam elements. The soil environment is represented by the surface model with parameters C1x , C1y and C1z [11]. 5.1 Constant distribution of friction parameter C1x The soil environment subject to horizontal deformation is characterised by friction parameters: C1x ˇ and C1y . On condition that the maximum values of tensile forces calculated pursuant to CSN 73 0039 [3] for equivalent section of the foundation are same these parameters can be determined [15] as follows:

373

Figure 3. Differential element of balance.

nxD [kN/m] 704.43 640.39 576.35 M L 512.31 K 448.27 J 384.24 H 320.20 G 256.16 E 192.12 D 128.08 C 64.04 B 0.00 A -55.31 min -110.63 max N

B B

A A A

AB B C

D

C E G HH EAC DG B

AB B CD E C

C

AB B D

C

B

C C

D

C

D

C

B C

B C

B B D E E D G D G D G H J C G E E C D D H E J K B G H J G D D K K JL L D D D JH G L KL EGJ L M M ML L K L J J G EDD D H L D G G H M N G K L N K N E H N M N MLL L K N N JJ N L L M M N J K K M N M G H JJ L N E JK L L M N N CE EGHGEG J J K L N L M B C M L E EGH L H B C L J K B DC EG G BDE E

E

B

G H

Figure 4. Distribution of normal forces in slab and walls in reinforced concrete space structure.

where Fx , Fy are horizontal forces, if any, Ec is modulus of elasticity of concrete, Acx , Acy are cross-section areas of the structure, Lx , Ly are lengths of an equivalent rod, and εx,max , εy,max are values of extreme deformation of landscape caused by the undermining in x and/or y axes. The similar method can be used to represent the deformation load caused by changes of the temperature, creep or shrinkage of concrete If it is assumed that the friction parameters C1x and C1y are constant for the whole structure, the ˇ distribution of tensile forces is not similar as if the method pursuant to CSN 73 0039 [3] is used. For typical cases, however, this method is sufficient because the maximum values of the friction parameters are not exceeded. If the state of stress should be analysed in detail, it is necessary to consider different values for the friction parameters C1x and C1y . To calculate the friction parameters now, simple analytical expression is not enough, since it is essential to make numerical calculation considering the balance of forces per element. 374

5.2 Non-liner distribution of friction parameters C1x If the resultant force of the normal forces is expressed as a one-dimensional (bar) isoparametric element with influence of the subsoil friction, the equation will be:

C1x can be expressed as follows:

The resultant force for the general non-lineal distribution of the shear force on an element can be expressed by the following equation:

For the linear and constant distribution of the shear stress, accurate integration can be carried out, using, for instance, a simple rectangle rule:

In order to calculate unknown equivalent parameters of the shear resistance of the subsoil (the friction in the foundation base between the foundation and sub-soil, influence of the sliding joint), it is possible to use another condition expressing the balance of forces in the element:

This condition can be substituted in the iteration calculation of the friction parameter C1x . The equation for the j + 1 iteration is:

If the rod is approximated using one element only or if an exact differential equation is solved analytically, the constant value of C1x can be used for the whole rod element. 6 EXAMPLE OF SOLUTION The reinforced structure consists of slabs, walls, columns, and reinforcing ribs. It is placed on the undermined territory classified as Group III site. Below are deformation parameters:

375

The thickness of the lower foundation slab is 300 mm. Dimensions of the reinforcing ribs are 400/900 mm. The reinforcing ribs increase the rigidity of the structure, enabling the load from the internal columns of the profile 400/400 mm to be transferred without any problems on the slab. The reinforced concrete space structure consists of planar wall/plate elements modelling the foundation structure, basement walls and floor slabs in individual floors, incl. staircases. See Fig. 2. The beam elements enable the stress of the structure to be analysed in the columns, reinforcing ribs, slabs, and foundation slab. For the calculation of the whole slab, the friction parameters C1x and C1y were regarded as constant:

Figure 4 shows resulting tensile forces and their distribution throughout the foundation structure, this means not only in the lower slab but also in side walls and upper floor slabs. This constructions resists well rock bumps [10] as well as influence of floods [7].

ACKNOWLEDGEMENT This paper has been prepared thanks to the support provided within the grant project CEZ ˇ 105/04/1424 and grant provided by the City of Ostrava in 2004. J17/98271200005, GACR

REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16.

ˇ CSN 73 1001 Foundations of Buildings. Foundation Ground in Shallow Foundations, 1987 (in Czech) ˇ CSN 73 1201 Foundations of Concrete Structures. 1986, a-89, 2-91 (in Czech) ˇ CSN 73 0039 Designing of Constructions on Undermined Territories. 1989, a-91 (in Czech) ˇ CSN P ENV 1992-1-1 (73 1201) Foundations of Concrete Structures. Part 1-1: General Rules and Rules for ˇ Structural Engineering CNI, December 1994 (in Czech) ˇ ˇ CSN P ENV 1992-3 Designing of Concrete Structures. Part 3: Concrete Foundations. CNI, February 2000 (in Czech) ˇ ˇ CSN P ENV 1997-1 (73 1000) Designing of Geotechnical Constructions. Part 1: General Rules. CNI Praha, 08/1996 (in Czech) Bradáˇc, J. et al: Regulations for Buildings in Flooded Areas. Draft, TU Ostrava, December 1999 (in Czech) Balcárek,V.-Bradáˇc, J.: Utilisation of Asphalt Insulation Bands for Sliding Joints in Structures on Undermined Territories. Building Construction 2/1982, p. 63-69. (in Czech) ˇ Bradáˇc, J. et al.: Designing of Constructions on Undermined Territories. Comments to CSN 730039, 1991 (in Czech) Bradáˇc, J.: Effects of Undermining and Protection of Constructions. Part I. and II. EXPERT Ostrava, 1996, 1998 (in Czech) NEXIS 32 – Set of Programmes for Designing of Beam and Wall/Plate Structures, FEM consulting Ltd., Brno, Czech Republic ˇ Cajka, R. Strengthening of Historical Structures on Flooded and Undermined Territory, International Geotechnical Conference Reconstruction of Historical Cities and Geotechnical Engineering, Saint Petersburg, 17–19 September 2003, Russian, ISBN 5-93093-204-2 ˇ Cajka, R. Numerical Analysis of Contact Pressure under Shallow Foundation. International Symposium on Shallow Foundations FONDSUP 2003, 5–7. November 2003, Paris, France, ISBN 2-7208-0355-3 ˇ Cajka, R. Lifetime Enhancement of Historical Structures on Flooded and Undermined Territory. Integrated Lifetime Engineering of Buildings and Civil Infrastructures, 2nd International Symposium ILCDES 2003, December 1–3, 2003, Kuopio, Finland, ISSN 0356-9403, ISBN 951-758-436-9 ˇ Cajka, R.: Influence of Friction in Subsoil on Stress of Foundation Structures. Eighth Pan American Congress of Applied Mechanics, PACAM VIII, January 5–9, 2004, Havana, Cuba, ISBN 959-7056-20-8 ˇ Cajka, R. Contact Subsoil FEM Element for Soil – Structure Interaction. The Second International Conference on Structural Engineering, Mechanics and Computation SEMC 2004, 5–7. July 2004, Cape Town, South Africa, ISBN 90 5809 568 1, ISBN CD-ROM 90 5809 698 X

376

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Gas explosion effects on buildings C. Bob & C. Badea “Politehnica” University of Timisoara, Romania

ABSTRACT: The paper deals with efficient solutions for rehabilitation of a block flat affected by gas explosion in December 2002. The first part of the paper is devoted to general aspects of characteristics of gas explosion. The second part presents a case study analysis and rehabilitation of a block flats. The damages due to the explosion are presented for the structural elements: those in direct contact with blast pressure and the surrounding structural members. Structural analysis comprises two important aspects: the behavior of the structural members subjected to direct blast pressure and the analysis of the state stresses into entire building. The rehabilitation solutions were chosen in accordance with the actual stage of the building.

1 CHARACTERISTICS OF GAS EXPLOSION 1.1 Introduction Any explosive detonation in a gaseous medium gives a sudden pressure increase in that medium which is characterized by an almost instantaneous rise from the ambient pressure to a peak incident pressure Pso. This shock waves are shock front travels radially from the burst point, typically at several times the speed of sound (340 m/second); gas molecules moves more slowly, at about the speed of sound, but both its temperature and pressure drop with the volume of ambient air into which it expands. This latter pressure is associated with a pressure formed by the shock waves which is known as „dynamic pressure”. As the shock front expands into open medium, the peak incident pressure at the front decreases and the duration of the pressure increase (Massa 2000). If detonation occurs inside a building and in a confined space, such as inside of a room, the hot gases from the detonation can not expand freely to mix with on ever-increasing volume of cooler air. If the shock wave fell on a rigid surface, a reflected pressure is instantly developed on the surface and the pressure is raised to a value that exceeds the incident pressure: it is reflected pressure Pr. This supplementary pressure Pr is a function of pressure Pso and the angle formed between the rigid surface and the plane of the shock front. The peak pressure associated with the initial shock front will be extremely high and will be amplified by reflections within the structure. In such cases the shock effects from the explosion will produce mechanical damages as break of windows, doors, walls and floors. On the other hand explosion induced fire generally are caused when something within the building is ignited by the explosion (Massa 2000). 1.2 Blast pressure characteristics The peak incident pressure Pso is characterized by almost instantaneous rice and will act on a structure element. The positive phases of the shock waves are extremely short leaving nearly no time for the element to respond before the negative phase begins (3–50 m sec). Blast pressure is characterize by the explosive weight Q, expressed in the equivalent TNT weight, and the radial distance R between explosion center and target. 377

A global parameter, named scaled range Z, is generally used to characterize the blast pressure:

The peak incident pressure Pso depend on scaled distance (Faust 1993). The reflected pressure Pr can be relating to the scaled distance as: Directly relating to the scaled distance by Low, N.Y/Hao, H; 1999 (Low 1999):

The effect of reflection from structural elements is expressed with (Biggs 1964):

where RL = 2–8 for low and big overpressure, respectively. 2 CASE STUDY 2.1 Overview On the beginning of December 2002 an explosion were reported into a block of flats in the town of Timisoara – Romania. This building is 5 stories with 100 bed-sitters and the explosion was located in the second flat at the 4th floor. The cause of explosion was the leaking and accumulation of propane gas from a gas bottle into the flat. When the resident woman start in the middle of the night, the lighting switch, the gas ignited and caused a tremendous explosion. The initial explosion totally destroyed the concrete walls, reinforced concrete floors, windows and door of the flat and caused serious damage of the others structural members. The resident woman dead, after a couple of week, at hospital. 2.2 Building structure The building is located in the periphery of Timisoara and was built in 1976. The building plane dimensions are 43.55 × 14.75 m and it has a sub-basement and five story of 2.72 m high. The structural composition of the building consists of: vertical members built of longitudinal and transverse reinforced (over the borders) concrete walls with width of 30 cm for concrete facing slab and 15 cm of the interior panels; horizontal members of 14 cm width consisting of prefabricated reinforced concrete floors. The reinforcements of prefabricated floors are φ 6 mm at 15 cm over the short span and φ 6 mm at 20 cm over the long span. 2.3 Building damage assessment The damages due to the explosion are as follows: (i) The flat directly affected was completely destroyed – the transversal walls as interior concrete panels were totally destroyed (Figure 1); – the concrete facing slab was moved, by the force of the blast, from the initial position by 55 mm; – reinforced concrete floors, both bottom and upper slab floor, were deformed with a maximum deflection of 280–300 mm; – window glass and the door were broken. 378

Figure 1. View of transversal walls.

(ii) The surrounding structural members were affected more or less seriously: – the concrete facing slabs located over and under the flat directly affected were moved from the initial position by approximately 10 mm; – reinforced concrete floor over the flat placed near the flat with explosion at the 4th storey has had a maximum deflection of 50 mm; – the transversal walls and floors of the flats located surroundings the flat with explosion, from the 1st floor to the 5th floor, on one part of the gangway, were cracked by dynamic effect of the shock front.

2.4 Structural analysis The structural analysis take into account three aspects: the non-destructive testing of main structural members, analysis of the flat’s elements subjected to blast pressure and analysis of the state of stress into hollow structure. The non-destructive testing were performed with test hammer and pulse velocity measurement on the main elements surrounding the flat directly affected (Chge 2003). The concrete class found by using of the combined method (test hammer-pulse velocity) was C12/15 and C 16/20. The structure elements, which have been affected directly by blast pressure and were seriously damaged, are presented. The elements analyzed are: concrete walls of 4.83 × 2.75 × 0.15 m which are assumed as interior panels without reinforcement in the middle of the wall; reinforced concrete floors with the dimensions 4.60 × 3.44 × 0.14 m and with four edges discontinuous; window glass of 1.00 × 0.50 × 0.0015 m with four edges discontinuous. The results concerning the relevant mechanical parameters of the structure members are presented in Table 1. The values of the bending moments, static loads and static defection, were calculated according to well known formulas. The natural period was obtained from formula with the values of static deflections due to gravity load applied in the vibration direction. The ductility factor was calculated as the ratio between ultimate strain of compressed fiber for plain concrete and reinforced concrete respectively and ultimate strain of tensioned fiber. The pulse time of the blast pressure was appreciated at 25 msec as mean value of the positive phase. The peak load Pso was obtained from the correlation ratio of the equivalent static load to peak load with ductility ratio and pulse time to natural period ratio (Kinney 1988). The assumed model of blast pressure and structure elements response is presented in Figure 2. 379

Table 1. Relevant values of the structure members subjected to blast pressure. Structure members

Parameters

Concrete walls

R.C.floorsbottom floor

Window glass

Cracking bending moment mcr, kNm Maximum bending moment, mmax , kNm Maximum static load Pst , kN/m2 Static defection ast , mm Natural period T, msec Ratio pulse time on natural period: td /T Ductility factor ρ Peak load Pso , KN/m2 Dynamic coefficient µ = Pso /Pmax

7.13 7.13 21.00 0.15 24.6 ∼1.0 1.9 30 1.43

5.17 11.88 12.80 1.23 70 0.36 4.67 47.5 3.70

– 2.25 0.81 1.0 63.5 0.4 1.0 1.13 1.39

P, kN/m2 50 47.5

Load

40

Legend:

30

µ=1.43

24.6 20

Assumed blast pressure

µ=3.70

R.C. floor Concrete walls

}

Element response

12.8 10 0

10

20

30

td=25 msec.

t, msec

40

Time

Figure 2. Assumed model of blast pressure and structure element response. Table 2. Efforts in some transversal walls. Initial stage*

After explosion*

Level

N

T

M

N

T

M

1 2 3 4 5

740 576 452 329 164

106 106 93 74 45

848 572 444 321 122

578 414 290 0 0

111 124 132 0 0

694 361 180 0 0

After strengthening* r 0.78 0.72 0.64 0 0

1.05 1.17 1.42 0 0

0.82 0.63 0.40 0 0

N

T

M

896 693 550 407 203

160 151 132 101 59

1189 781 609 433 160

r 1.21 1.20 1.22 1.24 1.24

1.51 1.42 1.42 1.36 1.31

1.40 1.37 1.37 1.35 1.31

* Value in kN and kNm.

The analysis of the hollow structure has been performed according to Romanian norms for seismic action at present-day level at three stages: initial stage-before the explosion, after explosion and finally after the strengthening of the affected members. The dynamic analysis was performed by using the program SAP 2000. Some results of the analysis are presented in Table 2. The data 380

refer to more significant values of the efforts (axial force N, shear force T and bending moment M) in the elements affected by the explosion. The ratio r between efforts after explosion as well as after strengthening to initial effort is also given in the Table 2. From the tables with all data obtained with the FEM analysis of the entire structure some conclusions were pointed: – affected structural elements have lain surroundings the flat with gas explosion: left and right flats from the first to fifth level located on part of the gangways; – the strengthening of the damaged elements have no sensible influence on the efforts due to the supplementary torsion moment; – the explosion did not significantly affect the walls and coupling beams in longitudinal directions. Similar conclusion with those obtained with dynamic analysis by FEM procedure can be obtained from the comparison between the horizontal seismic load, as an equivalent static load with the blast pressure due to explosion. The horizontal equivalent static load was found: Hs,tot = 1680 kN - for entire building; Hs,3 = 420 kN - for 3 flats (6 for each storey) from foundation to the roof. The structure response at blast pressure on the walls (PSo in Table 1) gives the following horizontal load in transversal direction: He = 282 kN From the above data it can be concluded: the explosion has no sensible effect on entire structure since only 20% of the horizontal seismic load in achieved; for 6 flats (3 on each part of the gangway) the horizontal load due to explosion represents 67% of the horizontal seismic load which is a source of structural elements damage.

2.5 Strengthening solutions In accordance with the building damage assessment and structural analysis, the strengthening of the affected zone, located surroundings the flat with gas explosion was performed in 2003. The strengthening solutions have been chosen to obtain technical and economical advantages as: safe behavior under seismic action; slight change of general stiffness of the structures; easy strengthening technology and short period of refurbishment (4 months); small cost of rehabilitation (approximately 175,000 a). The replacements and strengthening have been made on the following structural elements: – new reinforced concrete floors, with the same geometrical and reinforcement characteristics as for the existing members, at levels 3, 4 and 5, in total 5 elements; – new reinforced concrete walls at levels 4 an 5 which represent 4 transversal walls (with the same width as of the corresponding walls with new shirts) and 4 longitudinal – lateral walls; – local strengthening with R.C. shirt (5 cm on each side), columns (20 × 60 cm in the gangway and 20 × 30 at facade) and longitudinal beams at all levels (20 × 30 cm), vertical elements from ground floor to the roof, incorporating the new walls, on the one part of the gangway; – rehabilitation of the cracked elements with CFRP (Sika wrap) for 10 transversal walls and 12 floors (Sika 2003) (Figure 3). The chosen solutions were in accordance with the real stage of the building: the new elements have been used instead of completely destroyed structural members; the shirts and columns were necessary to assure continuity and stability of vertical members with serious damages after explosion; longitudinal beams have the role of connecting the new vertical members; utilization of CFRP for strengthening of slight damaged elements have the advantages of easy technology and a short time of erection. 381

Figure 3. Aspects of strengthening of walls with CFRP.

3 CONCLUSIONS The main ideas which emerge out from this study are summarized below. 1. The peak incident pressure as well as the maximum reflected pressure are characterize by the scaled range parameter which represents the ratio between the explosive weight and the third root of the radial distance between explosion center and target. 2. The effects of explosions on buildings, personal and others are in function of peak incident pressure. 3. The design procedures as wells as the rehabilitation solutions of the buildings subjected to blast loads were adopted from those already used of existing structures in seismic zones. 4. The damages due to the explosion into a block of flats were function the position of the structural elements: those in direct contact with blast pressure were completely destroyed; the surrounding structural members were affected more or less seriously damages being: slight movements, average deflections and cracks with different width. 5. Structural analysis refers at two important aspects: the behavior of the structural members subjected to direct blast pressure and analysis of the state stresses into entire building. An assumed model of blast pressure and of structure elements response is presented. It presents an important tool in judging the parameters of the explosive detonation. From the FEM analysis of the entire structure the conclusions were very similar to the building damage assessment. 6. The strengthening solutions have been selected for obtaining technical and economical advantages: safe behavior under seismic action; slight change of general stiffens of the structure; easy strengthening technology and short period of refurbishment; small cost of rehabilitation. REFERENCES Avram, C. Bob, C. Friedrich R. & Stoian V, Numerical Analysis of Reinforced Concrete Structures, Elsevier, Amsterdam, 1993: 121–148. Amjad, M.A. 1993. Performance of Building Structures affected by Scud Missile Attacks in Riadh. Magazine of Concrete Research, Vol. 45, No. 165, December 1993: 183–190. Biggs, J.M. Introduction to Structural Dynamics. London, Mc.Grow Hill, 1964. Bob, C., Rehabilitation of Existing Structures in Seismic Zones, Progress in Structural Engineering and Materials, 2001: 3: 353–359.

382

Bob, C., Jurca, A., Florea, V., Palade, C. & Mur˘ara¸su, O. Efficient solution for Rehabilitation of a Building affected by Explosion (in Romania), Romania Academy, Timi¸soara, May 2003: 32–41. Bob, C. Evaluation and Rehabilitation of a Building Affected by Gas Explosion, Progress in Structural Engineering and Materials, 2004. Chge, J.K. & Matalanga, N. NDT Application in Structural Integrity Evaluation of Bomb Blast Affected Buildings, SAMCO NEWS, DG Research of the European Commission, Issue 8, January 2003: 1–7. Faust, B. Evaluation of the Residual Load-Bearing Capacity of Civil Structures using Fuzzy – Logic and Decision Analysis, Universita der Bundeswehr, D-85579 Neubiberg, Germany: 3–25. Kinney, G.F. Explosive Shocks in Air. Springer Verlag, 1988. Kirby, R.E. Major Propane Gas Explosion and Fire Perryville, Maryland, Federal Emergency Management Agency, Technical Report Series; Report 053, 1991: 1–14. Low, H.Y. & Hao, H. An Investigation of Dynamic Response of R.C. slab with Stochastic Properties Subjected to Blast Loading. Asia-Pacific Conference on Shock and Impact Loads structures, Singapore, Nov. 1999: 267–272. Massa, J.R. Vulnerability of Buildings to Blast Damage and Blast (2000) – Induced Fire Damage, U.S. Fire Administration, Technical Report Series, The World Trade Center Bombing: Report and Analysis: 148–150. Odello, R.J. Polymer Composite Retrofits Strengthen Concrete Structures, The AMPTIAC Quarterly, Volume 6, Number 4:.25–29. Romania Ministry of Public Works and Teritory Planning. Code for Aseismic Design of Residential Buildings, Agrozootehnical and Industrial Structures, P100-92, English version: 24–52. Sika Carbo Dur, System of consolidation based on carbon fibres (in Romanian), Technical Data Sheet, 1st Edition, 2003: 1–5. U.S. Department of Defense. Effects of Explosions and Permissible Exposure, DOD 6055.9-STD, DOD Ammunition and Explosives Safety Standards:.2.1–2.12.

383

Session 7: Urban design 1

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Current trends in tall building construction S. Kind University of Applied Sciences, Darmstadt, Germany

ABSTRACT: Essential tendencies in the international field of tall building construction are shown. The article gives details on the complexity of a variety of requirements originating from esthetics, functionality, economical feasibility and safety, as well as planning, erection, usage (and reusage) and demolition. On the whole, the tendencies are described as follows: increasing flexibility, filigree full glass facades, integrative planning, improved room climate, modified vibration behaviour, optimised supporting complexes and materials, increased safety requirements. 1 FLEXIBILITY – ERECTION OF FLEXIBLY USABLE FACILITIES Tall buildings are being built for many decades. A life duration of at least 75 years can be taken for granted during the designing phase. Where the facade is concerned, improvements can be required even after a period of 30 years. The indoor technical systems (water, electricity, telephone, EDPcable connections) are totally obsolete after a period of only 10–15 years (some earlier, some later). The user (tenant) changes in even shorter time intervals. In addition, floor covering and painted surfaces and, in some instances, partition walls, floors and ceilings have to be renewed. This calls for flexible structures in order to facilitate mixed usage (hotels, restaurants, offices, apartments and various forms – such as cell office, nomad office, telecommunication work-places, office cluster, satellite locations). Constant change of the users, or even the mode of usage requires a good documentation (digitalisation) of the planning documentation. There is a tendency in the direction of computer-aided facility management. And the life cycle of the facility has greater emphasis during the planning phase than in former times (life cycle management). 2 FILIGREE TRANSPARENT FULL GLASS FACADES Glass facades are being increasingly called for over the entire floor height, meaning, they are not only placed in front of the hole façade. With the planned use of illumination and colours, unmistakable structures can be created which have completely different impressions during the day as well as at night. The planning of such facades requires a considerably higher scale of knowledge because all requirements (of an esthetical, structural–physical and static nature, such as increased heat input, anti-glare for computer equipment, crash safety) must be fulfilled by the glass structure itself. 3 MODIFIED ROOM CLIMATE – REDUCTION OF ENERGY COSTS One of the main objectives will be to improve the ventilation comfort in tall buildings also. This requires individual intervention options where the room climate is concerned. A further objective is to provide for daylight offices as far as possible. Increasingly larger glass facades call for increased requirements with regard to sunray protection and cooling. The anti-glare effect must be guaranteed (particularly in such places also where computer workplaces are involved). In order to have the heat input under full control, the possibilities of passive cooling must be utilised. For this reason, the application of full glass facades calls for 387

greater ceiling masses if these are to be effectively used as buffers. The tall building in Munich of Murphy/Jahn is a good example here. In addition to passive cooling, the possibilities of pre-cooling and pre-heating the air by means of geothermal heat and geo-cooling are being increasingly adopted. Furthermore, there is an evident tendency towards the double façade (ESG solid fixation outside – VSG tiltable inside). 4 INTEGRATIVE PLANNING If requirements for flexibility are to be fulfilled, the coming usage options must be “perceived in advance” during the planning stage. In this particular case, the later usage management must be incorporated in the planning phase. Moreover, the entire lifecycle of the facility: planning – erection – usage – demolition must be given due consideration as far as possible. The supporting structure must also take over partially the tasks of the room climate (in hole facades, relatively massive walls support heat storage – ceilings can be thinned out in the process – in full glass facades, more massive ceilings are required for heat storage) and the façade must also fulfill safety requirements (e.g., anti-crash). Subsequently, the requirements with regard to the specialised knowledge of the architects and engineering planners increase also. The compulsion towards integrative planning will continue to increase. The necessity to reduce water consumption calls for e.g. considerations on the multiple usage of the water, for example for cooling purposes and for WC-flushing. This example also shows the necessity for close cooperation between the planners involved. 5 IMPROVED VIBRATION BEHAVIOUR A decisive criterion for the dimensioning of the supporting structure is the limitation of the horizontal deformations and the horizontal accelerations. The required acceleration limit values are normally upheld when the horizontal deformations do not exceed approx. 1/500 of the building height (USA: approx. 1/400, China approx. 1/850). The dynamic effects originating from earthquake and wind (but also from shocks) can be at least partially counteracted by means of various vibration suppression devices and neutralisers (e.g. liquid traps). Therefore, slimmer structures can be built and unpleasant horizontal accelerations can be significantly reduced. 6 APPLICATION OF OPTIMISED STRUCTURAL MATERIALS AND SUPPORTING STRUCTURES AND STRUCTURAL MATERIALS The proportionate costs for the preliminary structure are on the decline where tall building designs are concerned. They amount to approx. 25% (as also the proportionate costs for facade, facility technical systems and completion). Whereas the facility technical systems and the facility completion change many times during the service life of the edifice, this hardly applies for the preliminary structure. Due to the long-term effects on the overall edifice, planning must be carried out here with particularly great care. The objective here at all times is to ensure flexibility at a later date and to have a largest possible net surface available. In dependence on building outline design, manufacturing costs, desired design height and installation routing, requirements for sound and fire protection as well as the manufacturing process, various ceiling systems with different surface weights are available. The objective is normally to limit the weight for each square meter so that girders/trussing, supports and foundations can also be sized in a more economical and filigree manner. Composite floors and (-beams) in combination 388

Steel cables 6 meter. dia (660-to) steel ball consist of 30 slices of steel Hydraulic dampers

Figure 1. Taipei 101.

Figure 2. Damping system of Taipei 101.

high [m]

600 Taipei 101

500 Sears Tower

Petronas Tower Jin Mao Center

WTC

400

Bank of China Empire State Building John Hancock Center

300

Chrysler Building

Eiffel-Tower

Messeturm

200

100

0 1875

1900

1925

1950

1975

2000

2025 year

Figure 3. Tallest buildings in the world in the 20th century.

with the use of light weight concrete (lc) provide for the lightest surface weight. For the purpose of installation routing, the steel beams are to be provided with large web openings. In addition, slim-floor-ceilings are being increasingly planned. In Germany there is a continued wide-range use of flat ceilings which are (partially-) fixed-positioned in the core and in the hole façade. In recent years the tendency away from the hole facade was clearly evident because, with it, it is not possible to fulfill the increased requirements with regard to transparency. (For safety reasons there could be a partial renewed reflection on the hole facade because it offers redundant supporting 389

a) core

b) frames

c) connected d) tube cores

e) tube in tube

f) bundled tubes

g) outrigger- h) megai) megasystems structures- structurestrusses megacolumns

high [m]

Figure 4. Stiffening systems. 600 Tube in Tube Bundled Tube

Kern+Outrigger Mega Sturcture

Truss and Frame Tube

Braced Tube

500

400

300

200 1960

1970

1980

1990

2000

2010 year

Figure 5. Stiffening system of the tallest buildings of the world.

systems, meaning, if one supporting element fails, the neighbouring supporting elements take over the loads (refer to 6.)). Heavy duty concretes are being increasingly used for the supports. These concretes offer increased supporting capacity and (by way of a modified elasticity module) reduced deformations at the same time. Slimmer cross-sections are possible in this way. For composite supports, various cross-section types are being used in the meantime. In this case also, the tendency is in the direction of the use of high-tensile steels in combination with high-strength concretes. The bearing behaviour of such HH-columns is increased by the use of high-strength steel fibres (improvement of the ductility) and by the use of polypropylene fibres (improvement of fire behaviour). The combined application of various concretes, steel and composite supports in conjunction with various stresses leads to vertical difference deformations which continue to require intensive observation and attention. For the decisive load case “wind”, various stiffening systems are customary for tall buildings. The following three systems have gained acceptance internationally (among all possible variants): a) If the buildings have approximately the same outlines over the height, the “tube in tube” stiffening system is suitable (fig. 6). If the outlines are graduated over the building height, several bundled tubes can be applied. 390

Figure 6. Messeturm, Frankfurt.

Figure 7. Jin Mao Building, Shanghai.

Figure 8. Commerzbank, Frankfurt.

Figure 9. Bank of China, Hongkong.

b) A more free facade arrangement is possible with the use of “outrigger systems (fig. 7).” In this case, core and facade can be coupled on a positive locking basis continually, but mostly discontinually (in the height of the technical floor). Very high forces (particularly shear forces) occur in the zone of the connecting elements. This must be taken into consideration during breakthrough planning. c) Maximum flexibility of the structure is achieved when the loads are discharged by way of mega supports or mega columns (standing in the corner points of the building) (fig. 8). The large supporting widths are tensioned over by means of frameworks or open-web girders. The subsequently resulting high vertical loads overpress (for the greater part) the tensile forces in the supports. The mega supports are being increasingly interconnected shear-strong only in certain levels. The levels lying in between can then be flexibly hung in by means of suspension rods. A belated alteration of the floor height would then be theoretically possible. A considerable number of static problems must be observed with these progressive variants of tall building construction. Other mega structures are shown in fig. 9. Up to a few years ago, the victory march of the mega supports could be forecasted with relative certainty because they form considerable static advantages with a maximum possible degree of 391

flexibility and transparency. Increased safety considerations requiring more redundancy (refer to 6.) allow the expectation of (at least temporarily again) a greater portion of hole facades. There are no clear tendencies evident for the selection of the foundation. Essentially, plate foundations, pile foundations and combined pile-plate-foundations (KPP) are available for selection.

7 IMPROVED SAFETY Edifices are to be dimensioned for realistic strain and stress. Buildings can only be partially designed against extraordinary (terrible) stresses resulting from explosion, various terrorist attacks. It would otherwise mean that bunkers would have to be built instead of light structures. However, the attack on tall buildings represents such a great danger potential that special considerations are imperative. A suitable solution would be to reflect on the choice of redundant designs where, if one supporting element fails, the overall structure does not fail (immediately) as a result, but other supporting elements take over the tasks of the failed component (at least temporarily) (redundancy). This is particularly well possible with hole facades. From failing facade areas, loads can be transferred to the adjacent façade areas. However, the failure of a mega support is a different matter. Further examination is required here in order to combine the requirements for greater transparency with the requirement for increased safety. The mixed usage of tall buildings with business units, restaurants, hotels, offices and apartments as required by urban planners also places high demands on the safety concept. From the aspects of safety, restaurant and hotel usage must be strictly evaluated. REFERENCES Bachmann H., 1995, Erdbebensicherung von Bauwerken, Basel, Birkhäuser-Verlag. (Earthquake safety of edifices, Birkhäuser Publisher, Basel, 1995.) Hausladen G., De Saldanha M., Nowak W., Liedl P., 2003, “Bauklimatik und Energietechnik für hohe Häuser”, Betonkalender 92, Berlin, Verlag Ernst & Sohn. (“Structural Climatics and Energy Technology for Tall Buildings”, Concrete Calendar 92, Ernst & Son, Berlin, 2003.) Kind S., 5–7 October 2003, “Some Trends in Tall Building Construction”, Conference Tall Buildings and Transparency, Stuttgart, Germany. Kind S., September 2004, “Trends in Tall Building Construction”, IABSE-Symposium, Shanghai. König G., Laubach A., “Innovative Entwicklungen von Hochhausbau – Neue Konzepte und Werkstoffe für die tragende Konstruktion”, Trends in Tall Building, Frankfurt, September 2001. (“Inovative developments from Tall Building Construction – new concepts and materials for the supporting structure”, Trends in Tall Building, Frankfurt, 2001.) König G., Liphardt S., 2003; “Hochhäuser aus Stahlbeton”, Betonkalender 92, Berlin; Verlag Ernst & Sohn. (“Tall buildings from steel concrete”, Concrete Calendar 92, Publisher Ernst & Son, Berlin, 2003.) Lange J., Kleinschmitt J., “Stahl im Hochhausbau”, Stahlbetonkalender 4, Verlag Ernst & Sohn, Berlin, (“Steel in Tall Building Construction”, Steel Concrete Calendar, Ernst & Son, Berlin, 2002.) Liphardt S., September 2001, “Tragkonzepte für Hochhäuser”, Trends in Tall Building, Frankfurt. (“Supporting concepts for tall buildings”, Trends in Tall Building, Frankfurt, 2001.) Remmel G., Schnell J., September 2001, “Technischer Fortschritt bei Konstruktionen und Ausführung von Hochhäusern”, Trends in Tall Building, Frankfurt. (“Technical progress for de-signs and erection of tall buildings”, Trends in Tall Building, Frankfurt, 2001.) Robertson L.E., 1988, “Structural systems for the new bank of china building, Hong Kong”, Proceedings of the Fourth International Conference on Tall Buildings. Schreiner H., Nordhues H.-W., 2003, “Fassaden”, Betonkalender 92, Berlin, Verlag Ernst & Sohn. (“Facades”, Concrete Calendar 92, Ernst & Son, Berlin, 2003.)

392

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Aesthetic of historical towns and innovative constructive techniques M. Fumo & M. Naponiello Department of Building Engineering, University of Napoli Federico II

ABSTRACT: The technological innovation in the building domain engrave on the image of the architecture and on the aesthetics of the urban environment. The technological progress involves the employment of new materials and new techniques that can be “hidden” or “visible”. Brilliant example of an intervention of “visible” recovery is represented by the Museum Guggenheim to Bilbao; for this Museum Frank O. Gehry is called, therefore, to plan a form that turns to the symbol of the renewed artistic and architectural reality that Bilbao proposed itself to reach.

1 INTRODUCTION (M.F.) The technological innovation in the building field doesn’t engrave only on the mechanical and functional performances of the single constructive elements end/or of the whole building, but also on the image of the architecture and on the aesthetic aspects of the urban environment. The technological progress involves the employment of new materials and new techniques that can be “hidden” or “visible”: in this last case modern technology is able to set in evident contrast with the traditional technique and to produce buildings in total formal contrast with those of the past. Such contrast is very more evident if the “technologically advanced” building is built in a historical center, or rather in an built homogeneous environment characterized by the employment of sand-gravel mixture materials. Before the advent of the reinforced concrete and to its diffusion as structural material from construction of rapid execution and great flexibility of use, such to produce new building volumes and new expressive forms, the traditional architecture has produced very similar buildings in the volumes, so much to be been able to affirm that the difference of aspect among those was limited to the leaning parts (balconies, mouldings, platforms …) and to those decorative (superficial treatments, frames, plasters, fixtures …). The discrepancy among the “traditional” cities and their outskirtses has gone defining since the planning of these last. Conceived in the second postwar period, generally as negation of the “old” preexisting urban system already hystorical, the peripheral districts were declared the alternative to the historical center for the best hygienic and environmental quality: hygienic services in all the lodgings, openings (windows and balconies) better proportionate with the surface of the inside environments, ampleness of the roads dimensioned by the diffusion of the car and the vehicular traffic, attention to the orientation of the residential buildings in comparison with the sun … From such evolution of the city has sprung that the perception of the contemporary city, in greater part of its outskirts, clearly makes distinguishable from the traditional city: the design of the building volumes, potentially very taller those more recent, the relationship among the full ones and the voids in façade and, lingering over this last that shapes the element of interface with the urban context, the absence of elements of decorative apparatuses that have characterized the architecture up to the advent of the armed concrete. Just the development of the use of this construction material has involved a revolution in the planning and in the design in architecture for the sure extraordinary advantage to conceive a different spaceness thanks to the characteristics of plasticity of the material, has behaved in the time the reduction of the creativeness of the planners (motivated more to do a lot that to do well) and the 393

affirmation of a serial design characterized by the repeatness of the solutions. Also the necessity to reconstruct whole urban districts, following the second world war, has behaved an “economic” employment of the technology of the concrete above all for the buildings of residence, realized with little attention to the aesthetical quality. From the concrete in work to the prefabricated one the aesthetical aspect of the constructions has not certainly earned, neither those to housing use neither those industrial, so much that the most greater part of the sheds of the XIX century, realized in bricks with coverages in iron, were very more parish priests in the decorative aesthetical aspect in comparison to the industrial sheds prefabricated of the second halves the XX sec. These reasonings, for the aesthetical quality of the traditional classical city can conduct to projectural attitudes also exasperated as the proposition of traditional elements recognizable in the contemporary architectures. The traditional design of decorative elements of classical matrix (like mouldings, eardrums, ashlars, columns, pilaster strips) has not only been proposed by the post-modern but also in the “restauration” of anonymous buildings in masonry that have gone enriching themself of these false elements of decorative merit. For this reason assists to the bricks production and of tiles “aged”, of forms in glass-resin for Corinthian capitals to leaves of acanthus, etc. running into the taste and the applications of the public that from some year has increased the question of “ancient.” Also in Italy, where the culture of the maintenance is deeply rooted, the technological innovation has run over wide layers of industrial production to imitate the external aspect of the ancient constructions through productive shrewdness able to furnish a modern industrial product (under the aspect of the most greater quality and durability) that has the appearance to be handicraft and therefore “ancient.” In fact the New Urbanism, considering the traditional city as model of urban spaceness for its equilibrium among the common and private spaces and the presence of strong signs of identity that produce architectural quality and comfort of the fruitoris, curtains to the proposal of the signs and design of the classical and the rhythms that are its own even though with contemporary materials and techniques. (Fig. 1) The general “perception” of urban environment springs from the sum of perceptions connected to the volumeter relationships both between the building and the road and, in the same building, among the parts “floods” and those “empty”, as the characters of homogeneity are fundamental or rather of recurrence of the building elements and the “dissonances” in comparison to a harmony. Such perceptive feelings are also referable to the color of the materials or to their chromatic variety and their ability of reflection of the light: even before to notice the chromatic range, our mind perceives the monotony or multiplicity of the colors and their brightness identifying the chromatic contrasts and the dominant tones. A threat is certainly represented by the slavish proposal of hystorical rules and from a rigid interpretation of the tradition: this last has to be considered a point of departure but not a point of arrival. The knowledge of the materials, of the local resources, of the traditional constructive techniques and of the building wisdom of the people must not be dispersed on the contrary sought after and safeguarded, but the guardianship of the culture is not had to identify with the guardianship of all of its products and the creative and experimental immobilism. The technological innovation and the maintenance of the traditional aesthetical characters are perfectly conjugable, as the actual production shows us: f. ex. the external aspect of a façade with windows in natural wood can be preserved improving the performances of the lockings, replacing it old windows with new fixtures with loom in natural wood to the outside, double-glass for a best thermal isolation and inside loom in pre-painting aluminum for a best maintenance (Fig. 2). The technological progress has involved in the last decades an abandonment of the sensitivity to the natural environment, typical of the constructive tradition of the past centuries, but today we are assisting to an inversion of tendency: the local resources are again considered as the principal environmental component and technological contribution, often sophisticated, cannot leave out of consideration from the problem list of insertion and adaptation both in urban scale that in building scale. 394

Figure 1. Tipiousa, Spetses (Greece), Traditional type of buildings with modern tecnologies, project Porphirios Associates.

Figure 2. Publicity of a fixtures.

The lesson that comes us from the traditional city can be a example for the knowledge of the sites, for the intelligent employment of the local resources and for the constructive wisdom tied up to a reverential and timorous observation of the natural phenomenons, always channeled in the furrow of that we today define environmental “sustainability”. In the following paragraphs introduces it an examination on two types of projectural attitudes when intervenes in the historical centers: the contrast or the mimesis, that we synthetically will define as visible or invisible innovation. The first type of project (visible innovation) is what makes visible the innovation, both in the intervention on an element of the building (for example the consolidation or the substitution of an attic) both in the intervention to urban scale (for example the realization of a new building in a historical center) while the second type (invisible innovation) is that in which the actual intervention, also realized with advanced and innovative technologies, it is not declared and it is not comprehensible to any observer. 2 THE VISIBLE INNOVATION (M.N.) Bright example of an intervention of “visible” recovery it is represented by the Museum Guggenheim to Bilbao, planned by Frank O. Gehry and inaugurated in 1997. Together with the Building of the Congresses and the Music of Federico Soriano and Dolores Palacio, the airport Terminal of Santiago Calatrava, the Subway of Sir Norman Foster, the plan of shores setup of Nervino, for the reconversion of the buildings along the river in structures for services, commerce and finance, the Museum makes part of the plan of revaluation of Bilbao, that has had beginning in 1989 and that sees to operate also other great names what Pelli, Stirling and Wilford. The objective of the plan, founded upon the infrastructural retraining and on the environmental and urban transformation giving ample relief to the human resources, was that to give back life to a city that borns as commercial, therefore place of exchanges and passages, and that it grows during the industrial revolution up to the end of the seventies, when an economic crisis crushes it, making it a city disseminated of industrial fittingses in disuse as furnaces, chemical and iron establishments, shipyards; this has produced a progressive impoverishment and degrades of the whole city. The spirit of the city changes, and in the nineties Frank O. Gehry is called by Guggenheim Foundation, therefore, to plan a form that rappresents in the whole the renewed artistic and architectural reality that Bilbao had reached. And thanks to the use of new technologies that the planner succeeds in his intent. 395

Figure 3. Projects of the museum Guggenheim of Bilbao

The Museum, in fact, for the mathematical and composite complexity given by the intersection of well nineteen volumes and from the use of the titanium, as material of covering not belonging to the building tradition, has required for its planning CATIA, a software among the most modern for the planning and the calculation. Used by the aerospace industries of the Dassault in France for the planning of military airplanes, CATIA has been used by the italian enterprise performer of the works for the realization of the Museum, the Permasteelisa of San Vendemmiano (TV) that can develop the definitive and executive project of an so imposing work not to be able to be realized otherwise (Fig. 3). Such grandeur, that confers great architectural strength to the Museum Guggenheim, succeeds without difficulty in inserting itself in the context, without annihilating it, rather contrarily embellishing it thanks to the sculptural value of the form underlined by the use of the titanium as material of covering. The architect, induced also from the opinion of some artists among which Thomas Krens, decides to plan a building with a well defined character, but he allows the integration of it with the site thanks to the game of volumes and the intersections with the neuralgic points of the city and still thanks to the audacious approach of covering materials that we can define as traditional, as the calcareous stone extracted by the caves of Huèscar of Granada and other “new” materials, as the titanium. Gehry, in fact, succeeds through soft lines in creating continuity among the different levels of the river and the urban center; moreover he places the principal entry of the museum at the end of the street Iparraguirre connecting it so to the boxto, center city, through one of its principal road artery that diagonally cuts the heart of Bilbao (Fig. 4). Moreover a ramp brings to the tower that has the function to integrate the museum with the Bridge of La Salve, one of the principal streets of access in the city, in fact it is center of a series of ramps of staircase that allow a pedestrian connection among the museum and the rest of the city (Fig. 5). The match among “old and new” once more doesn’t happen in traumatic way, in fact the tower opens itself embracing the bridge. A fundamental role in confering heat to such mighty structure is developed by the used materials, beginning from the 27.200 mqs of thick 5 cms plates in calcareous stone of beige-amber color set to covering of the galleries composed by volumes of more regular dimensions, passing to the plates of double thermal crystal, fit to develop the double protective function for the works both from the solar irradiations that from the heat and able to confer the lightness and the enough transparency that requires to the building, used for the glass walls (Fig. 6) and for the skylights and the glass coverages (Fig. 7) up to come the titanium, that submitted to a process of lamination, to polish and galvanic baths, has gotten a silvered color able to make proper the colors that surrounds it, assuming and characterizing all the tones of color of the river Nervion and of the sky. 396

Figure 4. Wiew from the road of the entry of the Museum.

Figure 5. Wiew of the inside of tower.

Figure 6. Glass walls between the volumes covered by titanium.

Figure 7. Glass coverages.

Sixty tons of titanium, 36.000 mqs, 33.000 thick 0.38 mms foils, constitute part of the packet of lute (80 mms mineral wool; barrier to the vapor, plate of galvanized 2 mms steel) anchored to a structure of support constituted by a sweater of tubular with slides to C shaped on the external wrap. The manifold bending of this, have forced the builder enterprise, that wanted also to standardize 397

Figure 8. Views of together of the museum and of the city.

its dimensions ( 1150 × 600 mms), to realize a considerable number of types of foils, around two hundred, and to prepare a further dimensional control after the cut of the foils. To this is added a operative system with topographical coordinates to realize the joints perfectly among the foils. The Museum Guggenheim to Bilbao, for this reason, can be considered a radiant example of recovery through new visible technologies, but not invasive, in fact a building with maximum height of 50 ms and with a surface of 24.000 mqs it perfectly results integrated not creating aesthetical fractures neither with the built landscape neither with that natural (Fig. 8). This has been possible for its elevated mechanical resistance, comparable with that of the steel to the carbon and density equal to halves of the steel. Besides it is also very supple to low temperatures and the spontaneous formation of a film of oxide that protects the material from further aggressions, able to instantly reform it in case of lacerations following a mechanical action, confers to the titanium an elevated resistance to the corrosion both in acid environments that basic and in chlorinated environments. The titanium finds application also in other ways in the field of the recovery (see it: Partenone, Eretteo, Propilei, Church of S. Cristina, inferior Plaza of the Basilica of S. Francesco, etc.) thanks also to the coefficient of thermal expansion comparable to that of the concrete and of the glass, therefore matchable with sand-gravel mixture or ceramic materials. 3 THE INVISIBLE INNOVATION (M.N.) Other topic must be given for Fiber Reinforced Polymer (carbon’s fibers, fibers of glass, fibers of boron, aramidics fibers), that are innovative materials and they constitute a mean to realize an “hidden” intervention of recovery, as it doesn’t bring serious aesthetical mutations of the built object. In fact the FRPs, constituted by two or more phases (the reinforcement, discontinuous element and from the elevated mechanical characteristics and the matrix, generally epoxide resin or polyester, continuous and with function only of support) and products in form of ribbons, fabric, foils, plaques and cylindrical bars, with elevated mechanical characteristics, of resistance to work and to chemical attacks of solvents, acids and bases, easily applicable, and if in form of ribbons or also plotted adaptable also to the most complex forms, compensating the low resistance to hortogonal efforts for instance with multidirectional fibers, they represent a valid alternative to the traditional materials, unlike which allows a type of completely reversing intervention. Such materials, have in particular the characteristic to have little weight and small thickness, which allows to realize interventions of recovery of ruined or damaged structures and of adjustment e/o static improvement without heavily inflicting on the object. In fact they are perfectly calibrated in consideration of the slot picture, they don’t modify the inertia of the masonries under conditions of exercise, but they attend only in presence of particular conditions of load, neither they alter the mass, the volume neither so much the form and the aesthetics of the support also because they can be plastered, painted or covered, so that practically result “invisible”. 398

It succeeds so to preserve the structure and the architectural and historical characters of the built one, that constitute a tangible testimony, rather, enjoyable of the historical and cultural identity of a city. To testimony there are the numerous studies and interventions of recovery realized on structures in armed cement, also prestressed, in steel, in masonry and in wood with the use of the FRPs as for instance the recovery of masonry vaults of the basilica of S. Francesco in Assisi, effected with a multidirectional fabric of aramidic fibers applied with epoxide resin so that preserve the present frescos in the intrados both by drainages that from water of bathing or the recovery of the masonry cross and skew vaults in the Palazzo Comunale in Assisi, where it’s intervened to the extrados applying ribbons of CFRP (carbon’s fibers) with bicomponent resin. A further valid example of materials that help to recover staying “hidden” it’s offered by the new materials for the Daylighting, among which we shortly remember: • The holographic films, transparent self-sticking films endowed with elevated abilities rifrattive and able to diffract the light, partly screening it and partly addressing it in well precise points. In such way, applying the holographic films to the normal glasses of the openings, he is able better to use the heat exchange that happens for irradiance; • The TIMs (transparent insulating materials), transparent solid and not good transmitters of heat also spreading the radiation solar it give direction to the reflection. It’s possible to limit the dispersion of heat because the TIMses decrease the conductance of a window realized with double glasses up to 1.2 W/mq◦ K, lowering it therefore to around the 25% of the middle value. With such materials it can improve therefore the bright performances and to regulate the climatic conditions of an environment avoiding, in some cases, the installation of bulky of system of conditioning that for the more one they deface the aesthetics of the buildings, leaving more than a simple trace on the façades. LITTLE REFERENCES Basso Peressut, L. 1999. Musei architetture 1990–2000. Milano: Federico Motta Editore Perman, H. 1998. Contemporary World Architecture. London:Phaidon Berlage, Bergeijk, 1985. Architettura, urbanistica, estetica. Bologna; Zanichelli Menna, 1968. Profezia di una società estetica : saggio sull’avanguardia artistica e sul movimento dell’architettura moderna. Milano: Lerici Panza, Franzini, 1996. Estetica dell’ architettura. Milano: Guerini Soto, Pons, Cuito, 1998. Guggenheim. Cavallermaggiore: Gribaudo Avorio, A. & Borri, A. & Corradi, M. & Celestini, G. 1999. Miglioramento sismico: sperimentazione e analisi sull’utilizzo dei materiali compositi nel1e costruzioni in muratura, L’Edilizia, n. 9–10 settembre–ottobre 1999, anno XIII, pag. 60–71 Avorio, A. & Borri, A. Il miglioramento sismico delle volte, Recupero e conservazione, n. 42 novembre– dicembre 2001, anno VII, pag. 50–56 Borri, A. 1999. Il consolidamento degli edifici in muratura, L’Edilizia, n. 2 settembre–ottobre 2001, anno XVI, pag. 22–26 Faccio, P. Foraboschi, P. & Siviero, E. 1999. Volte in muratura con rinforzi in FRP L’Edilizia, n. 9–10 settembre–ottobre 1999, anno XIII

399

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

New living concepts for late 19th and early 20th century town housing B. Pahl, T. Hobusch, S. Kruger & C. Pluto Dept. of Planning and Constructional Design, University of Leipzig, Germany

ABSTRACT: The NEWOG project’s main research goal was to develop new concepts capable of combining two factors – property redundancy in and the demand for alternative forms of inner city living. Numerous new ways of rebuilding late 19th and early 20th century town housing were developed based on an analysis of certain selected cases. Priority was given to reducing the number of flats, the integration of sought after, individual living concepts and an investigation of criteria for sustainability. Building work was to be concentrated and concepts applicable elsewhere. Initial variations were valued. Types of building work were compiled in a catalogue, with a detailed description and cost analysis. Within the framework of UTN II, an EU network project, the execution and financing of initial pilot projects are now being prepared. The aim of these pilot projects is the demonstration of new living concepts on an accessible and habitable scale, the exchange of information and multiplication effects.

1 INTRODUCTION The current residential property market in the former eastern German States presents a unique situation. The large number of empty flats in town housing and high rise has led to very low rents, creating advantages for tenants and disadvantages for owners. The various trends in flat-hopping simplify analysis of the market, showing dissatisfaction as well as preferences. Loft conversions, the continual renovation of housing in unpopular locations and of flats with unfavourable floor plans has increased property redundancy by 20%. Leipzig has about 58,000 empty flats [1]. The social structure of whole areas is being shifted. At the same time, certain types of accommodation are in demand but lacking. The spatial layout of old town housing, so important in creating the city’s urbanity, is being thinned out through demolition, itself speeding up the further erosion of certain areas. The drastic population drain in the former eastern German States since 1990 has been intensely researched in recent years and can be traced to three main causes. A continually depressed economic climate has led to a strong migration to western Germany. Five years after the Fall of the Berlin Wall all but 15,000 of 100,000 jobs in industry in Leipzig remained [2]. From 1990, migration to the West set in. From 1993, there was a drain from the city into the hinterland. Since 1998, there has been a tendency to move house within the city. The demographical process of falling birth rates causing a massive reduction in population right across the country will intensify in the next few years. Since 1989, Leipzig’s population has sunk from around 500,000 to 400,000 (registered in the 1989 city borders). The desire to own a single family home led to a sprawl effect, a process still underway today. At the moment this process has stagnated, countered by stabilisation in the city caused by increased consumption of larger flats, new flat ownership by families and moves back from the outskirts. NEWOG research interest has been in the massive property redundancy on one hand and in the search for a sustainable approach coupled with the retention of urban structure on the other. Only some of the available accommodation matches demand in terms of size, standard and floor plan. When evaluated by potential clients, accommodation within the city falls well short of that in the outskirts in terms of location, building type, price and ownership potential. And this even though the 401

Figure 1. Town housing in Leipzig.

Figure 2. Overview of housing categories.

same clients see the outskirts as less than ideal. One of the most frequent reasons given for a move out of the city is the desire for house ownership. This need can however also be fulfilled in the city. In order to improve the image of locations within the city, more positive aspects of city living need to be defined. These must include attractive surroundings, less density and individual living concepts with enough open space. Demographical developments, changes in our social fabric and the pluralism of contemporary lifestyle must lead to new concepts for living. A more individualistic society needs individualistic spaces to retreat to. This requires a multifaceted supply of accommodation. Work on the NEWOG research project has shown that the new living concepts are sufficient to satisfy formulated needs and are being watched with great interest. More often than not, though, the marketability of these concepts has been doubted, as the qualities achieved in executing these projects could not be fully envisaged. On the one hand there is a clear need on the part of investors eg: housing corporations to develop new products and establish new markets. On the other hand we perceived a reluctance due to limited experience with these radically new concepts, which made their development risk in a difficult situation hard to assess. This is where the work within the framework of the UTN II project begins. The UTN II network project aims, with its EU partners and other interested cities, to promote a consistently high EU standard at all levels of urban infrastructure. To achieve this, projects are being implemented which offer model approaches to address various urban problem areas and which can be applied to different cities. For each approach, so-called “guidelines” are being produced to facilitate the exchange of experience and knowledge between the individual partners. They can be made available to initiate action in interested cities and promote international collaberation. The development of pilot projects on the theme of “New forms of living in late 19th and early 20th century buildings” is one of these projects. 2 AIMS One of the main questions which confronted NEWOG was how to sustain the unrenovated old town housing. Our aim was to develop strategies capable of sustaining and stabilising the old town 402

Figure 3. Matrix of variations.

Figure 4. Sections.

housing while taking into account demographic developments, the conditions of the buildings and the requirements of potential users. Using Leipzig as a case study, our intention was to improve living standards while retaining the urban fabric of the neighbourhood block, this through precise interventions reducing the amount of accommodation on offer. Concepts for different types of town housing, maisonettes and whole floor flats for various user groups otherwise uncatered for on the property market were to be developed. Unusual methods were to be implemented (the reversal of floor plans, the creation of new development land, the demolition of parts of buildings) in order to significantly reduce the number of flats available by changing their size and constellation. Personal needs (a flat’s size, standard, orienation and location), forms of ownership, surroundings (quiet and green) and flexible floor plans were to be important considerations, as were urban layout, user access to the direct surroundings, location and the maximising of a building’s potential in economical terms. 3 IMPLEMENTATION A catalogue of aims was compiled as a result of evaluations made from existing research material, from evaluations, from surveys about living requirements (UFZ, IFL …) and by limiting the object of investigation [3] [4] [5]. This included an examination of the old town housing based on data about building condition, amount, location, size, floor plan, occupancy and number of flats. Potential user requirements were compiled. The relation between the desired size of accommodation and financial situation, the type of ownership and standard preferred, the importance of open space and options like open kitchen, bathroom with daylight, separate WC, storage space and garage were collected and analysed in surveys. Next, a limitation to the relevant building categories was set and data concerning building materials and planing parameters was prepared and categorised. Houses with average floor plans located on busy roads were given priority. The initial results concerning the definition of target groups, future inhabitant structures and the selection of building categories and their problem profiles were discussed in a first set of meetings with experts. As a result, the aim to renew threatened old town housing was defined (and termed potential and conceptual development). Floor plan variations were developed for the selected building categories, including a new vertical and horizontal categorisation of flats (reduction of number of flats while maintaining urban layout). This sought to create previously lacking living standards and open spaces for different groups (single parents, young families, partnerships, singles, working people). The creation of variations was defined by combinations of the most influential factors such as organisational form, development potential of the building, building category and 403

Figure 5. Some examples for new floor plan with fewer apartments and fewer square meters in the building, maximizing buildings potential.

Figure 6. Building elements.

building use. An important criteria for the development of each concept remained its exemplary character and its potential for application to other buildings in the same category. These variations were then analysed with respect to the constructional intervention, building method, surroundings and expected costs. On the basis of this analysis, a selection of priority variations was made. Sustainability was given special priority (economical factors, living and design qualities, use of resources) and illustrated in diagrams for each of the concepts. Of the many variations, detached housing, maisonette flats and open plan, whole floor flats constituted important forms of living accommodation. As a next step, ways of reducing construction work while maintaining each plan’s full potential were sought. This involved combining the various plan’s aims with renovation methods (“step by step”, user contribution, extensive renewel of individual sections). On considering all necessary extensive renovation work on all housing, it became clear that a realistic level of user/owner contribution would make up a maximum of 10%. Taking into account the type of client likely to be seeking ownership and the type of constructional work necessary for part renovation, this contribution need not be considered in a cost calculation. The basic principles of span, method and organisation of rooms found in the buildings which were investigated (the RII and RIII category) can be found in other types of housing, making the application of RII and RIII concepts possible elsewhere [6]. Repeated building work such as staircases, ceiling openings, removal of interior walls, facade openings, roof deconstruction, hollowing out and rebuilding, set back facades, balconies and installation of large plant containers were compiled in a catalogue. Each type of work was drawn in detail, with a written construction description and cost estimate. To simplify its application to other building categories, a general cost per square metre register was drawn up. 404

Figure 7. Leipzig Window.

Figure 8. Models of transformation.

A number of variations plan a large, enclosed vitrine-like space fronting the street over one or two floors. This effective measure was given a typology, named “Leipziger Window” and investigated in depth. This enclosed, temporate volume creates a new light, open space for the surrounding flats on the one hand and a physical buffer soaking up noise from the street on the other and thus increasing the quality of life in the house. The physical and energy-related aspects of this intervention were investigated by an expert. As part of an economical study, cost was seen as being more important than the current market situation. The costs of rebuilding a flat in a late 19th or early 20th century town house in the city were compared to the costs of a new single family home on the outskirts. That the town house refit costs 20% more is due to the extra work necessary in reducing the amount of flat space, the precision of the constructional interventions and the larger volumes spread over more floors. These extra costs become relative when seen in relation to the sustainability of the urban structure (necessary demolition costs, under capacity infrastructure, cost of connecting new land in the outskirts to infrastructural networks). Considering the extent of property redundancy, it seems counterproductive to continue with traditional renovation practices. The percentage of empty flats in renovated housing is running at 23% [7]. A maximum of 10% is considered economically acceptable by the housing associations, allowing for house moves and market reserves. Various city redevelopment strategies were considered in relation to the wider urban situation in Leipzig. The almost uncontrolled, punctual demolition of buildings causing an increased erosion of urban structure was termed “Perforation” (A). By creating a “Priority Case” (B), a specific block can be retained while another is demolished. “The Compact City” (C) illustrates a desired shrinking from outside to within, though this strategy is unrealistic due the complexity of the situation, the well advanced renovation programme and ownership structures. NEWOG concentrated on a fourth strategy, an “Evolution” (D) of the existing structure: an adaption of the existing accommodation to 405

Figure 9. Urban development scenarios.

user requirements leads to a significant reduction in the amount of accommodation while retaining the urban structure. Property redundancy should be countered using a combination of targeted demolitions and an evolution of the old town housing. These results and conclusions were discussed with experts in a second meeting. In particular, the changes to government building subsidies currently being debated nationally were discussed intensely and approved. Home ownership should only be subsidised in existing buildings. Grants should only be awarded to areas designated for renovation and to other subsidised areas for measures which support a reduction in the amount of accommodation, dissuade from loft conversion and benefit area and neighbourhood block structures. The retention of the subsidies for protected buildings should be extended to extensive measures intended to sustain urban ensembles. Large scale demolition funding should also be available for partial demolition. There should be no subsidies for the costs of connecting new building land to infrastructural networks. The last part of the research project sought to define the necessary requirements for residential surroundings and for the urban context. In Leipzig, a market orientated selection process is determining two types of district, compact and perforated, each influenced by their surroundings, either close to the city centre or to the Auewald woods. A current willingness to move house is having a negative effect on “weaker” areas. In order to successfully implement new strategies in late 19th and early 20th century town housing, it is necessary to consider the city’s urban development on the whole and to analyse the concrete situation in the district as well as improving conditions in the block. By developing concepts for whole blocks, parallel improvements can be made to the near surroundings. The demolition of courtyard tenements, far reaching concepts for open space, parking solutions for whole blocks and specifically targeted offers can lead to improved living standards and to a stronger market position. Where local authorities, owners and users share common interests, subsidies should be made available. Where a single owner seeks to implement a concept for a whole block, subsidies should be used soley as an initial impulse, generating a process of self-run renewal. While the renewal strategies proposed here are capable of stabilising certain urban areas, their economical success must be based on the quality of their location, surrounding infrastructure and the district’s social fabric. It will be necessary to undertake pilot implementation projects in order to experience the improvement of standards and the application of strategic thinking first hand. Reliable costing and a functional funding mechanism must be capable of guiding potential target groups across initial hurdles. 4 UTN II – PILOT PROJECT The UTN II Network offers an international forum for moderating problem definition and solving and achieving an exchange of experience and interests with other countries and cities. The first step took place at a Kick-Off Meeting in Vienna in March 2004 which attracted not only the project partners but also a large number of East European cities. 406

A further workshop has been scheduled to coincide with the “Denkmal” (monument) Exhibition in Leipzig. A broad audience can thus be reached through the work within the network. The actual realisation of pilot projects offers the opportunity to test the concepts and cross the first hurdle for new developments. An important part of the project work is the search for potential investors and the moderation of the process. Within UTN II, real start-up work is being done – matched by subsidies from communal and federal sources – thus partly buffering the investor’s risk in testing “new waters”. It is expected that once these initial projects are accessible and habitable, their innovative qualities will enable further unsubsidised development. Parallel research will log, check, evaluate and document experiences and results over the long term and turn them into guidelines. These can be made available to aid those wishing to emulate this approach to late 19th century and early 20th century town housing. Currently, there are intensive negotiations with a number of investors taking place and financing and subsidy scenarios for the projects are being compared. The realisation of the first pilot project will commence in Leipzig shortly.

5 A SUMMARY OF RESULTS The massive amount of property redundancy in the former eastern German States makes a reduction of the amount of accommodation on the market unavoidable. The NEWOG research project proposes a radical transformation of the old town housing currently off the market while awaiting renovation. This transformation should be implemented in such a way as to reduce the number of flats in the buildings on the one hand and to integrate new living qualities sought after by specific target groups on the other. In order for these houses to compete with single family homes, private home-like characteristics and immediate access to open spaces need to be created. By combining these strategies with the advantages offered by an inner city location (public transport network, infrastructure, culture), it should be possible to keep potential home owners in the city and stabilise the existing urban fabric. NEWOG proposes concentrated interventions in existing structures, these creating significant improvements in living standards in less density. Existing funding mechanisms should be adapted to create more incentives. These proposals can trigger sustainable urban development, putting to use existing architectural and cultural resources and supporting the model of the compact, sustainable city with its short distances and effective infrastructure. The realisation of these new living concepts is eagerly awaited by potential users and has been judged as marketable by many parties involved. Previous experience shows nevertheless that to ensure the desired start to the process, clear moderation of the processes, information exchange and the haptic experience of the conceptual realisation through pilot projects, in conjunction with a retrospective in terms of sustainability, as well as urban and economic success, are all necessary. REFERENCES Aring, J. & Herfert, G. 2001. New Patterns in Residential Suburbanism. Leipzig: Institut für Länderkunde. Fritzsche, A., Kabisch, S., Müller, S. & Steinführer, A. 2000. An Evaluation of an Survey of Inhabitants in Gohlis-Süd in Leipzig. NAWO Bulletin Nr. 5.12. Leipzig. Pahl, B. & Widmann, A. 2000. Late 19th and early 20th Century Building Types in Leipzig, Categogies – Potential. Leipzig: University of Leipzig. Tschaschel & Wollkopf 1996. A geographical excursion through the city and its outskirts. 2002. City of Leipzig, Dept. of Town Planning and Building. Monitoring Report. A survey of citizens – an overview of results from 1999, 2000, 2001. Office for Statistics and Elections. Leipzig. 2000/2003. Urban Development Plan for Housing and Urban Renewal (STEP). Leipzig.

407

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Using MBT in transformation of multi-family prefabricated buildings A. Rybka Rzeszow University of Technology, Rzeszow, Poland

ABSTRACT: The structural systems used in Poland in the period of 1950 to 1990 for industrially constructed residential buildings have been shortly presented. Main structure types were applied: block type, large-size panel. Main structure characteristics have been also indicated. Problems with large slabs dwellings houses were described. Discussion of possibilities of transformation of these buildings has been done. Nowadays requirements of the existing multi-family prefabricated building were described. Example of such existing building was shown. Project of possible changes in such a building have been shown. Some problems of such old large panel dwelling buildings were described. 1 INTRODUCTION In Poland, we have approximately 300 dwellings per 1000 inhabitants nowadays. This coefficient is one of the lowest in Europe. The first problem is that, we have to demolish about 800 000 dwellings because of their technically poor condition level. Also, we must build about 1500 000 dwellings because of our social needs in Poland. So in total, we need about 2300 000 new dwellings in Poland in near future. The second problem is, how to accommodate to new social needs dwelling houses constructed, between 1945–1990, with using of prefabricate technologies. After II World War Poland was strongly destroyed. The first step of rebuilding in Poland was build dwellings with using traditional technologies. This traditional method was used between 1945–1950. Second step was start to used prefabricated technology. Dwellings were built in large blocks technology. Third step was start to used large slab technology. In Poland, prefabricate building technologies were used in order to meet the social dwelling requirements, together with the cost reduction. The simplest solution was mass-production of dwellings. In advance preparation of the typical building elements speeded up the time of assembly and increased the quantity of putted to use houses – that were the matter. Must the like houses, which have existed for dozens of years, be change? If we look at present day dwelling and service development, that direction is unavoidable. Due to still high cost of development, the large slabs buildings are going to be used. One more possibility remains – adaptation of the buildings to actual housing and servicing social needs. MBT can help to solve these problems. The existing situation in large slabs dwellings houses does not fulfil, nowadays tenants demands. Main problems we can find in tree domains: Architectural aspects Structural aspects Physical aspects Increasing of thermal comfort in existing large slabs houses is needed. Because of climate conditions, change of composition of the existing roof ’s covers is needed. In the low buildings change of roofs shape seems to be rational solution. While changing of the roof ’s shape at the same time we can varies building shape and make its renovation or we have possibility of increasing number of flats in existing building. 409

Another thing is the problem of changing first floor area in existing large slabs dwelling houses. Nowadays social needs show that we need more space for retail trade or for handicraft in neighbourhood of the flats. We can find the area for it on the first floor in existing houses. Another thing is the problem of changing aesthetic of these houses.

2 ANALYSIS OF PREFABRICATED DWELLING SYSTEMS USED IN POLAND In Poland large blocks and large panel dwellings buildings were raised by industrial method. Constructions of them were characterized by: arrangement of bearing elements, size and solution of prefabricated elements, solution of connections in knots as well as used of materials. In 50th year, first prefabricated buildings were realized in Poland. In large blocks buildings lengthwise arrangement of bearing walls was applied. On lengthwise load-bearing walls, channel ceiling plates were put. Lengthwise arrangement of bearing walls in that type of multi-family buildings makes considerable difficulties to do functional changes. First large slab buildings were done at the end of 50th year. Crossways arrangement of bearing walls was used. Ceiling plates were crosswise too. Internal walls were executed from concrete. The possibilities of changes that type of multi-family buildings is very limited. Open dwelling system with large-size slabs and walls elements, Szczecin system, was used in Poland started from 1967. Crosswise configuration of load bearing walls was the main arrangement used in this system. This arrangement of load bearing walls was used in several systems: W – 70 system, Wk – 70 system, OWT – 67 system and OWT – 75 system. MBT can help to do functional changes in this type of building. Scale of problem causes necessity of modernization of these objects. To consider it belongs itself changing relate of occupants requirements of functional solutions, install equipment, suitable internal climate. Demolish of this buildings is alternative for modernization. Yet in view of lack of flats in Poland as well as limited possibilities of financial state, and also as well as small affluence of society, it is not to accept. Decision about survival and modernization of existing large slabs buildings is well founded. Number of habitable buildings realized in this technology in years 1950–1990, it is about 50% of all buildings. Possible ways of adaptation of such buildings are: – – – – – –

join of neighbouring flats superstructure in aim of obtainment of new habitable surface adding of new loggia and balconies installation of additional lifts casing of balconies and loggia change of architectural view of buildings.

3 MODERNIZATION OF EXISTING LARGE PANEL DWELLING BUILDINGS Checking of theoretical possibility of introduction of architectural changes is aim of design works in buildings of this type. For qualification of full possibility of modernization of such buildings, is necessary to realization of suitable constructional calculations, which presented design does not hug. In every case of modernization workings one should to get every agreement and required permission regulations of building law. Problems of superstructure of habitable buildings join with following questions: – with lack of habitable surface connected with demographic changes and needs of new generation of occupants – adaptation of existing functional solutions to new needs 410

– tendencies of limiting of new terrains under housing building from economic regards and using of existing infrastructure (rule of balanced development). Modernization creates new possibilities of architectural view of building. Realization of full modernization undertakings together with adding new balconies and loggia, to enter high roofs, suitable form it creates change of elevation. General conception of superstructure should fulfil requirements and standard decisions. Order of operation hugs following stages: – – – –

study of technical evaluation of chosen object study of architectural conception study of constructional conception study of conception of using of materials.

Conception of superstructure foresees usually solution of technical additional storey. First step based on technical state of object from point of constructional sight, and so checking, whether bearing walls and fundaments can to safe transfer of additional burdens. Founding positive result of computational analysis (fulfilled condition of carrying capacity), it is possible to approach to fulfilment of remaining basic requirements. Question of perpendicular transport stays in 5-storey buildings. In this case external lifts are possible to use. Convenient access to individual flats gets itself into this way. Superstructure, and then sale of new flats to can be source of financial resources maybe onto modernization of large slabs buildings. These investments are remunerative in view of zero costs of terrain, foundations and existing technical infrastructure. 4 EXAMPLE OF MODERNIZATION OF EXISTING MULTI-FAMILY LARGE PANEL DWELLING BUILDING Shape of existing building is shown in figures 1, 2 and 3. Adding new elements to existing building does changes. New level of flats and new roof was added. Also new loggias and balconies were added. New lifts help to better communication in the building. Project show how this changes can be done in existing building and what are final results. See figures 4, 5, and 6. 5 MODERNIZATION OF GROUND FLOOR IN LARGE PANEL MULTI-FAMILY BUILDINGS Concrete comes perfectly true in Polish weather conditions. The use of prefabricated technology (the internal constructional walls, ceilings, the steps, elements of lift-shafts, roof elements) reduces

Figure 1. Repeatable level of existing building.

411

Figure 2. Section of existing building.

Figure 3. Facades of existing building.

about 10–20% cost of building. Buildings done in years sixtieth and seventieth, executed in large panel technology had a lot of defects. Architectural solutions, were very monotonous, what caused sad appearance of buildings. Today, prefabricated elements are executed to individual projects, eliminating former defective solutions. Existing objects done in years sixtieth and seventieth should be subject of changes which will adapt them to new needs and permit them more far exploitation.

6 THE ADAPTATION OF GROUND FLOOR IN EXISTING LARGE PANEL DWELLING BUILDINGS TO COMMERCIAL FUNCTIONS Nowadays, such old buildings about exclusively habitable function exhibit from side of street banal, monotonous elevations. One of way of the modernization of these buildings is the proposal 412

SEGMENT 2 28 955 67

91

150

112

SEGMENT 1

90

121

150

119

240

15

585

75

150

91

269

150

121

15

345

600

90

120

240

90

150

210

15

91

150

240

465

3 600

89

121

150

120

239

15

150

91

270

585

4 800

149

121

15

90

120

239

345

6 000

97

90

210

15

3 600

142

465

15

480

75

17 145 90

82

90

150

90

150

90

210

150

150

150

150

90

260

150

7

150

480

4 800

210

15 90

465

90

15

15

128 15

150

465

2 400

120 210

A

4 800

15

90

150

225

150 240

90

15

90

210

90 210

150

150

150

467

160

150 300

150

20

150

211

150

90

90

90

150

60 179

150

6

150

199

210

112

16

B

28

co

co

8

232

15

345

15

SYPIALNIA

90 70

20

147

139

2 400 15

297

222

88

20

279 287

150

53

80

210

210

180 20

163

247 319

28

480

401

33

15

180 95

27

48

222

20

150

34

280

20

20

402

251 300

30

4 800 15

202 120

67

465

232 150

2 400

82

15

232 150

82

15

222

20

258 150

108

279

20

263 150

29

400

249

20

299

30

10 20

200 120

131

65

150

120

76

20 126

235 150

85

150

51

17 255

Figure 4. Project of changes in repeatable level of existing building. +18, 50

+17, 17

2 800

97 143

210

396

+17, 52

125

1 050

% 15

53

%

140

53

10

+15, 07

10

10

,5 17 5x 27

16 20 52 30 3

8x 18 27 ,8

10 82 10 64

228

100 82 54 82 10

210

540

,5 17 8x 7 2

18

+5, 60

100

52

8x 17 27 ,5

82 10

210

540

18

+2, 53

34 2 00 0

2 270 10

30 85

410

Figure 5. Project of changes in section of existing building.

Figure 6. Project of changes in facades of existing building.

413

-1 460

-2 , 45 -3 230

40 427

-1 , 07

-1 , 07

Dy l a t a c j a

30 400

82 185

-1 050

,5 17 8x 27

-2 450

-3 230

2 100

6x 17 27 ,5

10

2

7

±0, 00

15

126 16

400

20

600

Dylatacja

400

+0, 05

-2 450

10

210

257

2 100

,5 17 8x 27

40

600

52

8x 17 27 ,5

100

2 580

,5 17 8x 27 +2, 80

+1 40 0

-1 460

1 991

100

18 52

8x 17 27 ,5

+4 20 0

+2 84 0

1 351

10

210

2 580 2 580 16

60

,5 17 8x 27 +8, 40

+7 0 00

+5 64 0

16

60

8x 17 27 ,5 +9 80 0

+8 44 0

16

60

1 988 1 591

+11, 20

52

16

60

2 580

+9 80 0

,8 18 8x 7 2

+11 240

400

16

20 30 3

+14, 23

38

17

150

90

150

90

150

90

150

210

150

150

90

210

90

210

80

210

8X 17. 5 X 27

8X 17. 5X 27

70 210

WC

210

10 1 500

150

1 400

10 120

10

0 15

150

110

420

10

462 211

10 1 400

180

3 000

250

420

268

101

4 900

210

77

90

65 482

0

3 425

3 000

251 150

210

90

65 250

275

1 400

12

KUCHNIA

275

240

210

SYPIALNIA

150

150

150

120

10

20

150

150

1 400

1 500

150

150

150

150

150

150

150

120

1 500

10

20 2 400

62

150

150

150

150 210

90 210

138

1 000

10

225

210

90

210

210

20

227

90

65

1 400

1 800

90

210

110

6 65

8X 17. 5X 27

8x 1 7 . 5 X 27

80

4 900

90

10

co 165

80

PRZEDPOKÓJ PRZEDPOK

210

70

co 20

20

28 980

3 000

238 150

222

SY PI AL N I A

150

B 490 84

15

10

A 20

248

1 400

275

20 1 400

171

180 81

3 300

65

3 425 100 900 20

1 500

210

1 500 1 750

K U CHNI A

co SYPIALNIA 15

315

1 800

230

3 425

220

210

210

80

8X 17. 5X 27

210

10 1 400

110

2 700

65

10

210

70

K U CHNI A

167

10

1 400

600

150

150

150

150

150

150

150

180 140 120

150

150

1 200

150

10

270

200

164

90

210

60

250

275

150

150

200

10

1 400

POKÓJJ DZIENNY POK

90

210

SYPIALNIA

0 12 0 15

150

8X 17. 5X 27

4 900

354

10 co

0 15

10

SYPIALNIA

co

0

1 400

2 225

4 900

210

80

4 900

90 65

29

1 125 80

210

80

210

4 900 110

270

4 900

8X 17. 5X 27

8X 17. 5 X 27

80

210

1 000 7

210

223

65

90

SYPIALNIA co 20

250

210

90

1 480

1 200 310

13 115

65

10

232

20

222

5 15

0

150

co

15

12

407

249

15

210

90

65

1 620

2 475

WC

KUCHNIA

90

210

70

210

co

65

90

P R Z E D P OK ÓJ

210

KUCHNIA

167

SYPIALNIA 210

SYPIALNIA

AZ I E N K A

90

210

210

SY PI AL N I A 2 100

0 15

SYPIALNIA

63

co

210

KUCHNIA SYPIALNIA

90

210

90

SY PI AL N I A

12

210

POKÓJ POK J DZIENNY SYPIALNIA

90

210

KUCHNIA

165

210

90

90 210

90

90

210

P R Z E D P OK ÓJ

210

90

90

210

70 210

210

WC

210

90

90

210

210

PRZEDPOK PRZEDPOKÓJ

70

WC

90

AZ I E N K A

210 co

AZ I E N K A

P R Z E D P OK ÓJ

210

90

PRZEDPOK PRZEDPOKÓJ 70

1 795

70

210

210

WC

70

210 co

P R Z E D P OK ÓJ

AZ I E N K A

90

10

90

70

210

WC

210

70 210

90

70 210

AZ I E N K A

210 co

POKÓJJ DZIENNY POK

210 AZ I E N K A

POKÓJ POK J DZIENNY

POKÓJJ DZIENNY POK

210

POKÓJJ DZIENNY POK

70

WC

210

90 210

SYPIALNIA

1 500

SYPIALNIA 90

POKÓJ POK J DZIENNY

2 100

210

90

co

co

70

210

90

90

150

90

90

150

210

150

150

90

210

150

150

150

150

SYPIALNIA

co

195

co

co

465

POKÓJ POK J DZIENNY

SYPIALNIA

2 230

co SYPIALNIA

345

210

2 330

co

250

of introduction, in part of them, in place of residential flats on ground floor to put commercial functions. This idea was motivated of liveliness of streets and existing public spaces among these buildings. Another idea was feed of the pedestrian’s draughts, which can become the attractive trade passages.

7 PROBLEMS OF CONSTRUCTION The functional arrangements of flats in old large panel dwelling buildings were realized in adapted to of that time requirements systems and it marks it comparatively large the saturation internal constructional walls, as well as, small spreads of ceilings between transverse bearings walls, though larger spreads in open newer systems overweighed already (e.g. 5,4 m and 6 m). The adaptation of space of ground floor on commercial function in such buildings should be adapted to existing arrangement of constructional walls, because of the behaviour of arrangement of building static in state of equilibrium depends there from base story. The larger openings of interiors by puncture in internal walls, the large openings or the super session the large fragments of walls e.g. threaten with steel horse heads the loss of sedateness. First of all the constructional elements, which nowadays state is fortunately good, but they requires the protections before any unfavourable load variation. The largest adaptive difficulties can cause the necessity of strengtheners of ceiling over cellars, according to enlarged usable burden in many cases.

8 ARCHITECTURAL FORM The external wall face of ground floor should be adapted to commercial functions in measure of possibility without exchange of existing large panel elements. It will be possible, obviously, to enlarge the surface of windows or to exchange on entrance doors by simple intervention and removals below the window side. It the desirable aesthetical effect and obtainment the new architectural form of ground floor will be possible to reach by addition from outside homogeneous protection side, e.g. from heavy-duty dyed glass, affirmer simultaneously protection before excessive insulation. The architectural solution of elevation of ground floor has to hug simultaneously the transformation unsightly in general, the cellars of visible zone over level of terrain and the suitable form to service premises the entry which the floor will stay on exists ceiling over cellars.

9 POSITION OF THE FLOOR ON GROUND FLOOR IN SUCH OLD LARGE PANEL DWELLING BUILDINGS It marks the flat realized in old multifamily buildings done in large panel technology the ground floor lie about 1,2 m over level of terrain near entry to building. Such solutions were used general in all such buildings. That came into being the huge supplies of dwelling buildings from the ground floors upraised on so high-level in relation to terrain. Nowadays, this is essential difficulty in assurance of access to possible commercial premises. To these rooms would belong to add external stairs, because common entry through flats operator staircase exclude sanitary and safety usable regards.

10 HEIGHT OF ROOM ON THE GROUND FLOOR LEVEL IN LARGE PANEL DWELLING BUILDINGS Almost in all large panel dwelling buildings the height of store carries out 2,8 m (with ceiling), that is 2,5 m in light between floor and ceiling, which answers the required height of habitable rooms. Designed the requirements the relating heights of every rooms on the men’s stay, with what 414

buildings and their location should answer admit to receive the height 2,5 m in light for designed to not burdensome work rooms or it falls the exercises in rooms in which is in draught of day of work the steels it more than 4 person, and on every of them the similar functions, at least 15 m3 of free cubature of room. The room in which to spend time larger number of persons it has not burdensome conditions, they had to have height, at least 3,0 m in light, and in burdensome conditions of minimum 3,3 m. In these technical and functional regards it is very difficult to find new function for this ground floor space. One of the possible is to consider rather the possibility of obtainment on the larger height of commercial room by draw down the floor e.g. to 0,15 m above the terrain in new added elements.

11 PROBLEM OF VENTILATION In flats in old large panel dwelling buildings, gravitational ventilation was used. “Collect canals” were executed from concrete notepads. The assurance of correct of gravitational ventilation, particularly in incident of use of collect canal, it requires the usage in rooms the inlets in which are to combustion lines or summary lines of gravitational ventilation the summary combustion lines and smoke as well as individual exhauster. In such solutions installed in room of device, in peculiarity the using up air, they it cannot to call out disturbances in ventilation and to limit her effectiveness. Peaceably with recipes, to service of service rooms in ground floor of dwelling buildings become necessary the use of mechanical ventilation would be or even the air-conditioning – e.g. in incident lowering of required height of such rooms. Lack of any reserve of internal surface in large panel dwelling building makes problem with execution from commercial rooms the lines of winding ventilations in ground floor by laid higher the room or staircase.

12 PROBLEMS OF NOISE AND TREMBLING Use service rooms, in dependence from their character, the enlargement for occupants of higher stores difficulty with regard on noise higher level can cause considerable or trembling. In large panel dwelling buildings, near thickness of the prefabricated reinforced concrete ceilings and walls, their acoustic isolation were very low, what marks, that they fulfil the low requirements. The enlargement the noise level would require trembling in commercial rooms additional acoustic isolating, at least ceilings over commercial rooms in ground floor. It is not suspension the suitable layer of isolation under ceiling or realization of additional acoustic isolation on them. However in this case, it would lead to inadmissible limitation of height of commercial or habitable rooms, below 2,5 m, in light.

13 SELECTION OF SOLUTION The above mentioned objective conditions of possibility of adaptation habitable ground floors to commercial functions in large panel dwelling buildings prove, that this folded task is difficult. It is practically possible. So to consider the possibility of location on ground floor of this type of building of not burdensome services, possible to service near solid workers and small servants’ number, satisfaction individual persons’ needs – the highest several customers simultaneously. They can then be the points of reception on different executed in specialist workshops services every orders for example or directly at ordering. The change the service workshops will not it raise however in significant prepare the architectural attractiveness for pedestrians, shaped through large panel dwelling buildings about habitable character, public draughts the proper housing estates the mono function. The other thing is elevation program attractiveness and aesthetical the not only farmers and townies calculate, but also potential service providers which the effect of obtainment of impression of settlement can add prestige. 415

14 CONCLUSION The use of new technologies (MBT), now aiming at new instruments available at planner, allows more precise checks when develops the idea of project. With help of using Mixed Building Technologies existing large panel, multi-family houses can be change and their shape can be totally new. REFERENCES Collective work, 1998, Concrete building, Arkady, Warsaw W¸eglarz M., 1972, collective work, Dwelling construction systems, Arkady, Warsaw Concise Statistical Yearbook of Poland 1995, 1999. Polish Statistical Office, Warsaw 1996, 2000. Januszaniec B., l985, Concrete building for architects, Bialystok University of Technology, Bialystok Kobiak J.,1991, Ferro-concrete constructions. Arkady, Warsaw Krol W., 1969, Ferro-concrete constructions. PWN, Warsaw Lewicki B., 1961, Big size elements for dwelling buildings, Arkady, Warsaw Parczewski W., 1995, Concrete building for architects: finishing works elements, Warsaw University of Technology Publishing House, Warsaw Rybka A., 1995, Central Industrial District and the Polish avant-garde town planning between wars, Rzeszow University of Technology, Rzeszow Rybka A., 2001, Evaluation of MBT buildings – Modernisation Aspects and an estimation of a heat insulation of multi-story houses made of large-sized prefabricated units, COST C12, Bled Rybka A., 2003, Large-panel dwelling building, currant stage – way of construction and architectural modification, 207–216, Improvement of building structural quality by new technologies, Proceedings of the international seminar Lisbon, 19 and 20 April 2002, Universidade de Coimbra and Universidade de Minho, Portugal, COST C12 Starosolski W., 1985, Ferro-concrete constructions, PWN, Warsaw Zaleski S., 1997, Remonty budynków mieszkalnych, Arkady, Warsaw

416

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

The architecture of the buildings in Düsseldorf Harbour Region as a master guide for all Y.K. Aktuglu Dokuz Eylül University Faculty of Architecture, Izmir, Turkey

ABSTRACT: For Rhine is a very important water way for transportation for the people, live in Dusseldorf, the new planning strategy of Dusseldorf Harbour Region is adding a great income for the city from every point. The structures of the buildings are challenging with the olds. The new athmosphere in the area is giving a new hope for the people, both who are working there and also who are visiting the area. The architecture of the buildings gives an inspiration to other architects and designers. While these are taking place, the city is becoming more attractive, also with its new airport building. Then the trade capacity and speed are being double with such kind of efforts on construction materials, building structures and technology. In the paper, the most attractive buildings with their structural solutions will be defined in urban context near a river, Rhine.

1 INTRODUCTION While designing a new building on earth, the environment has a great vital role in design. The characteristics of the region gives the clues to the architect to shape the product. Düsseldorf is a very important city in Germany in the name of being a settlement near Rhine and also it is a very busy business city. The banks of the Rhine are potential places to attract the investor and the client and the tourists. The urban gateway as Stadttor, the Westdeutscher Rundfunk, the Landtag and other lots of new and some old buildings call for a new urban design project to be constructed in both sides of Media Harbour in River Rhine. All of them are having a different language in construction and in design. And all of them are attractive and worthwhile to be photographed and also to be visited. This project can be a good urban design project erection for the people to have more efficient areas to work and to visit. Also along the way from the Altstadt of Düsseldorf to the Media Harbour, all of the buildings constructed on the bank of Rhine are still in service with their well-designed facades and plans. The principles in designing of this Media Port area can be observed over the buildings, together with their architects and structural solutions. In the paper, some of these interesting buildings and structures will be announced with their principles to be added to this project as a master guide for future.

2 IDENTITY FEATURES OF DÜSSELDORF Düsseldorf is located on the beautiful Rhine River, and is a home to more than 100 museums and galleries, as well as lively cultural scene with venues for music and dance. And also Düsseldorf is a popular venue for conventions and conferences. The Düsseldorf Conference Center(DCC), which is located in the North of the city and only 3 kilometers from the 417

Photograph 1. General view of River Rhine welcoming to the Media Harbour Area (Aktuglu, 2003).

airport, provides an excellent venue for the Congresses, with modern facilities, comfortable meeting rooms, spacious exhibition halls and a restaurant with a beautiful view of the Rhine. The comfort and convenience of the DCC is enhancing the experience at the Congresses. It can be seen that at the upper part, there is a skyline full of church spires; and at the lower part there are picturesque cobbled streets. The Konigsalle (The Kö) is maybe the most opulent shopping street in Western Europe. Lots of statues and sculptures dot the city landscape. Düsseldorf Airport is the third largest in Germany and welcomes about 15 million travelers each year. And new additions made Düsseldorf easy to get to-and close to the Congress activities. The Rhine River is being accepted as one of Europe’s most important and romantic-waterways. Dusseldorf which is located on the banks of the Rhine, is the capital of North Rhine Westphalia Germany. In past, the city dates back to 1159 and radiates the confidence and glamour of a blue-blooded aristocrat. Dusseldorf has rebounded from wartime destruction with an indefatigable resilience that translates into pride among residents. By day, crowds line city’s fashion runways and promenades; by night – propriety is cast aside as shoppers flock to the 500 or so pubs in and around the city’s Altstadt. While investigating in Düsseldorf as a tourist, you can visit the Altstadt, Königsalle, Rhine River, Media Harbour and most museums by beginning the journey from Jan-Wellem-Platz. At the Altstadt while walking from the middle section of the Heinrich-Heine-Alle, when you walk towards the Rhine, there can be found narrow lanes and old churches, quaint breweries and ancient pubs, hip bars and clubs and good german food.

3 ARCHITECTURE IN MEDIA HARBOUR The Media Harbour became a meeting point for connoisseurs for architects from all over the world with the projects of famous architects such as Frank O. Gehry, David Chipperfield, Joe Coennen, Steven Holl and Claude Vasconi. While walking along the Rheinuferpromenade (Rhine Embankment Promenade) toward the Stromstrasse, all of these products are in view. 418

The list of buildings in Medien Hafen starts with Stadttor, by Overdiek, Petzinka & Partner architects as project design, in 1998. The list continues with Neuer Zollhof 1–3 in 1998–1999 by Frank O. Gehry, and as the 50th building, at Hammer Strasse, two buildings are in planning stage. Most of the buildings are dating after 2000 and there are some buildings from 1982 as Rhein Turm by Prof. Herald Deilmann, from 1991 as Westdeutscher Rundfuk by Parade and Partner. And maybe one of the most interesting buildings, constructed with a very courageous steel structure, is “PEC-port event center” designed by Norbert Wansleben, Köln, in 2002. “Colorium” at Speditionstrasse 9, by William Alsop, GB, has a 62 m height, built in 2001. The building at Hammer Strasse 19, designed by Petzinka, Pink and Partner, built in 2002 has a different closed glazing gardens in two sides. 4 THE MOST INTERESTING BUILT EXAMPLES 4.1 Stadttor Stadttor has a composite structure to carry the building, having a height up to 80 meter. It was erected in 9 months as a “prism”. The two 19-storey office towers are rising above the entrance to the two road tunnels along the River Rhine and are linked at the top by a three-storey cross-structure (Strodthoff, 1999). Here, steel allows great distances to be spanned, and this, in turn, means greater freedom in the layout. The two towers of the gateway, are linked at the top by a bridging structure and encased in an all-enveloping skin off glass, and they form a portallike opening – a 58 metre-high glazed hall with shops and restaurants at the base. The primary load-bearing structure, which is a trussed frame with steel columns filled with concrete in a composite form of construction, is standing independently in the atrium behind the glass façade. With the creation of a new state parliament, by routing of traffic underground and by landscaping of the Rhine promenade, new life has been injected into this riverside area of Düsseldorf, to which this prismatic structure is forming a striking gateway. 4.2 FrankGehry’s buildings Gehry buildings shape the image of the new media mile from every side. At the moment, in tourist information center in Düsseldorf, Gehry Buildings are being announced as the symbol for the city as in the shape of mugs, etc. These three towers are warped, they are crooked, they are interlocking and full of nooks and corners. They are really impossible in first look. If the sun will shine on the metallic cladded tower, then the picture in front of us is calling to visit it. The “Neuer Zollhof” is forming a unique ensemble – an architectural masterpiece of its own kind (www.duesseldorf-tourismus.de/article). The buildings designed by Frank O.Gehry were completed just before the turn of the millennium and have become the modern landmark of the state capital. The wavy skin of the building, curvy, winding, asymmetrical shapes, strange angels, distorted facades, protruding window frames and more are celebrating the harbour area for its new image. 4.3 PEC – Port Event Center This interesting building with a steel structure outside is settled at the inner corner of the harbour. It flys over an old building by cantilevering several meters away. The composition of the structure, constructed at the outside of the building is giving lots of clues about the capacity of a structure with steel. PEC is designed by Architects Norbert Wansleben and completed in 2002. 419

4.4 Colorium Alsop Architects plans an 18-storey multi coloured office block in Düsseldorf. The 11 million sterlin tower has three multicoloured facades and black and white asymmetric stripes on the forth (AJ-29Nov.2001). This large illuminated overhang glows red at night and the building can be seen from 7 km. away.

5 CONCLUSION To shape the image of the new media mile, first it is decided, then the buildings are designed in a very high quality from the point of attractiveness, then they are built in a very new engineering technologies to be immortal longer. The result is as a frame on the wall to say “yes this is the picture of our power in urban design with the high quality of the living conditions in the meaning of having a motivation both while inside the buildings and also outside of the buildings, mostly by using steel in construction. REFERENCES Strodthoff, W. 1999. Das Düsseldorfer Stadttor. Stahl und Form.Stahl_informations-Zentrum. Düsseldorf. Voigt, C. & Leisner, R. 2001. Der MedienHafen Düsseldorf. Meile der kreativen Künste. NOBEL-Verlag. Essen. Facades en Acier Inoxydable. 2002. Série Batiment, Vol.2. Euro Inox. The European Stainless Steel Development Association. Belgique. The Architect’s Journal. 29.Nov.2001.Stirling 2001 – Potential Winners. http://2003.osgiworldcongress.com/ Some papers about the buildings in Media Harbor presented in the Stahlforum, took place in November 2003 in Düsseldorf. A visit to the area during the time when the Stahlforum took place in November 2003 in Düsseldorf.

420

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

The structural features of Millennium Bridge in London, connecting St. Paul and Tate Modern, as a very successful Urban Design Project Y.K. Aktuglu Dokuz Eylül University Faculty of Architecture, Izmir, Turkey

ABSTRACT: Millennium Bridge, which is 4 m wide, is a suspended footbridge between St. Paul’s and the new Tate Gallery on the other side of the river. It is the first new bridge across the Thames in one hundred years. In these days, it has been a popular route across the river and also it has been enlivening two river banks that have been rather belighted by poor Access. For walking on the Millennium Bridge, people are visiting St. Paul’s first and then after crossing the Thames, they are also visiting Tate Modern only by walk. From the beginning time till now, Millennium Bridge has always a happenning to be solved or to be discussed from structural point to carry the pedestrian in a safety position and also from aesthetical point, to add quality for life. In the paper, millennium bridge will be defined from the first day till now, with reasonable topics about its structural design and architectural design. And also the advantages of such a bridge to the city will be explained with its benefitial credits for the city. 1 INTRODUCTION For the aim is to connect St. Paul and Tate Modern with a pedestrian Access, an international competition took place in 1996 by the Financial Times. Also it was celebrating the Millennium. The winners of the competition was Foster and Partners/Sir Anthony Caro/Ove Arup & Partners. The proposed was an innovative and complex structure, featuring a 4 m wide aluminium deck flanked by stainless steel balustrades, supported by cables at each side. After a break in service for two days and twelve months, the exact nature of the problem was identified and the dampers were installed, then problem was solved. The result was looking like a blade of light during night time, having a largest span at the centre as 144 m between the North and South Piers. Then it has 330 meter length, and it is in the “A list” of London attractions. 2 IDENTITY FEATURES OF MILLENNIUM BRIDGE • • • • • • • • •

It is designed by architect Sir Norman Foster with sculptor Sir Anthony Caro and engineers Arup. It is openned at 10th of June 2000. It is closed at 12th of June 2000. It is reopenned at 27th of February 2002. It has a length of 330 m. It has a width of 4 m. Its height above river at high tide is 10.8 m. Its handrail height is 1.2 m. The piers are from concrete and steel. 421

Photograph 1. The whole view of Millennium Bridge from Tate Modern (Aktuglu, 2003).

3 STRUCTURAL FEATURES OF MILLENIUM BRIDGE 3.1 Spans Millennium Bridge has a span at the centre as 144 m between the north and south piers. On the southwark side, the south span is 100 m and on the city, the north span is 81 km. 3.2 Cable sag The cables, tensioned, sag only 2.3 m over the centre span, which is a span to dip ratio of 63:1. And 6 times shallower than a conventional suspension bridge structure where a ratio of around 1:10 is common. 3.3 Cables There are four 120 mm diameter locked coil steel cables which are used on each side of the bridge. These cables provide the bridge with stiffness in both vertical and lateral directions. They maintain around 2300 tonnes of force(22.5 MN) in total. 3.4 Pier head and pier body and pier foundations The pier heads are hollow and to form the required shape, they are formed of 70 mm thick flat steel plate folded and welded. The anchorage of the pier head to the concrete body below is made by a series of 75 mm diameter high strength steel bars. To connect the pier head to the foundations in the river, the pier body is made from reinforced concrete and required specially constructed moulds to form its tapering elliptical shape. 422

For the foundation, there are twin 20 m deep 6 m diameter caissons into the river bed to support each pier. These caissons were constructed as shafts dug from the top down in a sealed cofferdam in the river, and these shafts were then backfilled with reinforced concrete and topped with a 3 m deep pilecap.

3.5 Saddles The saddles of Millennium Bridge support the cables as they pass over the piers. Clamps which are on each side of the saddles prevent the cables from sliding, thereby adding stiffness to the structure. They were made from molten steel cast in specially formed moulds.

3.6 North abutment The north abutment connects the cables, which terminate in a steel anchorage which is connected to the pilecap via two concrete walls, to the foundations in the ground below. And underneath the pilecap, twelve 28 m deep 2.1 m diameter reinforced concrete piles transfer the structural forces to the underlying clay.

3.7 South abutment On the south abutment the cables are anchored to a steel “strut and tie” system which is fabricated from steel and the ties formed from 100 mm diameter bars. This steel structure is fixed onto the 3 m deep pilecap where below the pilecap sixteen 28 m deep 2.1 m diameter reinforced concrete piles are used to transfer the cable force to the clay below.

3.8 Deck The deck is 4 m. wide, and it is formed in extruded aluminium box sections which link together via tongue and groove fittings. The structural sections span onto steel edge tubes on each side of the bridge. Arm: The arms of the deck which are steel hollow box sections tapering from 450 mm square to 225 mm square at the cables, the clamps are used and they were cut from 350 mm thick steel plate. Edge tubes: Edge tubes are round steel and they span between the arms and support the deck. Aluminium decking: Extruded cellular aluminium sections which are for decking span between the edge tubes to form the walking surface. Lighting: To create the “blade of light” when the bridge is viewed from a distance, proprietary light pipes along both edges of the deck. And light is carefully directed and reflected from the deck to provide a safe and comfortable level of ambient illumination on the bridge level of without the need for overhead lightning. Ballustrades: Ballustrades are from stainless steel and enclose the deck and support a wide stainless steel handrail. Movement joints: To allow the deck to expand and contract as the cables more, movement joints are installed every 16 m on the bridge deck.

3.9 Bifurcation At the south end of the bridge, the deck bifurcates and is known as the “needle” due to its shape in plan. For enabling this to happen, the deck supports tubes increase in size to resist overturning and movement by using torsional stiffness. 423

4 CONCLUSION In everyday’s life, we are used to do some vital things as surviving in life by working in a job, eating and sleeping for our physical requirements. And everyday new buildings and structures are being joined to our artificial environment where we are taking our breathes in. As we know very well, that if we will live in a well prepared world, then our products will be more smiling. While a bridge is accessing Thames to connect two sides of the river, having very attractive points such as St. Paul by Sir Wren and Tate Modern by Herzog & de Meuron, then the connector, by Foster, Caro and Arup should be very attractive at least as the others are. Of course, the products which will be always in front of the eyes, should be taken care very much than the others. Here the solution is very elegant as a sculpture. When you look to the bridge from every window and balcony of Tate Modern, every time you are feeling yourself very good with a feeling that there is a piece of from your world in 2000’s in front of the old. While walking above the bridge you are behaving as flying humans by thinking that you are walking above Thames. This is a great fantasy. It looks like a view that while driving a car in highway from Izmir to Cesme, in three lines in one direction, for the cars are going with a speed more than 170 km, you observe them as small ships moving in the sea, because of my speed which is 90 or 100 km. If a building or a structure built with steel can be designed according to the capacity of material as being the best for tension as it is in steel, then the product is unbelievably looking like a sun, smiling always in Cesme. Steel as a material, cables as structural members, dampers as technology bring you a piece of sweety taste of life for helping the development of a city, finally of our world. REFERENCES Allinson, K. 2003. London’s Contemporary Architecture + A Visitor’s Guide. Dallard, P., Fitzpatrick, T., Dr.Flint, A., Low, A. & Smith, R.R. 17 April 2001. The millennium bridge, London: Problems ans solutions. The Structural Engineer. Volume 79/No8:15–17. Fitzpatrick, T. & Ridsdill, R. August 2001. “Stabilising the London Millennium Bridge”, Earth, Air and Water. Ingenia, No:18–22. Parker, D. 23 November 2000. Marching to a new tune. CNew Civil Engineer, No:1378,12–13. Parker, D. March 2001. Swaying opinions. New Civil Engineer.No:92: 8–9. Prof.Wise, Chris. January 2000. “The Birth of London’s Millennium Bridge” – innovative design. Ingenia. 9–12. Sample, I. 31 March 2001. The bridge of sways. New Scientist:38–41. Seaman, J. & Hayward, D. January 2000. Poetry in Motion. Ground Engineering. Yuan, F. November 2000. Pressure relief for the London Millennium Bridge. CGround Engineering. 42–46. 5 December 2000. Pedestrian-induced vibrations of footbridges. The structural engineer-volume 78/No23/24. Several technical visits to the bridge in between 28 June–3 August 2003 with a third British Council Scholarship to make researches about steel buildings and steel structures in London and in England. A meeting with Prof. Chris Wise in his office to talk about his projects and also to talk about Millennium Bridge.

424

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Improving the performance of buildings Heli Koukkari & Pekka Huovila VTT Building and Transport, Espoo, Finland

ABSTRACT: In the building performance analysis as developed at VTT, the characteristics of buildings are classified according to three main categories that are A) conformity, B) performance and C) cost and environmental properties. Each category has a hierarchy of indicators and their values. A software tool has been developed to support multicriteria decision-making from the whole life costing perspective. Its database includes a number of preset definitions and performance levels as well as description of verification methods. KEYWORDS: building performance, eco-efficiency, cost-efficiency, building process, management, software. 1 INTRODUCTION 1.1 Needs for performance approach Methodologies to manage construction projects from design to use are nowadays of greater importance than ever due to increasing complexity of buildings and building processes, rapid changes of user needs and market environment, goals of sustainable development and demands for faster delivery schedules. The buyers in the supply chain have traditionally been regarded as the customers of the construction sector and the performance of the completed building with respect to the needs of the occupants have been of minor interest. New social and technological drivers of the sector have brought along ideals and business models emphasizing the performance of the end product. 1.2 Benefits of performance analysis The performance analysis is a hierarchically organized set of characteristics with which a building concept can be qualitatively described and technical and/or economic target specifications established. Performance analysis may be used as an integrative platform for different experts to make common decisions. For different building concepts and projects, a generic model of hierarchical performance objectives can be used to systematically analyse the different goals. The model becomes an advanced development and evaluation tool with concrete information and specific methods. Software solutions help to identify the customers’ needs, analyse the performance objectives and manage the phases of design process. 2 PERFORMANCE ANALYSIS 2.1 Application of value tree analysis For the management of the buildings performance, the value tree analysis offers a logical way to organize the various performance objectives, to evaluate their value and relations, to generate technical criteria and potential solutions and to incorporate rating and verification methods in the framework (Figure 1). 425

REQUIREMENTS A conformity A1 LOCATION A2 SPATIAL SYSTEMS A3 SERVICES

ENVIRONMENTAL PRESSURE

B performance

owner user society

B1 INDOOR CONDITIONS B2 SERVICE LIFE B3 ADAPTABILITY B4 SAFETY B5 COMFORT B6 ACCESSIBILITY B7 USABILITY

PERFORMANCE

DESIGN CONSTRUCTION

C life cycle costs and environmental pressure

USE AND MAINTENANCE

C1 LIFE CYCLE COSTS C2 ENVIRONMENTAL PRESSURE

COSTS

Figure 1. From objectives towards evaluation.

The building performance analysis from the point of view of fitness for use, or fitness for purpose, consists of the following steps: – Gathering qualitative objectives of a building and its immediate surroundings – Defining the performance objectives to levels so that corresponding technical solutions can be designed – Selection of evaluation methods, including environmental assessment and cost estimations. For a building, the performance objectives are often interrelated which call for an experienced use of the analysis methodology. 2.2 Hierarchy of performance objectives The buildings performance may be analyzed from different directions like from the social and cultural, economic, ecologic or usability viewpoints. The general methodology for a performance analysis provides applicability for different building concepts. The main purpose for a generic hierarchical model is to provide a common platform to define the desired qualities of a building and to develop a common language for different disciplines as well as to serve as a basis for development of design and technical solutions. The choice of the objectives in the hierarchical presentation shows also to some extent the values of the developer. In the performance classification developed at VTT (VTT ProP®), the hierarchical division of building’s properties is based on three main categories that are (A) Conformity, (B) Performance and (C) Cost and environmental properties. Their subdivision is as follows: A Conformity

B Performance

A1 Location A2 Spatial systems A3 Services

B1 Indoor conditions B2 Service life and deterioration risk B3 Adaptability B4 Safety B5 Comfort B6 Accessibility B7 Usability C Cost and environmental properties C1 Life cycle costs C2 Environmental pressure 426

The subdivision of main categories is divided further into sub-objectives when necessary. For building’s performance, the sub-objectives are: B1 Indoor conditions B1.1 Indoor climate B1.2 Acoustics B1.3 Illumination B1.4 Vibration conditions B4 Safety B4.1 Structural safety B4.2 Fire safety B4.3 Safety in use B4.4 Intrusion safety B4.5 Natural catastrophes The regulated requirements of a building have to be incorporated in the performance analysis. The safety and comfort of a building for its users and neighborhood are issues that most of the national building regulations deal with. Minimum levels for design are given for different technical properties. Also, the Construction Products Directive (89/106/EEC) of the European Community constitutes both the general and specific criteria with which construction works must comply. The criteria comprise six essential requirements (ER’s) each including durability: – – – – – –

ER1, mechanical resistance and stability ER2, safety in case of fire ER3, hygiene, health and the environment ER4, safety in use ER5, protection against noise ER6, energy economy and heat retention. For life cycle costs and environmental pressure, the sub-objectives are: C1 Life cycle costs C1.1 Investment C1.2 Operation C1.3 Maintenance C1.4 Demolition and disposal C2 Environmental pressure C4.1 Biodiversity C4.2 Resources C4.3 Emissions

The user of the performance analysis defines the target quality for different performance objectives. Commonly, the minimum levels of building code requirements are seen as the lowest possibility. Wider range of levels is possible for properties that are not covered by regulations. 2.3 Decision-making procedures The hierarchy of performance objectives opens many possibilities to develop the decision-making procedures of the actors in the construction sectors. At a general level, e.g. methods of quality function deployment can be used to make a priority list of targets and to evaluate design solutions. 427

Figure 2. Virtual building technologies can be used to support decision-making.

Based on the hierarchy of performance objectives and their targeted qualities, alternate design and technical solutions can be developed. The capability of different solutions to fulfil the performance criteria can be studied with verification methods. Verification methods are nowadays most often various simulation programs which handle large input data and use theoretically sound formulae. It is possible to get estimations for the most important objectives already in the initial design phases and to compare the influence of alternate technical solutions. For traditional technical and structural systems, expert knowledge is often enough for an estimation of quality of performance objectives. Rough cost estimation can be done in the initial design phase based on few variables of design. The developers and contractors have long collected pre- and post-calculated costs of their projects. The more detailed the design is, the more accurate cost estimation can be made. The amount of verification methods of single products (structures or devices) or technical systems is numerous. They may be needed already in the start of a building project, especially when new solutions are to be studied. Visualization (Virtual Reality) has long been seen as an important way to overcome the problems of different languages of different disciplines and especially the problems associated with the user’s involvement in building design processes. Novel software applications provide visualization to present the results of simulation, often with the possibility to visualize also the effects of changes dependent on time or other variables (Figure 2). Visualization is already extensively used in design and manufacture of different technical and structural systems of a building, and the predictable step is combination of these in order to evaluate the integrated performance. In this way, visualization can compensate the prototyping to some extent that is an easy industrial way to survey the users. 2.4 Cost-efficiency and eco-efficiency The performance analysis is a basis for a toolkit of calculation and assessment methods with which the performance objectives are transformed to a part of the cost-efficiency and eco-efficiency. The cost-efficiency of a building still is the most common principle when the basic design and main contractors are selected. The performance analysis adds qualitative properties of the completed building to the evaluation. More often the costs during the use of the building are also 428

included into the evaluation process. The cost-efficiency can be presented as

The eco-efficiency of a building is defined in Finland as follows: The eco-efficiency of a building is the ratio of performance and conformity to the environmental pressure induced by the spaces and technical solutions that fulfil the client’s requirements. The requirements cover both the required performance as well as the conformity in terms of location, rooms and services. The environmental impact is divided into three classes, which are the use of natural resources, harmful emissions and land use.

The eco-efficiency indicators have their starting point in the generic and internationally accepted concepts. 2.5 Support for product development The performance of a completed building is a result from the performance of different structural and technical systems and their interaction and influences on each others. In the future buildings, the various technical systems will be developed towards an intelligent building whose performance clearly results from integrated actions. The performance analysis together with verification methods can be applied for gathering needs and generating new ideas. For a construction product development, the reasoning with fitness for use instead of fitness for manufacture is a new kind of approach. Verification methods of single technical and structural systems and components include analysis and simulation software, design methods, experimental research and standardized testing methods. The systematic approach developed for CE-marking of construction products gives guidelines to plan the necessary project tasks after thorough market surveillance. 3 SOFTWARE TOOLS 3.1 Verification and evaluation tools At VTT, several software programs are developed that can be used for performance analysis of building products, facilities or built environment. Examples of such tools can be listed: EcoBoxx (a toolbox for the performance analysis of a community), VTT Talo (a test bed platform where different simulation tools are integrated), WinEtana (energy estimation) or LCA House (environmental assessment of building systems, materials and products). 3.2 EcoProP The EcoProP software, that is a standalone Windows application, has been developed at VTT based on the VTT ProP® performance classification. The software tool is aimed to serve in establishing the specifications that can be incorporated in design documents as word or HTML files. Content in the database has been implemented and tested for specific purposes, like different building types amongst the building stock. This software is mainly used by owners, developers and project management consultants in individual projects. EcoProP comprises a database of performance requirements and easy-to-use interface to the database. There are a number of requirements definition sets, which correspond to the possible requirements of different project types. The user can select from one to five pre-set performance 429

levels for each requirement and then add own comments. The application provides estimates the life cycle costs of the building. This analysis is based on the cost factors associated with different performance levels and the baseline information of the project. EcoProP can be used to manage performance requirements for new building projects, and it can also be used to some extent for evaluation the performance levels of existing buildings.

4 EXPERIENCES FROM SYSTEMATIC PERFORMANCE REQUIREMENTS MANAGEMENT WITH ECOPROP EcoProP has proven to be a valuable aid in implementing the performance approach, because the users are ‘forced’ to think their objectives through, before jumping into the technical solutions. EcoProP can be used in a team session or one user can set requirements. The team sessions improve the quality of the selected targets and goals of the project since participants challenge each other’s ideas and selections. Also the commitment of the project team members increases. The scope of controlling and managing the requirements has to be seen in a broader context, where customers can link all relevant information from their point of view to software managing performance requirements. This means that decisions can for example lean on to company’s business plan and relevant other information gathered. It leads to more transparent decisions and combine varied approaches which for example owner, user, designer and architect have.

5 CONCLUSIONS The performance based building provides a methodology to improve the quality of buildings from the point of view of fitness for use or fitness for purpose. Gathering, analyzing and evaluating the needs of different users of a building create the general framework for a performance analysis. Transforming functionality-based requirements into appropriate technical solutions will mean also new business models. The hierarchy of performance objectives and software tools developed at VTT give a strong support to the starting phase of a building project. They can also be applied throughout a building process and in the product development. REFERENCES EcoProp 2004. Espoo: VTT Building and Transport, Brochure. 2 p. Huovila, P. & Leinonen, J. 2001. Managing performance in the built environment. CIB World Building Congress 2001 – Performance in Product and Practice. Wellington, NZ, 2–6 April 2001. CIB . 8 p. Häkkinen, T. et al. 2002. Eco-efficiency in the building and real estate sector. Helsinki: Ministry of the Environment, Housing and Building Department. 165 p. (in Finnish) Leinonen, J. & Huovila, P. 2001. Requirements management tool as a catalyst for communication (2001). 2nd Worldwide ECCE Symposium. Information and Communication Technology in the Practice of Building and Civil Engineering. Espoo, FI, 6–8 June 2001. Association of Finnish Civil Engineers RIL. pp. 105–110.

430

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Effective use of cold-formed steel structures for low-story urban buildings E.L. Airumyan Melnikov Central Research and Design Institute of Steel Structures, Moscow, Russia

O.I. Boiko Exergia firm, Lipetsk, Russia

S.V. Kamynin Taldom Profile Trading House, Moscow, Russia

ABSTRACT: The paper considers some examples of economical steel structures developed in Russia using zinc-coated cold-formed sections for frames of buildings with height not exceeding 10 m. Structural material is coiled steel with the thickness range not exceeding 1.2–2.0 mm and double-side zinc coating of at least 18 µ. Cold-formed channel, C, Z and -shaped sections of increased stiffness with height varying from 100 to 280 mm are fabricated from this steel by a continuous rolling. Such sections can be used to assemble frames, columns, girths, trusses, beams and purlins. Connections of cold-formed sections and structural members are made using normal bolts, self-tapping screws and combined rivets. Of particular interest are load-bearing structures made of perforated wall sections allowing to greatly increase their thermal resistance within cladding structures. 1 INTRODUCTION Over the recent 8–10 years a new branch of construction industry has been formed in Russia: production of zinc-coated steel cold-formed sections. Taking into account the growing importance of this developing industry a National Association of Makers of Steel Cold-Formed Sections (NAMSCS) has been established incorporating more than 30 firms which produce about 1 million tonnes of zinc-coated steel cold-formed sections per year. The area of the widest structural application covers various-purpose lightweight and cladding structures for buildings and constructions used in all regions of Russia, including seismic regions and those located within the Polar circle. In civil engineering the most effective are zinc-coated cold-formed structures used in seismic regions, those used for low-story cottages delivered as complete sets and also for reconstruction of buildings: buildings of added mansard stories, creation of ventilated façades and replacement of flat roll roof covering with low-slope metallic roof panels with waterproof joints. The use of such structures instead of traditional ones made of reinforced concrete, brick, wood or rolled steel plates yields a considerable economic effect in the above industry owing to a reduction of dead and seismic loads, transportation costs and labor consumption during erection works, and of construction time as well. Below are given some examples of a successful realization of advantages made of zinc-coated steel cold-formed sections during construction and reconstruction of a number of civil engineering buildings. 2 FRAME BUILDINGS For fabrication of frames for buildings a production range of standard cold-formed sections is used with 3 types of cross-sections: channel, C and Z-shaped. The height of sections varies from 100 to 431

100-300

(c)

1-1 100-200

(b)

1

100-400

(a)

1

Figure 1. Types of cold-formed steel sections: (a) – flat web, (b) – increased stiffness web, (c) – perforated web. (a)

(b)

Figure 2. Two-storey cottage with mansard: (a) – erection of cold-formed section frame, (b) – general view of constructed cottage.

300 mm. To increase stiffness of sections subjected to local loads and torsion their walls are given

-shape while profiling (Fig. 1). The sections are fabricated from 0.7 to 2.0 mm thick coiled zinc-coated steel with the yield strength of 250 to 350 MPa and relative elongation of at least 18%. The overall mass of zinc coated on both surfaces of coiled steel shall be not less than 414 g per square meter, thus providing section corrosion resistance for at least 25 years provided that the frame is in service in non-aggressive or slightly aggressive environments. In order to make higher corrosion resistance of the frame elements sections may be fabricated from zinc-coated steel with applied polymeric or varnish coatings. For decreasing heat conductivity of cold-formed sections used within cladding structures walls are perforated during profiling to produce longitudinal slots. Heat conductivity of perforated sections (or “thermo-profiles”) is comparable with that of wood elements with the same cross-section area. 2 systems of outside walls were developed for low-storey walls: – load-bearing walls with the frame made of thermoprofiles; – plate assembly self-supporting walls or those using panels. Load-bearing outside walls comprise: – 0.7–1.5 mm thick zinc-coated steel perforated sections forming vertical supports and girths which are interconnected by means of self-tapping screws (Fig. 2); – effective incombustible insulation material (mineral wool, basalt, glass-fiber panels); – TYVEK-type film vapor seal; – gypsum wallboards used for sheeting; – outside facing using brick, siding sheets, wood panels with an air gap provided from the insulation material surface. 432

The wall thickness varies from 150 to 250 mm when the reduced resistance of heat transfer is of 3.2 to 5.1 m2 . ◦ C/Bt. The internal frame of a building is made of twin solid-wall sections forming supports with double-tee or box cross-sections. Beams for intermediate floors are fabricated using single or twin C-shaped sections with height varying from 150 to 300 mm which are connected to columns by bolts. A corrugated steel deck is placed on beams providing out-of-plane stability of beam upper flanges and serves as a stiffening diaphragm which replaces horizontal braces. The deck should be attached to supports using self-tapping screws for each corrugation. An attic floor includes a steel frame, diagonal braces, boarded ceiling and mineral wool insulation panels. The frame of the floor is made using 600 mm spaced thermo-profiles and purlins for the boarded ceiling on which the insulation material is placed. Load-bearing roof structures consist of roof trusses or beams which are made of zinc-coated cold-formed sections. 6–15 m span roof systems are used as follows: – simple and double-web triangular trusses; – trusses with parallel chords; – two-slant beams with ties. For fabrication of elements of roof trusses and beams channel, C and Z-shaped cold-formed sections are used. Trusses are fabricated with symmetrical cross-sections in relation to a vertical plane with attachment of lattice elements to the chords. As a rule, truss chords are made of single sections, and for lattice elements single or twin C-shaped sections are used. While designing trusses their elements are optimized for the shape and cross-section area so to minimize their overall weight. Considering that there are eccentricities in joints, chords and lattice elements are designed with account for bending moments. For out-of-joint loading chords are designed for a combined action of longitudinal forces and bending moments as provided for in Eurocode 3 (1996) and AISI Standard (2001). Element connections of roof structures are made using self-tapping screws the number of which is determined by a calculation. In the case when the size of section flanges and walls is insufficient for placing a required number of screws or bolts, the design of particular structural joints provides for 2–6 mm thick gussets and cover plates. Roof structure joints bearing on the frame supports can be designed as hinged or rigid. 3 FRAMELESS ARCH ROOFS The unique technology of fabrication and erection of frameless arch buildings using K-span steel cold-formed sections is well-known. It is being permanently improved and is widely used in Russia. Over the last 5–6 years this technology has been used for construction of about 60 buildings with spans of 12–24 m (roofed markets, railway station buildings, recreation centers, garages and cold store buildings). For metal consumption and construction time such buildings are much more cost-efficient as compared to similar ones (foreign or domestic) with the use of frame-type steel structures. Cold-formed sections performing load-bearing and cladding functions in frameless buildings are made from 0.8 to 1.2 mm thick zinc-coated steel on a construction site using two movable forming and curving machines. One of the machines forms a straight -shaped section while the other curves it on a given radius (not less than 2 m). The length of the section made is not practically limited, and for an arch roof, as a rule, it corresponds to the arc length of its crosssection. For this reason sections without transverse joints overlap entirely a building span. Sections are interconnected along longitudinal edges with a fold seaming machine without using fasteners. Walls and partitions in a building having a roof of this design are made of straight cold-formed sections which are manufactured and connected using the above described technique (Fig. 3). The field of application for such buildings depends on their sizes, design loads, temperature and moisture conditions, degree of environment aggressiveness and fire safety requirements. Strength, safety and effectiveness of such a structure depends greatly on how much accurately the design 433

(a)

(b)

Figure 3. Fabrication and erection techniques of steel roof shells: 1 – profiling machines, 2 – coiled steel, 3 – straight section forming unit, 4 – section curving unit, 5,6 – erection of panels.

Figure 4. Analysis and experimental deflections of arch specimens: a – unsymmetrical test loading, b – test results.

will consider specific structural behavior due to section wall thinness, surface corrugation of section faces after curving, final stiffness of folded joints, increased deformability of an arch roof (especially, for the case of unsymmetrical loading). The Melnikov Institute developed “BIPLAN” program which allows to evaluate enough accurately deformability and load-bearing capacity of arch roofs of the described design accounting for specific features of structural nonlinear behavior for various combinations of design loads, including snow, wind and seismic loads. As a basis for design a finite element method was used considering structural geometric nonlinearity, structural form initial imperfections (for example, corrugations on section faces), action of end walls and other factors influencing the behavior of roof shells. With the help of this program areas of rational use for arch roofs of the said type were defined depending on span ratios, curvature radius, material thickness, distances between walls or transverse load-bearing partitions and design loads by Airumyan & Yemelin (1998) (Fig. 4). Recommendations for design of frameless arch buildings developed by the Melnikov Institute were used in construction practice (Fig. 5).

4 BUILDING RECONSTRUCTION One of the most important tasks of reconstructing the existing buildings is to provide an additional heat insulation of outside walls and roof coverings which do not meet more stringent requirements for energy savings. For solution of this task involving thin-walled cold-formed sections the system known as a “ventilated façade” is used. The description of the system is given below. Steel sliding brackets are attached to the outside wall surface protruding at a required distance. Between the brackets insulation material panels are fixed to the wall using self-tapping stop elements with plastic press washers. Then parallel guides made of channel or C-shaped cold-formed sections are 434

(a)

(b)

Figure 5. Buildings with frameless arch roof: a – sports complex, b – trading center with spans of 25 and 20 m, respectively.

connected to bracket free ends. The distance between the insulation outside surface and the plane of guides is within 50–80 mm. For façade facing metallic or plastic siding sheets are used providing a naturally ventilated air gap that gives the name to the whole system. All frame elements of the system are made from 0.8 to 1.2 mm thick zinc-coated steel and these are connected with the use of self-tapping screws. The system is erected without using cranes (Fig. 6). This makes it possible to carry out works without resettling people that is of growing social importance. The reconstruction of a roof includes the replacement of roll or other damaged roofing sheets with low-slope metallic ones. Metallic roofing sheets are erected without disrupting the integrity of the existing roll carpet, except for places needed for installation of roof supporting columns. The columns are made of C-shaped single or twin cold-formed sections with height of 100–150 mm and are installed with spacing of 2.5 to 3.0 m. Column bases are fabricated using rolled steel angles to be fixed to a concrete layer or slabs of the existing top floor by means of 150–200 mm long anchor bolts. The column height is taken depending on a required thickness of additional insulation material and a 30–35 mm gap provided for natural ventilation in the space between roofing sheets and the surface of insulation material. Strings made of 0.8–1.0 mm thick steel twin cold-formed sections with channel cross-section and height of 100 mm are attached to the columns. These are located along a roof slope with 1.0–1.5 m spacing. Roof purlin elements are fixed to the strings. The elements are -shaped cold-formed sections with height of 40 mm and spacing of 300–500 mm, except for 1.0 m wide areas located along the roofing perimeter, where the spacing is reduced to 0.25 m since for those areas the design load of wind suction doubles in accordance with design code requirements. For fabrication of roofing sheets cassette-type sheet cold-formed sections made of 0.55–0.6 mm thick zinc-coated coiled steel are used. The above steel has a varnish coating on the outside surface and a priming coating on the inside surface. Sheet sections are placed on cold-formed roof purlins which are 300–600 spaced depending on snow loading and these are attached to the purlins with zinc-coated steel cramps with thickness not exceeding 1.0 mm (Fig. 7). Roofing sheets are interconnected along their longitudinal edges with a fold seaming machine forming a double welt with a simultaneous fixing of cramps in a joint. Tests showed that this joint provided complete water tightness of sheet connections without using a sealing material for a roof slope of at least 7%. For smaller slopes sealing materials used as pastes and mastics are added in longitudinal joints. When a roof slope length is great (more than 12 m), in order to exclude joints in roofing sheets which are located across the slope, the sheets are manufactured directly on a construction site and their length is taken as not less than the slope length. For this purpose section forming machine with a mass of about 7 t is installed directly on the building top floor or on an inclined platform located at 15–20 m distance from the building under 435

Figure 6. Erection of ventilated façade for 9-storey residential building.

Figure 7. Erection of cold-formed steel section frame for low-slope metallic roof of residential building.

Figure 8. Mansard built over the existing building with flat roof.

Figure 9. Erection of mansard cold-formed section frame using enlarged blocks.

reconstruction. Roofing sheets are rolled in a continuous way and then are delivered to a conveyor gallery provided with roller beds. Rolled cassette sections are fed as a band to guide rollers which are located along the roof slope. When the length of a section band reaches the width of a building, further profiling is suspended and a roofing sheet is cut at the roof ridge and eaves on the roofing site. The cutoff parts of roofing sheets are placed along the roof slopes, then interconnected along longitudinal edges and at the same time these are attached to the roof purlin elements using steel cramps. Construction practice is familiar with the cases when the length of roofing slopes made as profiled sheets using the above described technique reached 108 m without transverse joints. In Russia mansard construction is especially urgent because about 20% of residential buildings which are in use for over 40 years need to be reconstructed. Construction of one or two-storey mansards will allow not only to give a modern architectural appearance to these buildings, but also will provide additional flats and penthouses (Fig. 8). An 436

original structural scheme of mansard load-bearing structures will allow to assemble them on site using zinc-coated cold-formed sections and erect by enlarged sections with no use of heavy-duty load-lifting equipment (Fig. 9). All the connections of the above structures are made using selftapping screws, combined rivets or bolts. Welding of such structural elements is not permitted. This erection method will allow to reconstruct buildings in short time without evacuating people from below located stories.

5 CONCLUSION The fabrication and use of zinc-coated cold-formed sections in mass production civil engineering becomes one the priorities of construction industry development, being in keen demand in Russia because of their evident advantages. REFERENCES Airumyan, E. & Yemelin, E. 1998. Analysis and tests of frameless roof shells of cold-formed steel sections. Proceedings of International Congress ICSS-98, Moscow. AISI Standard. North American Specification for the Design of Cold-formed Steel Structural Members. Washington, 2001. Eurocode 3. Design of Steel Structures. Part 1.3. Supplementary rules for cold-formed thin gauge members and sheets. ENV, 1996.

437

Session 8: Research and development concerning mixed building technologies

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

On design of composite beams with concrete cracking J. Bujnak & J. Odrobinak University of Zilina, Faculty of Civil Engineering, Zilina, Slovakia

ABSTRACT: The paper deals with behaviour of steel-concrete composite girders under negative moment. The experimental measurements on the specimens of composite girder acting with slab under tension are presented. Further, the non-linear analysis of 3D finite element analysis has been executed. Experimental results are then confronted with values obtained from both the non-linear analysis and a simplified design method.

1 CONCRETE CRACKING CONSIDERATION 1.1 Introduction Composite steel-concrete girders are able to resist effectively mainly positive sagging moments. However, in a negative hogging region they are less structurally perform due to concrete cracking. In spite of these phenomena in these regions, composite beams continue to manifest normal structural behaviour. But cracks in concrete and tension stiffening should be carefully considered in the analysis of composite structure. The non-linear computer calculation taking into account concrete cracking and tension stiffening is required for more precise simulation of such problems. If negative bending regions are occurring, static analysis should be recalculated on the updated model due to change of the flexural stiffness of cross-sections through the length of girder. Anyway, there is uncomfortable to do such complicated and time-consuming numerical calculation in a practical design. 1.2 Proposed methodology The Eurocode 4 (2002) suggests replacing the non-linear analysis by at least two-step procedure. From the practical point of view, however, it seems to be more reasonable to proceed by the same calculation computer model and to use idealised cross-sections by considering transformed area of concrete. This requirement can be fulfilled providing that such construction will be reloaded by additional deformations. If no longitudinal restraints are present, only additional rotations due to changes in the curvature of unit elements have to be considered. The additional angle deformation ω [rad · m−1 ] due to tension stiffening can be derived on the basis of stress redistribution from the concrete part to the steel beam as follows

where the value Ea is modulus of elasticity of the steel, I1 (x) and I2 (x) represent the second moment of area of the composite section including the concrete part and neglecting the concrete, respectively. The necessary variable ξ(x) resents the ratio Mcr (x)/M(x), where Mcr (x) and M(x) is the cracking moment and the actual bending moment, respectively. Effect of tension stiffening is than successively investigated by means of additional load. In such calculation, the regions where bending moment exceeds cracking moment (M > Mcr ), the 441

1000

400

325

300

IPE 300 - 2600

1000

40 x 40-70

2400 2600

100

150

70

70

300

325

100 800

Figure 1. Dimensions and layout of the beams.

P 12x300

330

100

P 10x100

330

100

500

P 20x200 250

850

1300

850 250

3500

Figure 2. Dimensions and layout of a beam.

construction should be reloaded by the corresponding rotation load according to (1). In the case of statically determinate girders, only deflection is modified. Statically indeterminate continuous girders are additionally subjected to secondary bending moment (Bujnak et al. 2003).

2 EXPERIMENTAL RESEARCH The experiment has been mainly focused on the influence of concrete cracking on composite girder behaviour under stepped load inducing a tensile stress in the slab. For this purpose, two sets of physical models of composite girders were prepared. Firstly, the tested models of composite beams have acted with the slab under tension (Bujnak & Odrobinak 2004). Main dimensions and a static scheme are showed in Figure 1. Rigid block shear connectors have been used for transferring longitudinal shear flow between steel and concrete part of cross-section. Thus, an eventual influence of the slip on final strains was minimised. Alternatively, two pairs of models, which have acted as simply supported beam with overhanging cantilevers loaded at those overhangs, are analyzed (Odrobinak & Bujnak 2003). For transferring shear flow between steel flange and RC slab, the stud connectors φ16 mm (beams N09 and N10) and φ10 mm (beams N11 and N12) were used in each pair of beams, respectively. A scheme of testing models is given in the Figure 2. The built-up welded unsymmetrical girder was made of steel S 235. Usual reinforcement bars in concrete slab of standard quality C30 was used. All essential material characteristics were tested on specimens in laboratory. The cracking moment was evaluate as Mcr = 47,25 kNm. Elastic and plastic ultimate moment have values Mu,el = 200,79 kNm and Mu,pl = 297,57 kNm, respectively. 442

Figure 3. One half of the theoretical model in FEM, overall layout and reinforcement. Reinforcement

σ[MPa]

400 300 200 100

500 400 300 200

σ[MPa]

500 σ[MPa]

Concrete in com-

700 600

100 0

0 0

50 100 1000 x ε

150

0

50

100

150

1000 x ε

-45 -40 -35 -30 -25 -20 -15 -10 -5 0

σ[MPa]

Steel beam 600

-0,5

-1,5 -2,5 1000 x ε

-3,5

Concrete in tension 4,0 fct 3,5 3,0 2,5 0,6 fct 2,0 1,5 1,0 0,5 0,0 0,0 0,1 0,2 0,3 0,4 0,5 0,6 1000 x ε

Figure 4. Stress-strain curves for material models.

3 NONLINEAR COMPUTER FEM ANALYSIS In accordance with described tests, the non-linear numerical computer simulation, based on finite element method (FEM) has been executed in ANSYS program (2003). In the Figure 3, the numerical model of simply supported members with the concrete slab connected to the bottom flange of steel rolled section is illustrated. The real geometrical characteristics and load procedure were implemented into 3D finite element model in system ANSYS. Material non-linearity and actual stress-strain curves were considered for all the three used materials: steel, concrete and reinforcement. In concrete elements description, both cracking and tension stiffening have been taken into account during the computer analysis. Stress-strains relationships for material behaviour models used in the numerical calculations are presented in the Figure 4. In spite of using symmetry for saving computer time and memory space, the pure complete calculation reaching the beam failure necessitated nearly 200 hours.

4 RESULT PROCESSING 4.1 Simply supported members with the bottom concrete slab under tension Only a part of the results can be discussed in this contribution. The deflection of beams is shown in the Figure 5. Theoretically predicted deflections according to Eurocode 4 (2002) are designated as “EC4”. Deformations determined on the basis of the method described in the paragraph 1 are signed as “ω -method”. It can be probably stated more significant influence of shear on the vertical deformation than given value. Shear deformation has in the case of unusual high depth of composite cross-section important effect. Therefore, the effect of shear was taken into account in theoretical description of the overall vertical deformation. 443

250

N1

M [kNm]

200

N2

150 EC4 100 ω-method 50 0 -0,5

ANSYS 0,5

1,5

2,5

3,5 4,5 δ [mm]

5,5

6,5

7,5

Figure 5. Deflection at the mid-span. 50 N1

EI [MNm2]

40 30

N2

20

EC4

10

ω-method

0 0

50

100 150 M [kNm]

200

250

Figure 6. Flexural stiffness of the beam.

The flexural behaviour might be more appropriately illustrated by flexural stiffness EI modification. The dependence of stiffness decreasing as function of bending moment is presented in the Figure 6. Values received from measured deflections are compared with two prediction methods. From comparison of the curves, it can be concluded that approach based on ω-method fit rather good to the experiment results. A visible initial crack has been found at the girder slab of the model marked as N2 before loading. In this way, the absence of sharp change in the measured deflection development can be explained. An ability of the beam to transfer negative bending has been dissipating even before theoretically predicted first crack. The lines in the Figure 7 compare the experimentally and theoretically predicted strains at the top and bottom flange of the rolled IPE beam in the middle span cross-section. Basically, the good correlation in development of compression strains at the top fibres can be concluded. Prediction of bottom tensile strains is, especially at higher load levels, not enough accurate. Anyway, bottom flange can be markedly influenced by shear flow between steel and concrete part of the crosssection. Unsteady variation of stresses due to concrete cracking has essential importance on stress redistribution as well. Our simplified method overestimates load carrying capacity of the tested girders. However, in the calculation using this ω-method, it is impossible to consider several non-linear effects. The main influence is probably caused by bond changing between reinforcement and concrete and its complete degradation during cracking. In the Figure 8, the strains at the top surface of concrete slab are showed. For the measurement of experimental strain values, mechanical gauges were used. In the case of beam designated marked as N1, only four measurements have been executed. Two output lines of the numerical computer analysis are showed in the graphs. They correspond to the mean longitudinal experimental strains with the base length of 180 mm and 420 mm, respectively. 444

N1 N2 ANSYS ω-method -2500

-2000

-1500

-1000

-500

0

300

300

250

250

200

200

150

150

100

100

50

50

0

0

(b) M [kNm]

M [kNm]

(a)

N1 N2 ANSYS ω-method 0

1000

2000

ε[106]

3000 ε[106]

Figure 7. Strains at the rolled IPE beam: (a) top flange, (b) bottom flange. 250 N2

M [kNm]

200 150

N1

100

ANSYS (180 mm)

50 ANSYS (420 mm) 0 0

500

1000

1500

ε[106]

Figure 8. Strains at the top surface of concrete slab measured by mechanical strain gauges compared with non-linear analysis.

The presented relationship in the Figure 8 represents mean strains including the width of cracks inside the base length. Therefore, it coincides with mean strain of reinforcement, because the slip between concrete and reinforcement bars was not included into non-linear analysis. 4.2 Beam with overhanging cantilevers For assessment of flexural behaviour prediction of the models their deflection development can be compared. Deflections at the mid-span of the beams as well as at their overhanging cantilevers under loading point are shown in the Figure 9. Theoretically predicted deflections according to Eurocode 4 (2002) are marked as “EC4”. Deformations determined on the basis of the methodology described above are signed as “ω-method”. In the region of pure bending, both calculation approaches overestimate deflection. From the Figure 9, it is evident that each beam model reaches slightly greater deflection of cantilever than predictions. After detailed analyse, it can be stated that this difference is probably caused by more significant influence of shear on vertical deformation than regarding value. Shear deformation has in the case of unusually small depth of composite cross-section important effect. Therefore, complex flexural behaviour should take into account also effect of shear. Naturally, when a bending moment exceeds elastic ultimate moment, a deflection of the girders is becoming more non-liner due to plastic deformations in the most stressed fibres. The plastic behaviour in the flanges of steel girder is evident from the Figure 10, presenting strains development. Strains in the bottom fibres (compression) are predicted with good accuracy. The top flange (tension) is in direct contact with concrete part of cross-section through shear connectors. Therefore, differences between theoretical and experimental strains are more obvious. An initial micro cracks due to shrinkage also influence strains in steel girder, mainly at the steel-concrete interface. 445

EC4

N09 - studs 16

N11 - studs 10

EC4

N12 - studs 10

ω-method 400

N10 - studs 16

N12 - studs 10

ω-method

350 300

-3,0

-2,5

-2,0

-1,5

-1,0

-0,5

0,0

M [kNm]

N11 - studs 10

N10 - studs 16

M [kNm]

N09 - studs 16

400 350 300

250

250

200

200

150

150

100

100

50

50

0 0,5

0 0,0

2,5

5,0

7,5

10,0

12,5

15,0

17,5

δ[mm]

20,0 δ[mm]

350 300

400 350 300

N09

250

250

N10

200

200

N11

150

150

N12

100

100

50

50

0

0

theory -2000

M [kNm]

400

M [kNm]

Figure 9. Deflection at the mid-span and at the cantilever under loading point.

-1500

-1000

-500

0

N09 N10 ;

N12 theory 0

ε . 106

N11

1000

2000

3000

4000 ε . 106

Figure 10. Strains at the bottom and the top flange of steel girder at the mid-span cross-section.

5 CONCLUSIONS It can be observed good approximation of the time-dependent modification of strains by computer FEM modelling to the experimental values received by experimental research. In non-linear modelling with consideration of concrete cracking and tension stiffening, a good correlation of results can also be confirmed. Our simplified ω-method for stiffness changing due to concrete cracking can be seen as practical and easy for use in the most of common commercial CAD-FEM programs. The results obtained by this method are in a quite good coincidence with the real behaviour of composite girders. Moreover, application of the method can approximate complex processes in the concrete deck without need for a complicated and time consuming non-linear analysis of practically obvious steel-concrete composite bridges. REFERENCES ANSYS ver. 5.7 – Help & Theory Manual. SAS IP, Inc., 2003. Bujnak, J., Furtak, K. & Odrobinak, J. 2003. Influence of Tension Stiffening of Concrete on Flexural Behaviour of Composite Girders under Negative Bending. In Science, Education and Society at the Edge of Millenium„Proc. 11th International Scientific Conference. Section No. 1 Engineering, Zilina 17–18 September 2003, University of Zilina edition. Bujnak, J. & Odrobinak, J. 2004. Composite Girder with Concrete Slab under Tension. In Betonarske dni 2004 Proc,National conference, Bratislava 9–10 September 2004. University of Bratislava edition. EUROCODE 4 – STN P ENV 1994-2 2002 Design of Composite Steel and Concrete Structures – Part 2: Composite Bridges.CEN Brussels. Odrobinak, J. & Bujnak, J. 2002 Behaviour of Composite Steel and Concrete Beam under Negative Bending. In TRANSCOM 2003, Proc. the 5th European Conference ofYoung Research and Science Workers in Transport and Telecommunications, Zilina 15–18 June, University of Zilina edition.

446

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Design methodology of profiled steel sheets for composite slabs by FEM M. Ferrer, F. Marimon & F. Roure Dep. of Strength of Materials and Structural Engineering, Barcelona School of Engineering, Technical University of Catalonia, Spain

ABSTRACT: A methodology for building 3D non-linear finite element models has been defined, validated and applied, to simulate the longitudinal sliding failure mechanisms of composite slabs, for several commercial and own designs. Sensibility of shear resistance is analysed, mainly in relation to: friction coefficient, embossing depth and slope, shell thickness, inclination angle and length of embossment and profiling angle of rib shape. The model simulates locally the longitudinal shear failure in “pull-out” test. Concrete is treated as a rigid contact surface; steel is implemented using a multilinear plasticity model. Predominant effect is sliding with friction; so contact elements are implemented with appropriate friction coefficient. Conclusions of the parametrical analyses are summarized, regarding to as design as general behaviour matters.

1 INTRODUCTION Resistant failure of composite slabs is generally due to longitudinal slip between steel sheet and concrete – except for long span and very short span slabs. This slip is associated to shear forces developed in simple bending. In order to improve the composite slab strength for shear failure mode, current designs of cold-formed steel sheet present a repeated embossment pattern, of several shapes and sizes, all along the span of the composite slab. The present paper is the fourth one of a series, devoted to improve the understanding of mechanical behaviour of composite slabs and to implement a validated procedure for building numerical models; in particular, the local interaction between steel sheet and concrete. Thus, a FE methodology is established to design and improve steel sheets by analysing the sensibility of shear resistance in relation to: embossing depth and slope, sheet thickness, tilting angle and length of embossment, profiling angle of rib shape (from trapezoidal to dovetailed) and any other geometrical or physical parameter.

2 THE MECHANICAL BASIS For composite slabs, longitudinal slip between steel sheet and concrete – due to shear forces generated in simple bending – is prevented thanks to embossments on the sheet. Its function is the same as corrugations on reinforcing bars for concrete. However, the resistant mechanism is quite different. In reinforced concrete, wedge effect of corrugations, transforms the longitudinal slip to radial compression of the steel rebar and radial traction of concrete; both phenomena have a very high stiffness. In composite slabs, on the other hand, wedge effect of the embossments on steel, transforms the longitudinal slip to different actions between cold-formed steel sheet and concrete slab, such as transversal bending of steel sheet – especially important because of its very low stiffness – and posterior vertical separation between both elements. 447

Lateral forces, reproducing the effect of own weigh

Figure 1. Set-up of a composite slab bending test.

Figure 2. Pull-out test.

Therefore, the predominant physical phenomenon is sliding contact with friction between steel and concrete surfaces, with pressures induced by wedge effect of embossments. Already in preliminary simulations, yielding of steel was detected to be important and necessary to be implemented and concrete stiffness was proved to be much higher than sheet, so it acts as rigid contacting material.

3 BRIEF DESCRIPTION OF PULL-OUT TEST The test specimen consists of two confronted one-rib forms of composite slab. A thick steel sheet joints and stiffens both forms. Upper traction load is applied to the cold-formed steel sheet prolongation. Steel screwed bars are inserted in concrete and used as bottom traction elements. The Figure 2 shows the pull-out test setting. The pull-out test (Daniels 1988, Crisinel 2004) was designed to reproduce the longitudinal slip behaviour exclusively, on a reduced dimensions test specimen. In spite of obvious differences of behaviour between pull-out test and real slab bending, pull-out test is a very simple and cheap way to evaluate the longitudinal slip strength exclusively.

4 DESCRIPTION OF THE FEM MODEL 4.1 Material modelling Preliminary finite element models (Ferrer et al. 2002) treated concrete (20 node solids) and steel (8 node shell) as linear elastic materials. Obtained results showed yielding of steel to be important and necessary to be implemented using a multilinear elastoplastic model. A multilinear stress– strain curve, obtained from cold-formed steel sheet traction tests, has been introduced for steel. (E = 2.1 × 105 N/mm2 , ν = 0.28, fy = 370 N/mm2 ) On the other hand, concrete stiffness was proved to be much higher than steel sheet (Ferrer et al. 2003), therefore the elastic solid elements mesh were replaced with a rigid contact surface, resulting a much simpler model (Ferrer et al. 2004). Thus it is that concrete failure is avoided (beware of peel-off concrete failure, see 5.1 Embossing Slope). Contact with friction (Baltay 1990) is one of the most significant physical phenomena to be simulated, so contact elements have been implemented between both materials. Two cases have been considered, maximum µ = 0.6 (Veljkovic 1996) and minimum coefficient µ = 0.2 (Shuurman 2001). A model with no friction (µ = 0) has been calculated as well, to evaluate the influence. Neither chemical bond nor steel sheet residual stresses have still been taken into account. 448

Figure 3. Guidelines for FEM (HB design, concrete side view).

4.2 Element types and geometry Testing all available shell element types, modelling small fillet radius, decreasing elements size, modelling concrete as linear elastic, and changing hypothesis for boundary conditions have proved robustness of this finite element model. Simulated results were found to be reliable as well, by correlating with pull-out test results (Ferrer et al. 2003). Finally, the modelling has been carried out as shown in Figure 3. Four nodes thick shell element has been used to model the steel sheet (constant thickness), including shear deflection, first order shape functions and reduced integration option. The concrete is treated as a perfectly rigid contacting surface with friction. Two consecutive embossment patterns are needed for concrete to let steel sliding over it. A linear pre-stressed elastic element has been used to model lateral springs of the pull-out test. 4.3 Boundaries and loads Double longitudinal symmetry has been used to simplify de model. Moreover, the same behaviour has been assumed for all embossment patterns along the slab. Thus, model size and computation time have been reduced significantly. The boundary conditions are those for double longitudinal symmetry and cyclic symmetry of the embossment pattern. The middle thick steel sheet, used to joint the two forms for the pull-out test specimen, is stiff enough to completely fix cold-formed steel sheet transversally at lateral edges. The external loads consist of prescribed displacement in slip direction (Z), on both two cyclic symmetry lines. Therefore, these two lines remain flat and equidistant. 5 PARAMETRICAL ANALYSES RESULTS The aim of this work is to evaluate the dependency of shear resistant mechanics on geometrical design parameters; so, these numerical models have been parametrically built. Figures 4 and 5 shows a sketch and a concrete side perspective view of all commercial rib designs to be analysed with their identification names one own design, called “T80”. An optimised version of “T80” has been developed by this methodology and will be presented shortly (Ferrer et al., in prep). 449

Figure 4. Dovetailed rib analysed profiles.

Figure 5. Open rib analysed profiles.

Figure 6. Shell meshes of all analysed sheets.

The most significant profits of FEM results are the complete description and understanding of the sliding mechanics and its associated stress and strain states (Ferrer, in prep). By way of example, the von Mises stresses maps at the maximum shear load point are showed in Figures 7a–7e. Deformed shapes are five times magnified (QL60 ten times). Many parameters have been analysed for every design. The most significant parameters with regard to sliding resistance are presented below. 5.1 Embossing slope Undoubtedly, this is by far the most significant parameter with regard to the sliding resistance mechanism, and may be the most efficient parameter to be modified for improving the shear 450

Figure 7. Equivalent stress map at the maximum shear force point. (Displacements are magnified).

strength of a given sheet design. The dependency on embossing slope is extremely high, showing an exponential evolution. The steeper the embossment faces are, the more slide-resistant the pattern is. However, it is absolutely necessary to use this advantage carefully, since this improvement of shear resistance is always achieved at expense of ductility. Too steep embossments may be especially dangerous as a brittle peel-off cracking of concrete can occur, leading to a sequential failure from one embossment to the next one, so reducing drastically the shear resistance of the whole slab. On the other hand, there are manufacturing technological aspects, also limiting high slope values, not exposed in this paper. 5.2 Friction coefficient Related literature shows very scattered values of friction coefficient (from 0.2 to 0.6) and adherence stresses between steel and concrete. Simulated results, shows that this phenomenon is significant enough to be the only reason of the scattered results obtained from the inter-laboratory pull-out tests, carried out separately but identically by ICOM-EPFL (Lausanne-CH) and ETSEIB-UPC (Barcelona-ES). The results obtained from FEM simulations of these tests, showing a linear dependency between shear strength and friction coefficient, agree with this hypothesis, since shear resistance can be 451

180

α

(kN) b)

a)

160 140 120 100 80 60 40 20 0

0,7

0,6

0,5

0,4

0,3

0,2

0,1

10

0

0

20

30

40

50

60

70

80

4

4,5

Embossing Slope (º)

Friction Coefficient

180

(kN) d)

c)

f

160 140 120 100 80 60 40 20 0 1,4

1,2

1

0,8

0,6

0,4

0,2

0

0

0,5

1

1,5

Sheet Thickness (mm)

2

2,5

3

3,5

Embossment Depth (mm)

180

e)

f) 160

θ

140

α

120 100 80 60 40 20

130

120

110

100

90

80

70

60

0 50

30

40

50

60

70

80

90

100

Embossment inclination (º)

Profiling angle (º)

Figure 8. Shear Strength (kN) dependencies: a) Friction coefficient, b) Embossing slope, c) Sheet thickness, d) Embossment depth, e) Profiling angle, f) Embossment inclination.

nearly tripled from µ = 0.2 to 0.6 (see figure 8a). A lot of reasons could lead to very different coefficients of friction, such as cement type, dosing of concrete, type and size of aggregates, setting up and curing processes, moisture level, steel surface finish, temperature, etc. Another phenomenon included in FEM models by means of friction coefficient, is the local crushing of concrete around embossment ends, causing eroding grooves that softens the “effective shape” of the embossment. This is a too little scale process to be simulated explicitly in the presented FEM models. 452

This simplification leads to somewhat different evolution of the simulated response of shear forces in relation to slip, after the peak force point is reached (Ferrer et al., 2003). 5.3 Sheet thickness Shear strength has shown a linear dependency on sheet thickness within its usual values, from 0.75 to 1.25 mm. This agrees with European Standard, and other rules, for strength extrapolation. In fact, shear sliding strength should depend on squared thickness value, as corresponds to the plastic bending of a shell, but linear approximation is acceptable within the usual above said range of thickness, as can be observed in the Figure 8c. 5.4 Embossing depth A linear dependency on embossing depth has been found as well. The only design limitations to high depth values are manufacturing technological aspects, not presented in this paper. 5.5 Embossment length The sliding movement between steel sheet and concrete produces interaction forces located just on embossment ends, because steel sheet bends while concrete remains straight. So, the ends of large embossments, as well as spot embossments, are better located near the profile edges, since much higher forces are required to bend the flat sheet between edges. Thereby, variations on embossment length modify shear resistance since the embossment ends are being moved. Increasing the length of a centred embossment will increase involved forces. 5.6 Sheet profiling angle Maintaining the profile high and embossment length fixed, the more vertical the lateral walls are (90◦ ), the shorter they are, thereby, the higher the required forces to bend them are; in addition, vertical components of forces that tends to detach steel and concrete vertically decrease. 5.7 Embossment inclination Maintaining the embossment length fixed, the more vertical the embossment is, the nearer its ends are to profiling edges, and so, the higher the required forces to bend the sheet are. 6 CONCLUSIONS – On one hand, friction coefficient is a very scattered physical magnitude and usually out of control during construction process due to the large number of uncontrolled factors that may affect it, such as temperature of concrete and steel, steel surface state, ambient temperature, ambient humidity, curing process, shape and nature of aggregates, etc. On the other hand, results show that shear resistance depends linearly on friction coefficient, and low resistance is observed in case of null coefficient. Shear resistance can be tripled from µ = 0.2 to µ = 0.6. So, it seems necessary to normalize the test conditions of standard rules to the worst case; that is, near to null coefficient, for example, by coating with oil the sheet surface without embossments before setting the concrete. Something similar occurs with chemical bond, also being overcome by painting the sheet. – Embossing slope has shown to be the more significant parameter affecting the shear resistance. Few degrees variations of slope cause large variations on shear resistance. However, the improvement of shear resistance is always achieved at expense of ductility, and the brittle peel-off failure of concrete is limiting this parameter. – The verification of the embossing slope in standard certification tests and the definition of strict tolerances for manufacturing, are essential points to guarantee a reliable shear resistance. 453

– – –

–

–

An excessive manufacturing tolerance for this geometrical parameter is, probably, an important cause of test results dispersion. As standard rules postulate, shear resistance can be assumed to depend linearly on the sheet thickness within the usual thickness range, in spite of its real square dependency. Linear dependency has been found for embossment depth as well. From 1.5 mm depth to 3 mm depth, shear resistance becomes three times bigger. From all sheet profiling shapes (opened and dovetailed), it has been observed that the sliding movement between steel sheet and concrete produces interaction forces located just on embossment ends, because steel sheet bends while concrete remains straight. So, large embossments ends, as well as spot embossments, are better located near the profile edges, since much higher forces are required to bend the flat sheet between edges. Thus, large embossments are better than short ones, because ends are forced to be near edges and, in addition, a bigger inertia is achieved to resist transversal bending. No advantage is observed for tilted lateral embossments, in terms of shear resistance. Quite the opposite, vertical embossments works better than tilted ones (end locations maintained) since no asymmetries between ribs are introduced. The only advantage of tilted embossment is the smoother evolution of embossing forces during manufacture. Alternated embossing direction from one embossment to the next (inward/outward) has clearly reduced the dependency of shear resistance on tilting direction and, in addition, transversal bending stiffness is improved, since inertia is increased. So, the shear resistance of side-by-side ribs becomes more similar and the sliding behaviour of the whole composite slab becomes more uniform and higher. Based on this methodology, the new developed T80 profile has been optimised and will be shortly presented.

REFERENCES Baltay, P. & Gjelsvik, A. 1990. Coefficient of friction for steel on concrete at high normal stress. Journal of Materials in Civil Engineering 2(1): 46–49. Crisinel, M. & Marimon F. 2004. A new simplified method for the design of composite slabs. Journal of Constructional Steel Research 60: 481–491. Daniels, B.J. 1988. Shear bond pull-out tests for cold-formed-steel composite slabs. Rapport d’essais, ICOM194. Lausanne: EPFL. Daniels, B.J. 1990. Comportement et capacité portante des dalles mixtes: Modélisation mathématique et étude expérimental. Doctoral Thesis. These N◦ 895. Lausanne: EPFL. Ferrer, M. (in prep). Numerical and experimental approach to the interaction between steel sheet and concrete to improve shear resistance of composite slabs. Doctoral Thesis. Barcelona: Technical University of Catalonia (UPC). Ferrer, M., Marimon, F. & Roure, F. 2002. Losas mixtas (I): modelado mediante elementos finitos del fallo por deslizamiento longitudinal, XV Congreso Nacional de Ingeniería Mecánica: Libro de actas; Proc. XV national congress, Cádiz, 10–13 December 2002. Ferrer, M., Marimon, F. & Roure, F. 2003. Composite slabs (II): finite element modelling of longitudinal shear failure. Trends in the Development of Machinery and Associated Technology TMT2003; Proc. 7th intern. conf., Lloret de Mar, 15–17 September 2003. Ferrer, M., Marimon, F. & Roure, F. 2004. Composite slabs (III): parametrical analysis of the longitudinal slide failure mechanics using finite element models. Advances in Structural Engineering and Mechanics ASEM’04; Proc. 3rd int. conf., Seoul, 2–4 September 2004. Ferrer, M., Marimon, F., Roure, F. & Crisinel M. (in prep). Optimised design of a new profiled steel sheet for composite slabs using 3d non-linear finite elements. Eurosteel Conference on Steel and Composite Structures; Proc. 4th Europ. Conf., Maastricht, 8–10 June 2005. Schuurman, R.G. 2001. The physical behaviour of shear connections in composite slabs, Doctoral Thesis. Technische Universiteit Delft. Delft: DUP Science. Veljkovic, M. 1996. Behaviour and resistance of composite slabs. Doctoral Thesis. Luleå University of Technology. Luleå: Tekniska Högskolan I Luleå.

454

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Crack propagation at headed shear studs in composite beams M. Feldmann & H. Gesella University of Kaiserslautern, Kaiserslautern, Germany

ABSTRACT: Displacement controlled push-out-tests were carried out to investigate the crack propagation and the force-displacement behaviour of cast into concrete headed shear studs. The paper aims at the description of the fatigue behaviour of a composite joint by a crack propagation law by using a stud-force-slip relationship depending on the crack depth. Analytical models are given to describe the crack growth and its influence on the force-displacement behaviour. 1 INTRODUCTION In the past efforts have been made to investigate the high-cycle fatigue behaviour of into concrete cast headed shear studs in composite beams. Most of the investigations have not been performed at full-scale beam-tests but at small-scale push-out-tests, which were invariably subjected to fatigue loading. Using the results of the numerous investigators several design methods have been arrived. The common method to describe the fatigue behaviour of concrete-casted studs used in the EC 4, part 2 (ENV 1994-2-1 1997) bases on the nominal stress concept. A Wöhler-curve was derived by reanalysing international available push-out-tests, where the fatigue resistance is described independent of the diameter. Miner’s law is proposed as linear cumulative damage assessment. An alternative method has been developed by Oehlers (Oehlers 1995). Push-out-tests were reanalysed and complemented by his own tests. The results lead to a design method that is characterised by including the interaction between Ultimate Limit State and Endurance Limit State. The point of Oehlers’ work is that he observed a crack propagation in the push-out-tests, which starts within the first loading event (Oehlers 1985). This concept includes Wöhler-curves depending on the computed shank failure load for the stud connector subjected to static loads and the tensile range of cyclic shear load derived from a crack propagation law. Research investigations at the University of Kaiserslautern performed by Leffer (Leffer 2002) show that the EC4-prediction of fatigue life at two-span composite beams subjected to cyclic loads provides results that do not satisfy. This is the starting point of the work presented here. 2 PUSH-OUT-TESTS IN COMPARISON WITH COMPOSITE BEAMS Push-out-tests are commonly used to find out the static and cyclic behaviour of cast into concrete headed studs. Thereby the standardized (EC4) push-out-test is to be working as a small section of a composite beam supposing that the same shear forces as in the real beam are met. An interesting point is that static and cyclic push-out-tests vary significantly in one point: Static push-out tests are performed displacement controlled and cyclic push-out-tests are performed force controlled. To clarify the kind of control of the action effect in a composite beam the following two situations are discussed: 1. The equilibrium is enough to describe internal forces in a statically determined system. That means that action effect in the structure runs force controlled independent of any stiffness and condition to compliance criteria. 455

δ

P(t,x) (EI)c Ccomp. joint (EI)a Both tests show the same force-deformation behaviour: δ

F

F

δ

N

N

Figure 1. Equivalent force-deformation-behaviour of full-scale beam beam-tests and displacement controlled push-out-tests.

2. There are composite beams, in the simplest case external statically determined (single-span beams). But the composite joint is predominated by extreme statically indeterminate conditions. The composite joint as interface between concrete slab and steel beam provides a deformation condition to the structure. The inertial forces comply with ratio of the stiffness of the concrete slab, the steel beam and the composite joint. Therefore, the distribution of the internal forces depends on the deformation condition through the composite joint; that means that there is displacement control. The conclusion of this consideration is that the cyclic push-out-test has to be loaded displacement controlled in case a composite beam subjected to fatigue should be simulated, see Figure 1. Leffer accomplished displacement controlled push-out-test with a original collective from twospan composite beam tests (Leffer 2002) and gained the same fatigue life. Therewith it is shown that the push-out-test is adequate to simulate a composite beam test if the load-slip-condition is effected by displacement control. 3 CYCLIC PUSH-OUT-TESTS, DESCRIPTION Oehlers (Oehlers 1985) and Leffer (Leffer 2002) found out that the fatigue behaviour of the studs is governed by crack propagation rather than by classical fatigue behaviour with a predominant crack initiation-phase. That means that the whole fatigue life is dominated by crack propagation. Also other researchers gave first evidences to this circumstance. Mainstone and Menzies (Mainstone et al. 1967) are reporting of a reduced static strength in push-out-tests after cyclic loading, which is an evidence for having a crack propagation. For this reason it was necessary to perform further push-out-tests at TU Kaiserslautern to achieve data to analytically describe crack propagation and a force-displacement-relationship. 3.1 Test rig and push-out-test specimen Basically the specimen complies with the proposal made in EC 4 (ENV 1994-1 1992), see Figure 2a. As proposed in EC 4 (ENV 1994-1 1992) are arranged horizontal tension rods to avoid tensile forces in the studs. As a need the test rig was designed to control the magnitude slip very exactly. The displacement occuring in the hydraulic actuator is not suitable as control variable of the electronic control circuit, because this magnitude differs far to much from the displacement in the composite joint. The hydraulic actuator was regulated by means of two inductive displacement transducers placed at the level of the lower four studs. 456

200

Ø14

2 Ø14 o+u

4 Ø10

K-weld circumferental

Q335a

2 Ø8 o+u

620

HEB 320 S235

Q335a

250

Ø10

170

Ø10

125

150

Ø14

(a)

double I-weld

650

(b)

Figure 2. (a) Shuttering and reinforcement drawing and (b) push-out test rig.

For data acquisition efficient software was used to pack and analyse the collected data online. Thus the recorded data in form of slip at the position of the studs, the control slip, displacement and force of the actuator was kept in a reasonable quantity. The controlling slip, the displacement and the force of the actuator were logged in every hundredth cycle with 160 values per 2π and additionally the slip at the position of the studs in every fifth cycle with 240 values per 2π. A clamping fixation has to be arranged to hold back the specimen even it is subjected to pulsating slip range, as reverse forces occur being necessary to reach the initial displacement state. Another requirement is to achieve sufficient high testing frequency in view of testing time and cost. All push-out-tests were carried out with a frequency of the slip range of f = 5 Hz. Figure 2b shows the test rig. 3.2 Carried out tests Twenty push-out-tests so far have been performed. Subject of the investigation were headed studs from the same batch with a diameter of Ø22 mm and a tensile strength of fu = 523 N/mm2 . The concrete had a mean cube strength of fc = 55 N/mm2 (between 39 N/mm2 and 59 N/mm2 ) and the steel was a usual S235JR. The test series contains single stage tests with different slip ranges at s = 0,15 mm, 0,20 mm, 0,25 mm with slip ratio R = 0. 3.3 Test results and findings 3.3.1 Crack width a The crack width is the crucial magnitude. Due to the bad observability of the crack at the studcollar, it is difficult to acquire the crack propagation da/dN at cast into concrete headed studs during the test. For displacement controlled push-out-tests the production of rest lines is also difficult; notably for R = 0 (in terms of slip). On this account seems advantageous to carry out several tests on different stages of action effect with different numbers of cycles. By this it is possible to gain information about the crack propagation. Figure 3 shows the digital figure of a failed stud with two different cracks. The lower crack in Figure 3 was caused by the cyclic loading due the test. The upper crack is effected by artificially inducted brittle fracture of the residual area after the test caused by cooling down the steel with liquid nitrogen and chipping off the stud with a hammer. For analysis only the lower studs are taken into account, see Figure 4, because the crack width differs between the two levels. A reason for that might be the difference of the stud shear stress caused by different deformation lengths from stud to bearing. Further this gives rise to install the inductive displacement transducers for displacement controlling purposes at the lower level of studs. 457

former uncracked area

cracked area

Figure 3. Crack propagation of shear studs.

Crack depth a [mm]

25 20

R=0 ∆s = 0,25 mm Ø22

crack propagation starts from the other side

15 10 crack propagation stops a ≈ 13 mm

5 0 0,E+00

2,E+05

4,E+05 6,E+05 Cycles N [-]

8,E+05

1,E+06

only lower studs taken into account as here are higher forces

Figure 4. Crack propagation of studs subjected to s = 0,25 mm.

The evaluation for a slip range s = 0,25 mm is shown exemplary in Figure 4. The 50% fractiles with scatter-bars at the stages of the cycles N of the crack width are mapped in the chart. Attention should be paid to the issue that the crack front offers different geometric outlines. To achieve more accurate results, the crack area is converted to a crack width by assuming an ideal half-moon shaped crack with a straight line as crack front. Two failure modes occur: Type A is characterized by a crack between collar and shank of the stud; type B is characterized by a crack between collar an flange. The evaluation considers this point by relating type B to type A due linear correction of the crack geometry. Best results are derived by using a linear regression line to relate the crack width to the number of cycles N. By considering Figure 4 some issues get obvious: 1. The crack growth starts with the first cycle, no crack initiation occurs. 2. The crack propagation factor remains constant on a stage of slip range. 3. The crack propagation stops on a several crack width (s = 0,25 mm; a ≈ 13 mm). After that a second crack starts at the opposite side of the stud. The second crack due to cyclic loading grows till the hole cross section of the stud is cut off. There occurs no brittle fracture. The crack width a = 13 mm reduces the second moment of area to 8% and the cross section to 38%. Thereby the moment bearing capacity is out of action, only shear forces can be transmitted. Oehlers (Oehlers 1990) shows that the design resistance of shear connectors reduces linearly by the ratio of remaining cross section to the uncracked cross section of the stud. Thus the bearing capacity has been reduced to 38%. 458

Crack propagation rate da/dN [mm/cycle]

3,5E-05 3,0E-05 2,5E-05 2,0E-05 1,5E-05 1,0E-05 5,0E-06 0,0E+00 0

0,05

0,1

0,15 0,2 Slip range ∆s [mm]

0,25

0,3

Figure 5. Crack propagation rate da/dN against slip range s.

40

Stud force P [kN]

20 0

lower peak value of P

-20 -40 upper peak value of P

-60 -80

R=0 ∆s = 0,25 mm Ø22

-100 0,E+00 1,E+05 2,E+05 3,E+05 4,E+05 5,E+05 6,E+05 7,E+05 8,E+05 9,E+05 Cycles N [-]

Figure 6. Development of the peak loads of displacement controlled push-out-tests, slip range s = 0,25 mm = constant.

The comparison of the crack propagation rate da/dN on the three different levels of slip range s = 0,15 mm, 0,20 mm, 0,25 mm gives the crack propagation rate da/dN versus the slip range s, see Figure 5. This relationship shows the threshold value at sth = 0,10 mm. The single test S12_12 confirms the linear regression line, however this value is not part of the regression analysis of the line given in Figure 5. Thus the slip range s is the only parameter causing damage in form of crack propagation. 3.3.2 Stud force P The actuator force Pact was recorded during the tests as aforementioned. For the evaluation it is assumed that each stud maintains the same shear force. Figure 6 presents the peak stud forces P against the number of cycles for the slip range level s = 0,25 mm. The forces lie altogether very closely independent from the concrete crushing strength. After the loss of high initial upper peak stud forces because of the local concrete crushing at the collar, the force falls linearly with the number of cycles. The comparison between the 459

upper peak load and crack propagation shows a connection. The following equation describes the dependence.

CP assesses to −1,352 [kN/mm]. Thus the cyclic upper peak stud force is given with Equation 2; the rate depends exclusively on the crack width a.

The initial force Pini = 43,4 kN was found by a regression analysis. The cyclic lower peak stud force remains constant and results in Plower = 14,4 kN over all tests. It has to be pointed out that the cyclic peak loads are independent of the slip range s. 3.3.3 Simplified stud-force-slip-relationship of a composite joint with KD Ø22 By means of the recorded magnitudes control slip and actuator force it is possible to apply the loaddisplacement-curves for a composite joint with into concrete cast studs on discrete cycle numbers. With this data is a simplified stud-force-slip-relationship derived with a tri linear accretion. This tri linear accretion distinguishes elastic and plastic behaviour. The elastic behaviour is described by the two ambient-lines and the plastic forceless behaviour by a horizontal line in the middle between. The plastic part contains variability of the behaviour in time, which depends on the actual crack width a and the slip range s.

mupper = 550 kN/mm, mlower = 600 kN/mm. This relationship can be implemented easily in FE-Programs in order to compute composite beams with a flexible composite joint depending on crack width a. 4 COMPARISON BETWEEN FORCE CONTROLLED AND DISPLACEMENT CONTROLLED PUSH-OUT-TESTS Oehlers’ work gives adequate data for the comparison of the here presented results of displacement controlled push-out-tests to force controlled push-out-tests (Oehlers 1985). Oehlers describes indirectly a crack propagation due to the development of the static load capacity. He assesses a linear decrease of static load capacity of the studs depending on a linear crack propagation. This comparison is possible because the crack propagation offers two characteristics in force control and displacement control (see Figure 7): 1. The range of the cyclic stress intensity factor remains constant over much of the fatigue life, what causes a linear crack propagation. This is the precondition for a reasonable comparison. 2. Oehlers introduces the asymptotic endurance Nf . It would be the fatigue life if the fatigue crack was able to pass through the whole of the stud without the stud breaking as a result of fast fracture because of the peak load. The crack propagation law of this work with a crack depth of the diameter of a stud equals Oehlers’ asymptotic endurance Nf . The comparison is plotted in Figure 8, it shows the connection between displacement controlled and force controlled push-out-tests. It has to be annotated that the relation differs sensitively by varying the ultimate strength of the connectors. Also a linearization of the relationship is plotted in Figure 8. This linearization gives the equivalent threshold stud force range to Pth = 22 kN. 460

Kind of control of the action effect Force control

δ

Load-displacement behaviour

Crack propagation

in displacement and force control equal

F

δ

a

22 mm Nf

Displacementcontrol

F

da/dN = const.

F

N

δ

Equivalent stud force range ∆P [kN]

Figure 7. Principal behaviour of push-out-tests in force control and displacement control. 50 45 40 Linearization

35 30 Pth = 22 kN

25 20 15 10 5 0 0

0,05

0,1

0,15 0,2 Slip range ∆s [mm]

0,25

0,3

0,35

Figure 8. Relationship of force controlled push-out-tests of Oehlers and displacement controlled push-out-tests performed at TU Kaiserslautern.

5 SUMMARY The presented work suggests analytical models to describe the force-displacement-behaviour and the damage behaviour of into concrete cast headed studs in composite joints of beams. Shear studs with a diameter Ø22 were investigated. The concrete compressive strength varied from 39 N/mm2 to 59 N/mm2 . REFERENCES Feldmann, M., Gesella, H. & Leffer A. 2004. The cycle force-slip behaviour of headed studs under non static service loads – experimental studies and analytical descriptions. Composite Construction in Steel and Concrete V. Krüger Park, South Africa.

461

Leffer, A. 2002. Zum Ermüdungsverhalten einbetonierter Kopfbolzendübel unter realitätsnaher Beanspruchung im Verbundbrückenbau. Dissertation TU Kaiserslautern. Stuttgart: Ibidem. Mainstone, R.J. & Menzies, J.B. 1967. Shear connectors in steel-concrete composite beams for bridges. Concrete 1: 291–302. N.N. 1992. Eurocode 4: Design of composite steel and concrete structures Part 1: General rules and rules for buildings. Brussels: CEN. N.N. 1997. Eurocode 4: Design of composite steel and concrete structures Part 2: Bridges. Brussels: CEN. Oehlers, D.J. & Foley, L. 1985. The fatigue strength of stud shear connections in composite beams. Proc. Instn. Civ. Engrs. 79: 349–364. Oehlers, D.J. 1990. Deterioration in strength of stud connectors in composite bridge beams. Journal of Structural Engineering 116: 3417–3431. Oehlers, D.J. 1995. Design and assessment of shear connectors in composite bridge beams. Journal of Structural Engineering 121: 214–224. Schüler, W. 2004. Untersuchung des Kraft-Verformungs-Verhaltens einbetonierter Kopfbolzendübel in Verbundträgern and dessen schädigungswirksame Beurteilung auf der Grundlage einer analytischen Beschreibung des Rissfortschritts. Degree dissertation. Kaiserslautern: not published.

462

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Load-carrying capacity of anchor plates with welded studs Ulrike Kuhlmann & Markus Rybinski Institute of Structural Design, University of Stuttgart, Germany

ABSTRACT: For the load transmission into concrete constructions anchor plates with welded studs are frequently used. At the Institute of Structural Design two test series of reinforced concrete specimens were accomplished which show that reinforcement decisively influences the loadcarrying capacity and that the actually calculated design loads tend to underestimate the real resistance. On the basis of 27 test results and supplementary FE analyses a first mechanical model based on the component method is presented, describing the load-carrying capacity of anchor plates with welded studs in reinforced concrete elements. Due to transparency of load distribution and applicability to other situations this first mechanical model fits to the developments of European research work in steel and composite and the needs of practical engineers.

1 GENERAL 1.1 Design of anchor plates At present, anchor plates with welded studs are designed by technical approvals (DIBt 2000a, DIBt 2000b). The ultimate concrete strength is calculated by CC-Method (Eligehausen et al. 1997, Eligehausen & Mallée 2000), which describes very well the load-carrying-capacity of anchor plates in plain concrete. The ability of considering reinforcement is restricted. In addition if both shear and tension forces occur on studs, a conservative interaction relation has to be applied (Leffler 1998, Hauf 2003). 1.2 Research project The aim of the research project (Kuhlmann & Imminger 2004) was to identify the structural behaviour of anchor plates with welded studs considering supplementary reinforcement and to develop a first design model for practical usage. To ease the assimilation of the research results in actual European developments in steel and composite the design model is based on the component method.

Figure 1. Anchor plate with supplementary reinforcement submitted to shear or tension.

463

Distance of headed studs

Surface reinforceme nt

Load eccentricity

Load direction

S1

as - as⊥

e



0° 45° 90°

10 cm 15 cm 20 cm

0 1 2 3

C 20/25 C 25/30 C 35/45

hn

n Specimen

Concrete quality

4 cm 10 cm 15 cm

Length of headed studs

12/10-12/15 14/10-12/15 14/10-12/10

Number of additional stirrups

10 cm 15 cm 20 cm

Table 1. Varied parameters of anchor plates with headed studs.

B1 B2 R 1-1 R 1-2 R 1-4 R 2-1 R 2-3 R 3-2 R 3-3 R 4-2 R 4-3 R 5-1 R 5-3 R 6-2 R 6-3 R 6-4

Calibrated test load [kN]

R 7-2 R 7-3 R 7-4 R 7-5 R 7-6

Number of stirrups 800 624

600

700

695

R1-1

480

400 200

B1

0 R 1-1 0 Stirrups

R 1-2 1 Stirrup

B1 2 Stirrups

R 1-4 3 Stirrups

Figure 2. Test loads in dependence of supplementary reinforcement.

2 TESTS In the context of the research project 31 tests have been carried out. 21 of the tests were accomplished on anchor plates each with 6 identical headed studs. Supplementary reinforcement was installed in order to find out the influence of reinforcement on the serviceability limit state and the loadcarrying capacity of the anchor plates. Furthermore, 10 tests were accomplished with different combinations of connection devices for load transmission into concrete elements. Beside headed studs, connection devices like welded reinforcement and block shear connectors were used. Table 1 shows a list of the varied parameters for the 21 tests with headed studs. 464

Calibrated test load [kN]

Length of studs 1000

874

800 600

R 2-1

695 457

400 R 2-3

200 0 R2-1 100 mm

R 2-3 200 mm

B1 150 mm

Figure 3. Test loads in dependence of length of studs.

There were 7 test rows. Starting from a basic specimen, only one parameter was varied within each test row. So the influence of each parameter could be identified with a limited number of tests. More specific data of the tests is listed in Kuhlmann & Imminger 2004. Figures 2 & 3 show the comparison of test results for the influences of supplementary reinforcement and the length of headed studs. The load capacity of the anchor plates are calibrated to the same concrete compression strength of 35 N/mm2 . The load capacity increases by 30 percent when using additional stirrups within the concrete breakout cone. The use of a third stirrup has nearly no influence for the load capacity. A very effective way to increase the load capacity is to extend the length of the headed studs, see Fig. 3. Further test results and the evaluation of the tests are explained in Hauf 2003, Rölle 2003 and Kuhlmann & Imminger 2004.

3 FINITE ELEMENT ANALYSIS 3.1 Introduction Numerical studies were performed with the program MASA (Ož bolt 2003) developed at the Institute of Construction Materials, University of Stuttgart for non-linear calculations on concrete structures. The studies were performed deformation controlled similar to the tests. The concrete, structural steel and studs elements and also the intermediate layer between steel and concrete were modeled with an 8 node volume element. For modeling the reinforcement 2 node beam elements were used. More detailed information is given in Kuhlmann & Imminger 2004. 3.2 Verification of the FE Model Before using the non-linear FE model for the variation of parameters, it was verified by 19 tests with additional stirrups, see Table 2. Figure 4 shows the FE-loads compared to test results. Due to the satisfying correlation of test and calculated load capacity, the FE model is used for further analysis. 3.3 Structural behaviour Figure 5 shows a qualitative sketch of the normal stress σz on the anchor plate of the specimen B2 at maximum load. Due to the load eccentricity, there is a tension force which is covered by shear studs on the non-loaded side for this configuration. The integrated normal stress of the other studs is in every section nearly 0, so that these studs have no tension forces which have to be considered for the design of the inner moment resistance. Figure 6 shows qualitatively the distribution of the normal stress σy on the anchor plate of specimen B2 at maximum load. The shear force is transmitted into the concrete specimen by shear studs (V1 and V2 ) as well as by friction (Vf ) between anchor plate and concrete. Most of the shear force transmitted by shear studs is covered by the studs (V2 ) on the non loaded side. 465

1200 1000

Table 2. Verification of the FE model. Fu,Test

800 600

Test load

Fu,Test

FE load

Fu,FE

400

Ratio

Fu,FE /Fu,Test

200

Average value

0,998

Maximum ratio

1,20

Minimum ratio

0,86

Variation coefficient

8,7 percent

0 0

200

400

600

800 1000 1200

Fu,FE

Figure 5. Normal stress σz at maximum load.

Figure 4.

Comparison of test and FE loads.

Figure 6. Normal stress σy at maximum load.

In addition to the tests, a parameter study has been performed by the verified model by variation of different parameters. The influence of the parameters concrete quality, load eccentricity, edge distance as well as the diameter of the headed studs is medium to large concerning the load capacity of the anchor plate. However, nearly no influence was noticed for the diameter of the stirrups (dst = 10 to 14 mm), the distance of the first stirrup (sst = 25 to 75 mm) and an additional stirrup on the compression side of the anchor plate noticed.

4 MECHANICAL MODEL 4.1 General On the basis of the component method a design model has been developed and validated by the tests of the research project. It can be applied for shear, tension or combined tension and shear forces. Therefore, the effective components of the joint were identified, their structural behaviour was characterised and finally the load capacity of the whole joint was calculated. The design method is independent of the kind of fastener, e.g. headed studs or welded reinforcement, to transmit tension forces or, e.g. block shear connector or headed studs, to transmit shear forces. Within the scope of the research project only the maximum strength of each component was considered. The stiffness and the ductility were not taken into account. This simplification is possible if the inner load distribution is clearly defined, e.g. there is only one component for load transfer of the tension force. 466

ZR,1

ZR,1

Component 1: Steel failure NR,s

NR,s

Component 2: Pull-out failure NR,p

NR,p

NR,c

Component 3: Concrete cone failure without supplementary reinforcement:

NR,cp

with supplementary reinforcement:

NR,cr

Figure 7. Spring model of component group “tension”. VR,f

Component 1: Friction force VR,f

VR,1

Component 2: Shear connector 1 VR,1

VR,2

Component 3: Shear connector 2 VR,2

VS VR,f VR,1

VS

VR,2

Figure 8. Spring model of component group “shear”.

If in general shear forces are transmitted by several rows of headed studs, the stiffness of those fastenings has to be considered to calculate the inner distribution of the shear force. If in addition the connection device is designed according to its plastic resistance the ductility of the each component has to be considered as well. However, for the case considered here stiffness and ductility of the components may be neglected. The group of tension components is considered as a spring model with 3 springs in row, see Figure 7. The design of the single components is explained in the following text. The load capacity of the component group “tension” is defined by the weakest component. The group of shear components is considered as a spring model with 3 springs in parallel, see Figure 8. Thus, a failure of one component does not cause a complete failure of the whole joint. 4.2 Calculation scheme of maximum shear resistance VR The maximum shear resistance can be calculated by the scheme in Table 3. The maximum strength of the component group tension is assumed as minimum of the maximum calculated strengths of the different tension components. The load-carrying capacity of the component “steel failure” NR,s and of the component “pull-out failure” Nr,p can be taken of the technical approvals. The load-carrying capacity of the component “Concrete cone failure with reinforcement” NR,cr can be calculated by the mechanical model in Figure 9. It has to be distinguished whether the tension force ZR,1 is set into equilibrium only by D1 of the reinforcement or also by D2 of the compression zone of the anchor plate. The ultimate strength Nu,a of the reinforcement consists of the strength transferred by hook and the load strength transferred by composite action (Eligehausen & Mallée 2000).

467

Table 3. Calculation scheme. Steps of calculation 1 2 3 4 4.1 4.2 4.3 5

Design of component group “tension force” Calculation of compression zone height Calculation of inner moment resisitance Calculation of maximum shear resisitance Identification of friction force Calculation of VR Check of stud resistance Check of shear resistance

ZR,1

ZR,1 x MR VR,f VR VR,1 and VR,2 VR

DR ZR,1 DR

D1

D2 D1,k1 Zr,kr

Zr

D2,k2 w Zr

Figure 9. Load distribution of the tension stud force.

where Nu,1 = strength by hook effect; Nu,2 = strength by composite action; As = area of reinforcement and fyk = strength of reinforcement. If there is a compression zone under the anchor plate, not all the tension force of the headed studs is equilibrated by the compression force of the reinforcement. A part of the tension force is balanced by the compression force D2 caused by the compression zone under the anchor plate. For the load distribution of the tension force it is necessary to consider the stiffness k1 and k2 of both compression struts. According to the test results, the reinforcement takes between 17 and 25 percent of the tension load of the headed stud. But this assumption has to be checked by more tests. According to equilibrium orthogonal to the concrete surface the concrete compression force DR has to be equal to the force of the tension component ZR,1 . So the height of the compression zone and the inner moment resistance can be calculated, see also Figure 10.

where ZR,1 = maximum strength of tension components; fc = strength of concrete and bplate = width of anchor plate in compression zone.

468

VR

VR x

ZR,1 MR,i

VR,f

e2 VR,1

DR

e e1

VR,2

d

Figure 10. Determination of inner moment resistance MR,i .

Figure 11. Mechanical model to define the shear strength VR .

where ZR,1 = maximum strength of tension components; d = lever arm of tension components to the edge of the anchor plate and x = height of compression zone. The maximum shear strength consists of the friction force VR,f , see Equation (4), and the shear resistance of the studs VR,1 and VR,2 .

where VR,f = friction force; µ = 0.5 coefficient of friction used for calculation of tests and DR = compression force between anchor plate and concrete surface. After identification of the friction force,
the maximum shear strength VR can be calculated by the equilibrium of the inner moment M = 0, see Figure 11. The strength of the studs has to be checked by calculating their load capacity and their load share. Interaction between shear forces and tension forces has to be considered (Bode & Hanenkamp 1985), see Equation (5).

where VR,1 + VR,2 = load share of studs; VR = maximum shear strength calculated by the equilibrium of the inner moment and VR,f = friction force. If the ultimate strength of the studs is exceeded the load-carrying capacity of the anchor plate has to be calculated with a reduced tension force Z1 < ZR,1 . 4.3 Comparison of test results and mechanical model The test results show a good correlation compared to the results calculated with the proposed mechanical model. The mechanical model slightly underestimates the load-carrying capacity, see Table 4. In this comparison, only tests with additional reinforcement like stirrups were included. The specimen R 2-1 is not included because the studs were too short so that a load transmission between headed studs and stirrups did not happen. This non-ductile failure of the specimen R 2-1 accords with a specimen without additional reinforcement. 4.4 Application of the mechanical model for other situations There was an additional test row of 6 tests in concrete columns to gain first experiences of the load-carrying capacity of anchor plates with welded studs considering short edge distances. The 469

Table 4. Comparison of calculated forces and test results.

Specimen

Maximum test load Fu,test [KN]

Number of stirrups n [-]

Share of reinforcement ZR /ZR,1 [percent]

Distance between stud rows d [mm]

Load eccentricity e [mm]

Calculated force VR [kN]

Ratio VR / Fu,test [–]

B1 B2 R 1-2 R 1-4 R 2-3 R 3-2 R 3-3 R 4-2 R 4-3 R 5-1 R 5-3 R 6-2 R 6-3 R 6-4

690 700 624 700 874 779 801 732 1021 735 819 508 374 406

2 2 1 2 2 2 2 2 2 2 2 2 2 2

25 25 17 25 25 25 25 25 25 25 25 25 25 25

150 150 150 150 150 150 150 200 250 150 150 150 150 150

40 40 40 40 40 40 40 40 40 40 40 100 150 150

738 738 548 738 738 738 738 872 889 738 738 426 315 315

1,07 1,05 0,88 1,05 0,84 0,95 0,92 1,19 0,87 1,00 0,90 0,84 0,84 0,78

Average Value.

0,94

mechanical model was used to recalculate the 6 tests in columns. To take account of the edge distance, the stress distribution in the compression zone was calculated by parabola-rectangle diagram. Using this assumption the load-carrying capacities calculated by mechanical model showed a good correlation to the test results with an average value VR /Fu,test = 1,025. However, the procedure has to be verified by more test and studies.

5 CONCLUSIONS The transmittance of shear or tension forces by anchor plates with welded studs into concrete structures with reinforcement like walls or columns is very relevant for practical applications. The proposed mechanical model can be a first step for solving this problem. It complies with the needs of an engineer for use in practice: transparency of the distribution of the forces and transferability to other situations. This model is based on the component method and therefore can easily be integrated in actual European research work. More coordinated research programs between steel and fastenings researchers is required for verification of the results and a higher acceptance of this method.

ACKNOWLEDGEMENT We want to thank “Deutscher Ausschuß für Stahlbau e. V. (DASt)” and “Arbeitsgemeinschaft industrieller Forschungsvereinigungen Otto of Guericke e. V. (Aif)” for the financial support of the research project, just like Ed. Zueblin AG, Stuttgart and Karlsruhe, and the companies Peikko GmbH, Waldeck, Koester & Co. GmbH, Ennepetal and Goldbeck Bau GmbH, Bielefeld for the donations in kind and active support. REFERENCES Bode, H. & Hanenkamp, W. 1985. Zur Tragfähigkeit von Kopfbolzen bei Zugbeanspruchung, In Bauingenieur 60, 361–367, Berlin: Springer.

470

DIBt 2000a, Allgemeine bauaufsichtliche Zulassung Nr. Z-21.5-280, Verankerung von Stahlteilen mittels angeschweißter KÖCO-Kopfbolzen in Beton, Deutsches Institut für Bautechnik. DIBt 2000b, Allgemeine bauaufsichtliche Zulassung Nr. Z-21.5-82, Verankerung von Stahlteilen mittels angeschweißter Nelson-Kopfbolzen in Beton, Deutsches Institut für Bautechnik. Eligehausen, R., Mallée, R. & Rehm, G. 1997. Befestigungstechnik. In Betonkalender 1997, Teil 2, 609–753. Berlin: Ernst & Sohn. Eligehausen, R. & Mallée, R. 2000. Befestigungstechnik in Beton- und Mauerwerksbau, Berlin: Ernst & Sohn. Hauf, G. 2003. Neue Nachweisverfahren für Ankerplatten mit angeschweißten Kopfbolzendübeln und Zulagebewehrung, Diploma thesis, No. 2003-10X, Institute of Structural Design, University of Stuttgart. Kuhlmann, U. & Imminger, T. 2004. DASt-/AiF-Research Project. Final Report: Ankerplatten und Anschlussdetails zur Kraftüberleitung im Stahlbau, Deutscher Ausschuss für Stahlbau DASt, Düsseldorf. Leffer, A. 1998. Auslegung, Konstruktion und Berechnung von Ankerplatten zur Befestigung von Stahl bauteilen an Massivbauteile in der Mischbauweise. Kaiserslautern: Universität Kaiserslautern. Ozbolt, J. 2002. MASA3 – Microplane Analysis Program (Version 1/2002), Institute of construction materials, University of Stuttgart. Rölle, L. 2003. Konzentrierte Lasteinleitung mittels Kopfbolzen in Stahlbetonstützen, Diploma thesis, No. 2003-21X, Institute of Structural Design, University of Stuttgart. Ruf, A. 2003. Untersuchungen zum Tragverhalten ingeniuermäßig dimensionierter Einbauteile in der Mischbauweise, Diploma thesis, No. 2003-11X, Institute of Structural Design, University of Stuttgart.

471

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

An experimental study of the strength and stiffness of concrete-filled steel tubular column connections with weld and stiffener angles S. De Nardin & A.L.H.C. El Debs Department of Structural Engineering, School of Engineering at Sao Carlos, University of Sao Paulo, Brazil

ABSTRACT: This paper describes the results of an experimental program on two welded connections. The considered connections include two details of welded connections between I-shaped steel beams and concrete-filled steel tubular column. One of these welded details is stiffened by angles welded in the interior of the profile wall at the beam flange level. The specimens were tested in a cruciform loading arrangement with variable monotonic loading on the beams and constant compressive load on the column. The test results show that the angles welded in the interior of the profile wall increase the strength and stiffness of the welded connection detail. In addition, the curves of behavior of these connections are compared and some interesting conclusions are drawn. The test results are summarized for the strength and stiffness of each connection.

1 INTRODUCTION AND PREVIOUS RESEARCH The combination of structural steel tubes and concrete to form concrete filled steel tube columns (CFT) result in a structural element that has many attributes. The increasing use of concrete filled steel tubular columns in building structures is a justifiable tendency due to its high economic and structural performance. The advantage of using CFT composite columns is that the steel tube provides a permanent formwork for the concrete placement, thereby expediting construction. In addition, the steel tube provides confinement to the concrete inside the tube, whereas the concrete inhibits the local buckling form occurring in the steel tube. The result is a structural member possessing exceptional axial and lateral stiffness, strength and ductility. Figure 1 shows examples of the concrete filled steel tube columns. The major obstacle to the effective use of concrete-filled steel tubular columns is the lack of effective connection details. Therefore, the development of new beam-column connections is of major importance in order to guarantee an increasing use of the concrete filled steel tubular columns. The study of beam-column connections has been carried out in the last two decades. Expectations of technical and economic flexibility and cost-effectiveness that alternative feasible solutions may offer motivate research and development in this field. Owing to the complexity and cost of connection details it is very important to investigate the behavior of the easy connections, as the welded ones.

Figure 1. Examples of the concrete filled steel tube columns.

473

Beutel et al. (2001) placed the connection details in two broad categories: those that connect the beam to the steel tube only were denominated “external” connection and internal connections were the ones that attempted to distribute some or all flange forces directly into the concrete core by embedded elements. The external connection category includes flange and web cleat connection plates, full-strength butt welded to the face of the tube and the use of continuous stiffening rings or diaphragms at the top and bottom flange locations. Beam-column connections in concrete filled steel tubes are usually constructed by directly welding the steel beam to the steel tube (external connection detail). This form of connection is very simple, hence a cheap one and, like all external connections, offers no internal flow restriction to the pump filling the steel tube. However, this practice should be avoided for the following reasons (Azizinamini & Prakash 1993): – Large residual stresses because of the restraint provided by other connection elements; – Separation tube-concrete core due to the transfer of tensile forces to the steel tube; – Increase in the connection rotation by the deformation of the steel tube decreasing the stiffness of the connection. These problems can be reduced by using either the external or internal stiffener in addition to the welded connection. For example, the beam-column connection can be reinforced using two horizontal or vertical internal diaphragms (Sasaki et al. 1995). Another example of the external connection is the utilization of external stiffener rings connecting steel beams and circular columns as in Choi et al. 1995. The utilization of the reinforcing bars welded to the top and bottom flanges and embedded into the concrete core is a simple technique and test results have showed that the connection strength increases and the bars are very efficient in transferring both compressive and tensile loads directly into the concrete core of the composite column (Beutel et al. 2001). A type of through-bolt connection between steel beams and concrete-filled hollow steel sections was tested by Prion & McLellan (1994). Specimens were tested in a cruciform loading arrangement with variable moment-arm and the contributions of friction and bearing were determined by embedding some of the bolts in the concrete core while others passed through clearance sleeves through steel tube and concrete core. Test results showed that the bearing resistance bolts-concrete core was the most important load carrying mechanism and the friction between the steel wall and concrete core was significant, although less reliable. Further more, through-bolt connections can be a very efficient means of carrying high shear forces from floor beams and composite columns and the bolt shear capacity is the limiting factor with the bending in the bolts being insignificant. Similar beam-column connections were presented and discussed by Ricles et al. (1995). One of these connections was a structural detail with an interior set of diaphragms placed at the flange height of the beams and the diaphragms were welded along all four edges. The shear tab was welded and bolted to the beam web using bolts (Figure 2a). Another detail had no interior diaphragms and used structural tees to stiffen the steel tube (Figure 2b). Figure 2c shows the connection that utilized high strength bolts, endplates and shear studs. This connection and other details discussed by Ricles et al. (1995) are shown in Figure 2.

(a) Diaphragms

(b) Structural tees

Figure 2. Bolted beam-column connections investigated by Ricles et al. (1995).

474

(c) bolts, endplates and shear studs

Therefore, several connection types were previously studied by other researches including welded and bolted connections. The present paper describes the results of an experimental program developed at The University of Sao Paulo to determine the strength and stiffness of welded and welded stiffened beam-column connections. Two welded beam-column connections were tested and the results are summarized in this paper. 2 TEST SPECIMENS 2.1 Geometry Two specimens consisted of a square concrete filled steel tube column and steel beams. The “Welded” detail was a simple beam-column connection in which the beams were welded directly on the face of the steel tube. The “Stiffened detail” was identical to the “Welded detail” but four angles L50 × 50 × 130, 6.3 mm thick were welded by fillet weld on the internal surface of the steel profile. The angles were embedded in the concrete core and placed at the same level as the upper and bottom flanges of the steel beams. Each composite column had length of 1950 mm and the steel beams were 1650 mm long. The cross section of columns was 200 mm square. The dimensions of the Welded and Stiffened specimens are shown in Figure 3. The angles were welded to the steel tube with a continuous fillet weld using an MIG welder with gas immersion shielding and the steel beams were connected directly to the steel column with full strength butt weld. 2.2 Steel and concrete properties In order to characterize the steel material, three coupons were taken from each steel component: web and flanges of beams, endplates and steel profile. These coupons were tested according to E 8M – 00: Standard test methods for tension testing of metallic materials. The composite columns investigated in this paper were completely filled with concrete. The compressive strength (fc) and the concrete modulus were measured. The results of the steel and concrete properties are given in Table 1.

Figure 3. Geometry of Welded and Stiffened details. Table 1. Steel and concrete properties of the test specimens. Steel properties (MPa) Column

Flange beam

Web beam

Angle

Concrete properties (MPa)

Specimen

fy

fu

fy

fu

fy

fu

fy

fu

fc

Ec

Welded detail Stiffened detail

269 383

416 464

287 287

439 439

264 264

384 384

– 264

– 384

64.2 63.1

33711 33256

475

Figure 4. Test arrangement.

Figure 5. Displacement transducers arrangement.

Figure 6. Strain gauges arrangement on the angle.

2.3 Test arrangement, measurements and procedure In the test setup the specimens were subjected to static load on the columns and steel beams. All the concrete filled steel columns were subjected to constant axial loading equal to 450 kN (20% of the squash load, which is equal to the compressive strength of the concrete core plus compressive strength of the steel tube). This compressive load was applied using hydraulic jack. In all of the test specimens the nominal b/t ratio of the steel tube was 31.8. Vertical forces were applied on the beams at 1500 mm from the composite column face by computer controlled hydraulic actuators. The “Welded” and “Stiffened” details were tested in the Structures Testing Laboratory of the Structures Department at the Sao Carlos Engineering School, University of Sao Paulo (USP). The test arrangement is shown in Figure 4. A general arrangement of the displacement transducers for all specimens is shown in Figure 5. The displacements were measured at several points along the steel beams. In order to measure angle strains, strain gauges were placed in vertical and horizontal positions. Figure 6 shows the strain gauges arrangement on the four angles into the concrete filled steel tube. The strains were measured on the external face of the steel tube by strain gauges and their position is shown in Figure 7. The steel tubes used in the tests were filled in a vertical position and the concrete was dropped into the tube from the top. Progressive vibration was employed in order to eliminate air pockets in the concrete and also to give a homogeneous mix. No reinforcement bars were used in the concrete filled steel tubes. The speed of load application in the beams was 0.005 mm/second in the Stiffened detail and 0.01 mm/second in the Welded detail. 476

Vertical displacements (mm)

Left beam

0 -10 -20 -30

Welded detail Stiffened detail

-40 0

Figure 7. Strain gauges arrangement on the steel tube.

30 60 90 120 150 180 210 240 270 300 330 Length of Beam (cm)

Figure 8. Vertical displacements of the Welded and Stiffened details. (b)

(a)

-60

-60 left beam right beam average

-50 -40

Average moment (kN.m)

Average moment (kN.m)

Right beam

Column

Average

-30 -20

Right -10 Left 0

0

Average

-50 -40 Right -30 -20

left beam right beam average

-10 0

-30 -5 10 -15 -20 -25 Connection rotation (mrad): "Welded detail"

0

-2

-4

-6

-8

Left

10 -12 -14 -16 -18 -20

Connection rotation (mrad): "Stiffened detail"

Figure 9. Moment versus Rotation connection curves – (a) Welded detail and (b) Stiffened detail.

3 RESULTS AND ANALYSES Values of the vertical displacements of the beams are shown in Figure 5 for the “Welded” and “Stiffened” details. In Figure 8, the vertical axes show the vertical displacements and the horizontal axes show the length of the beams. Ultimate values of the vertical displacements of the end beams were utilized to evaluate the rotation of the specimens tested. The moment-rotation behavior obtained for tests is presented in Figure 9. The rotations were calculated from the vertical displacement of the beam ends, at 1500 mm from the concrete filled steel column face. In the tests, the joint moment is defined as the average moment at the composite column face, resulting as the product of the applied load and the distance at from the centre of load to the outside surface of the column. In both specimens, the left and right beams presented similar moment and rotation values. The behavior of the left and right beams rotation was similar and can be represented by average values. The behavior moment versus rotation connection curves was very different for the Welded and Stiffened details as shown in Figure 10. It is possible to see that at the initial stage of loading, the moment-rotation relationship exhibit a linear behavior. Table 2 presents the values of the ultimate moment and initial rotational stiffness. The ultimate moment capacities of “Stiffened” and “Welded” details were 49.2 kN · m and 43.5 kN · m, respectively. Therefore, the ultimate moment capacity of Stiffened specimen was 13% higher than Welded specimen and the angles increased the ultimate moment capacity and decreased the connection rotation. The resistances of the Welded and Stiffened connections were respectively 32.7% and 37% of 477

Average moment (kN.m)

Average moment (kN.m)

-60 -50 -40 -30 -20 Welded detail Stiffened detail

-10 0

0

-5 10 -15 -20 -25 Average connection rotation (mrad)

(a) Moment x rotation

-30

-60 Stiffened: beam

Stiffened: column

-50

Welded: beam

-40 -30

Welded: column

-20 -10 0 0

-10

-20 -30 -40 -50 -60 Connection rotation (mrad) (b) Displacements column x displacements beams

-70

-80

Figure 10. Moment versus Rotation connection curves – Welded and Stiffened details. Table 2. Values of the Ultimate moment and initial rotational stiffness. Specimen

Ultimate moment (kN · cm)

Initial rotational stiffness (kN · cm/mrad)

Welded detail Stiffened detail

4350 (1200) 4920 (1650)

663.0 2136.0

m= 1,0

M Mpb

RIGID 0,8

Unbracedframes Braced frames

0,6 SEMI-RIGID 0,4

Stiffened

0,2

Welded
=

NOMINALLY PINNED 0,0 0,00 0,05 0,10 0,15 0,20 0,25 0,30 0,35 0,40

E.Ib.
Lb.Mpb

Figure 11. Classification by strength and experimental Moment-rotation curves.

the bending capacity of the steel beam (Mpb = 13292 kN · cm). The bending capacity of the steel beam was calculated using the steel properties showed in Table 1. Both connection details can be classified as partial strength connections. The values between parentheses in Table 2 were the moment values that defined the upper limit of the elastic stage (Figure 10a). The rotation of a connection was also calculated using the displacement values measured by the transducer displacement numbers 11 and 12 showed in Figure 5. The distance between the center of the transducer and the flange beam face was 71 mm. These rotation results are showed in Figure 10b and the values were very different between column and beam in the Welded specimen. The column and beam values of rotation were very similar in the Stiffened specimen and the steel tube-concrete core separation was more intense in Welded specimen. In the Welded specimen it was possible to observe concrete-steel separation. As only part of this separation was measured by transducers (denominated Welded: columns in Figure 10b), those measurements were not utilized to evaluate the connection rotations. Figure 11 shows the stiffness classification and the experimental moment-rotation curve of Welded and Stiffened connections. A beam span of 6.0 m (Lb) was assumed to defining the boundaries. The Welded and Stiffened specimens are located in the semi-rigid domain and therefore the connections can be classified as semi-rigid. 478

-60

-60 Point 2

-40

Average moment (kN.m)

Average moment (kN.m)

Point 4 -50 Point 8

-30 -20 -10 0

Point 6 Vertical strains -500 -400 -300 -200 -100 0 100 200 300 400 500 Angle strains - (µε)

Point 5

Point 3

Horizontal strains

-50 -40 Point 7

Point 1

-30 -20 -10 0 -200

0

200

400 600 800 Angle strains (µε)

1000

1200

Figure 12. Strains in the angles – Stiffened detail.

Figure 12 shows the strains measured on the four angles versus connection moment. A similar behavior was observed for the vertical angle strains (for example, between point 2 and point 4). The strains on compressed angles (point 6 and point 8) were lower than the strains on tensioned angles (point 2 and 4). The behavior of the horizontal strains on the angles is entirely different to all the vertical strains. Horizontal strains on the angles presented similar behavior between point 1 and point 3, but strain values in point 1 were higher than in point 3. Points 5 and 7 presented the lowest strain values on the horizontal faces of the angles. The typical failure mode of the Welded and Stiffened specimens is showed in Figure 13. The vertical displacement of the Welded specimen was bigger than the Stiffened specimen. The displacement value in the end beam of the Stiffened specimen was 25.3 mm. The end beam of the Welded specimens achieved 40.2 mm displacement value at the ultimate moment value. Therefore, the angles decreased the vertical displacement of the end beams (a 37% decrease between the Stiffened and Welded specimens). A separation steel tube-concrete core occurred and was confirmed by the strain values on the external surface of the steel tube (Fig. 13). The behavior of the strains on the external face of the tube is shown in Figure 14. The strain values measured on the external face of the steel tube showed the separation tube-concrete core due to the transfer of tensile forces to the steel tube. The strains increased the connection rotation decreasing its stiffness. The angles decreased the strain values on the external face of the tube compared to the strains in the Welded connection. For two specimens tested, there was no evidence of cracking in the fillet weld connecting the beams to the concrete filled steel tubes, only excessive deformations on the face of the tube and separation between the steel tube and the concrete core of the concrete filled steel tube column. 4 SUMMARY AND CONCLUSIONS The results of the tests presented here investigated the performance of beam-concrete filled steel tubes connections varying the connection detail. Two welded specimens were tested. One of them was stiffened by angles welded in the interior of the profile wall at the beam flange level. The angles were L50 × 50 × 6.3 mm. The specimens were tested in a cruciform loading arrangement with variable monotonic loading on the beams and constant compressive load on the column. It was found that the angles welded in the steel tube of the composite column were very effective in transferring tensile loads directly into the concrete core of the column. The angles increased the ultimate moment and the rotational stiffness of the welded connection. Both Welded and Stiffened connections tested can be classified as partial strength and semirigid. The Welded and Stiffened details presented excessive deformations on the face of the tube and separation between the steel tube and the concrete core of the concrete filled steel tube column. However such deformations were reduced by angles in the Stiffened detail. Concluding, the angles contributed to the beam-concrete load transference and this detail can be utilized to connect the beam and the concrete filled steel tube columns. 479

(a) Typical failure: Welded specimen

Average moment (kN.m)

-60 Stiffened

-50 -40 -30

Welded

-20 -10 0

(c) Stiffened specimen

-200 -400 -600 -800 -1000 -1200 -1400 -1600 Steel tube axial strain (µε)

Figure 13. Typical failure mode for the Welded and Stiffened details.

Figure 14. Behavior of the strains on the steel tube nearly weld.

(b) Welded specimen

0

ACKNOWLEDGMENTS The authors would like to acknowledge Fapesp (Fundação de Amparo à Pesquisa do Estado de São Paulo) for the financial support given to this research. REFERENCES Azizinamini, A. & Prakash, B. 1993. A tentative design guideline for a new steel beam connection detail to composite profile column. Engineering Journal 31(1): 108–115. Beutel, J. & Thambiratnam, D & Perera, N. 2001. Monotonic behavior of composite column to beam connection. Engineering Structures 23: 1152–1161. De Nardin, S. 2003. Concrete filled steel profiles: a study of combined bending and compression and beamcolumn connections. Doctoral Thesis. The School of Engineering at Sao Carlos, University of Sao Paulo, Sao Carlos, 323p. (in Portuguese). Sasaki, S. & Teraoka, M. & Morita, K. & Fujiwara, T. 1995. Structural behavior of concrete-filled square tubular column with partial-penetration weld corner seam to steel h-beam connections. In Oxford, Elsevier (ed.), PSSC’95 Pacific Structural Steel Conference; Proceedings V.2 Structural Connections, Singapore, 1995, p. 33–40. (ISBN: 0 08 042265 9). Choi, S. M. & Shin, IL. B. & Eom, C. H. & Kim, D. K. & Kim, D. J. 1995. An experimental study on the strength and stiffness of concrete filled steel column connections with external stiffener rings. In Oxford, Elsevier (ed.), PSSC’95 Pacific Structural Steel Conference; Proceedings V.2 Structural Connections, Singapore, 1995, p. 1–8. (ISBN: 0 08 042265 9). Prion, H. G. L. & McLellan, A. B. 1994. Through-bolt connections for concrete-filled hollow structural steel sections. Annual Task Group Technical Session; Proceedings, Bethlehem, 1994, p. 239–250. Ricles, J. M. & Lu, L. W. & Sooi, T. K. & Vermaas, G. 1995. Seismic performance of concrete filled tube columns-to-wf beam moment connections. Annual Technical Session; Proceedings, Kansas, 1995, p. 83–102.

480

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Load-deformation and vibration-behaviour of new types of composite slim-floor slabs C. Butz, O. Hechler & H. Trumpf Institute of Steel Construction, RWTH Aachen University, Aachen, Germany

ABSTRACT: In the frame of a Korean-German research project various Slim-Floor systems have been evaluated in respect of bearing resistance- and serviceability demands, cost effectiveness and special Korean requirements. An evaluation of existing slab systems has been performed. On the basis of this results two innovative Slim-Floor slabs have been developed and optimised for Korean high-rise buildings to minimise the construction height of the cross section. These two composite Slim-Floor slabs with optimised cross section have been tested in different ground floor designs to investigate the bearing resistance and serviceability including vibration-behaviour. Particular emphasis is drawn to the dynamic properties and the vibration behaviour due to human activities. All tests are complimented by FE-models including sensibility analyses and parameter studies. Final conclusions are given to transfer of the research results to the design and execution of high-rise buildings.

1 INTRODUCTION On the Asian markets steel has become more and more important as a building material for highrise residential apartment buildings in the previous ten years. Since the Korean authorities started to restrict the building height for high-rise buildings the floor space index dropped significantly. Consequently the contractors aim to reduce the height of each floor and therefore increase the total number of storeys to gain back economical advantages in competition. The leading steel company in Korea POSCO, Pohang Iron & Steel Co. LTD., has awarded a consortium of research institutes and companies to develop sophisticated Slim-Floor decks for high-rise residential apartment buildings up to 70 storeys: – – – – –

POSCO Market Devision Steel Structural Team, Seoul RIST (Research Institute of Industrial Science & Technology), Seoul Studiengesellschaft Stahlanwendung, Düsseldorf 3L Architects, Menden Institute of Steel Construction, RWTH Aachen University (Co-ordinator)

At first the present situation of the Korean market has been analysed and a survey of existing SlimFloor systems has been carried out. The properties of existing Slim-Floor systems differ largely in terms of structural depth, span, resistance to action and fire, sound insulation, vibration behaviour and erection techniques. The research has mainly been focused on shallow slabs with main and secondary elements in one structural plane. For the complex evaluation two representative ground floor designs have been identified – The Low Price Market (LPM) and the High Price Market (HPM). The slab of LPM has got a span of 5,2 m and the slab of the HPM spans 9,6 m. The long spans enable a flexible ground floor design combined with steel frame and lightweight construction. This leads to flexibility and the possibility to adopt the ground floor design to the changing demands of the user in the future. 481

In the pre-design the following Slim-Floor systems have been considered for the two ground floor designs: – – – – –

In-situ concrete slabs and filigree slabs Pre-stressed hollow core slabs WING decks as a combination of filigree and pre-stressed hollow core slabs European profile decks (HOLORIB-, CORUS- and HOESCH-Additive-deck) South Korean profile decks (ALPHA-, ACE- and KEM-deck)

The static analyses and the designs have been carried out according to the Eurocodes. Various solutions of main and secondary beam-structures for each Slim-Floor system have been evaluated. Special requirements of the Korean market have had to be considered in the analysis e.g. a screed of at least 90 mm for the floor-heating and a crack width criteria of w < 0,15 mm. Static calculations and cost estimations for detailed comparisons have been carried out on the bases of real tender documents and bills of quantities. Thereupon optimised lightweight Slim-Floor systems have been designed with regard to the specific demands of the Korean market, high grade of prefabrication, fire resistance and building physics. In the following chapters the optimised Slim-Floor systems are shown and the results of the test on the bearing capacity and vibration behaviour tests are described.

2 OPTIMISED SLIM-FLOOR SYSTEMS FOR THE KOREAN MARKET For the LPM the cost evaluation and static analysis show, that composite decks of HOLORIB or the equivalent Korean ACE-deck system lead to the cheapest and slimmest slab solution. But they are not real Slim-Floor systems as the bearing girder is not integrated into the slab. Hence the deck systems has been optimised by designing a Slim-Floor slab with integrated beams and a unified floor depth over the entire ground floor, see Figures 1 and 2. The slab height is reduced by 20% still offering easy assembling and the advantage of a composite slab. The profile sheets are applied on secondary beams that are composite perforated T-section with concrete dowels for the serviceability state. The T-sections are propped temporarily during construction. The secondary

Figure 1. Grid & span direction for the LPM (shown ground floor design is a part of the total building). Screed and Floor Heating

90 83

ACE or HOLORIB Sheeting

150

51 16

Figure 2. Optimised cross section for the LPM.

482

Screed and Floor Heating HOESCH Deck 90 75 280

Figure 3. Grid & span direction for HPM (shown ground floor design is a part of the total building only).

205

Figure 4. Optimised cross section for HPM.

beam is placed on the bottom flange of the asymmetric main beam which is fully integrated in the slab. The provision of reinforcement through the web of the main beam generates additional bearing capacity and stiffness as well as reduces the crack width. The optimised total depth in 180 mm. The economic analysis based on a specification of tender that has been carried out by 3L Architect, a German contractor and a Korean contractor is resulting in an erection and construction price of appr. 180,-a/m2 . For the High-Price Market the HOESCH-Additive-deck represent an economic solution with sufficient span. The concrete topping of about 75 mm is added on the profile having a height of 205 mm. The low production costs combined with a small slab depth enable the development of an optimised solution for the HPM. Originally the HOESCH-Additive-deck is hanged by means of cams on the upper flange. By placing the cams beneath the upper flange and lowering the concrete zone to the level of the upper flange the main beam is fully integrated in the slab, Figures 3 and 4. Reinforcement bars equally spaced at 150 mm are stuck through holes in the main beam web to fulfil the Korean crack width criteria and restraint against torsion. The plane bottom view of the profile sheet and the lower flange of the main beam result in a total floor depth of 280 mm which means a reduction of ca. 20%. The erection and construction price for this optimised slab according to the specification of tender is about 190,-a/m2 . 3 STATIC TESTS Tests of the ultimate limit state (ULS) according to the Eurocodes have been performed for both Slim-Floor slabs. For testing representative parts of the ground floor designs have been chosen enabling the transfer of the test information to the whole slab system by numerical methods afterwards. Design details of the slabs are given in the Final Report [1]. During tests the service loads have been applied by eight single loads distributed uniformly on the test specimen. For the serviceability limit state (SLS) the crack width design has been controlled and the deflection of the slab (max. f = L/250) has been measured. All tests have been carried out at the laboratory of the Institute of Steel Construction at the University of Kaiserslautern. For the LPM one field of the total test specimen has been built with the SuperHOLORIB SHR 51/150-deck and the other field with the ACE-deck, Figure 5. For the ULS-test 12 upright position encoders have been installed under the slabs and along the main respectively secondary beams. At significant positions 63 strain gauges have been applied on profile sheets, reinforcement bars and particular beams. After several pre-loadings of 3,6 kN/m2 separately on each field the loading has been increased evenly on the whole slab. No ultimate failure has occurred, see Figure 6, thus the load capacity 483

Figure 5. Test set-up for the LPM and chosen slab system. 0 3 kN/m kN/m²² 6 kN/m kN/m²² 9 kN/m kN/m²² 14 kN/m kN/m²² 16 kN/m kN/m²² 18 kN/m kN/m²²

Deflection [mm]

20 40 60 80 100

midspan midspan 120 0.00

1.73

3.47

5.20

6.93

8.67

10.40

Span [m]

Figure 6. Load-span curves in the midspan section for uniformly distributed loads.

has been well above the design load for residential apartments of 2,0 kN/m2 . The serviceability limit for deflection has reached at a loading of 4,4 kN/m2 including 8,0 mm due to self-weight of concrete and screed. The different deflections of HOLORIB- and ACE-deck have been presumably caused by the loading procedure and the different bending stiffness of the sheets in transverse direction. Failure has occurred rather due to SLS criteria, namely deflection and cracking of the screed respectively the concrete. The slab has showed an enormous ductile behaviour due to the large amount of reinforcement in the slab. For the HPM a single field of 9,6 m × 9,6 m has been tested, see Figure 7. As slab the HOESCHAdditive deck supported by cams welded beneath the upper flange and a concrete topping of 75 mm has been applied. The deck has temporarily been propped during casting and hardening. For the ULS-test 17 upright position encoders have uniformly been distributed and 61 strain gauges on profile sheets, reinforcement bars and particular beams have been fixed. After the load has been increased failure occurs in terms of serviceability limit state due to large deflections and cracking of the screed respectively the concrete. Therefore, as expected, a super-elevation of the main beam for the final design is necessary to minimize the deflections under service loads. It is recorded that the main beam and the corrugation have been splitting apart each other at mid-span. This is related to the torsional deformation of the main beam due to the one-sided loading of the slab. For an edge beam in the final design fork end conditions should be provided for the connections to the columns. 484

Deflection [mm]

0

Figure 7. Test set-up for the HPM and chosen slab system.

0 10 20 30 40 50 60 70 80

1.2

midspan

2.4

3.6

Span [m] 6 4.8

7.2

8.4

9.6

0.7 kN/m² 1.1 kN/m² 2.0 kN/m² 3.2 kN/m² 5.0 kN/m²

Figure 8. Load-span curves in midspan section for uniformly distributed loads.

Extensive FE-sensitivity analyses and parametric studies have been carried out for both models to identify the influence of certain parameters (e.g. material strengths, lightweight concrete, span dimensions, end conditions). These studies show that the complete slabs in the final design will profit extremely from the end conditions at support and the continuous action in terms of reduced deflections [1]. For the LPM it has been validated that the slab system with HOLORIB- or ACE-deck satisfy all criteria of the ULS as well as the deflection limit and crack width criteria. The test has proofed the feasibility of the slab system for the application without any modifications. The test set-up for the HPM has not shown any failure according to ULS. Remarkable deflections have occurred at a load level above the design load. The FE-model has been extended to achieve the prediction of the complete slab system behaviour with continuous action according to the specified ground floor design with a super-elevation and fork end conditions of the main beams. The results show that deflections are reduced by these means to 30% of the measured values in the tests. Thus fork end conditions and an super-elevation of the main beams are recommended.

4 VIBRATION BEHAVIOUR As the designed Slim-Floor slabs are lightweight structures with long spans they have low natural frequencies and low structural damping. Hence they react very sensitive to dynamic loads. The dynamic reaction may introduce discomfort to the occupants but it does not imply any lack of structural safety. For serviceability checks vibration tests are carried out to determine the dynamic slab properties and to assess the vibration behaviour regarding comfort of users. Human perception to vibration is very sensitive. The exposure depends on several physical parameters such as amplitude, frequency and direction of the vibrations, the part of the body where the vibration enters, the duration of the effect and the body posture. Additionally the perception thresholds are influenced by age, sex, state of health, attentiveness and the kind of activity being engaged in, e.g. at work, under training, at leisure and the type of building (e.g. home, office, training centres, hospitals). The wide variety of the influence factors makes it difficult to define an uniform and universally applicable relationship between well-being and vibration acceleration combined with daily exposure. The comfort requirements concerning vibration behaviour are demanding for the Korean market, because many residents sleep on futon mattresses that are placed directly on the floor. The most usual and important dynamic excitation is pedestrian traffic on the floors. The vertical human ground reaction forces depend mainly on step frequency, walking velocity and the person’s weight. The walking pace frequency can vary between 1,4 and 2,5 Hz, but walking indoors is commonly performed with frequencies towards the lower end of this range. Running step frequency can rise to higher values but do not commonly exceed 3 Hz. Concerning their dynamic reaction to 485

Factor [-]

1 0.5

Table 1. Perception classification [4]. 0.1

RMS awT of the frequency weighted acceleration Description of the perception aw (t) [m/s2 ]

0.05

0.01

<0,01 0,015

0.005

0.001 0.1

0.5 Wk

1

5 Wm

10

50 100 500 Frequency [Hz]

Figure 9. Frequency weighting functions acc. to VDI 2057 [4].

< 0,02 < 0,08 < 0,315 >0,315

Not perceptible Threshold of perception Barely perceptible Easily perceptible Strongly perceptible Extremely perceptible

footfall forces slabs can be divided into two groups. Low frequency floors have natural frequencies that can be excited in resonance by human activities, whereas the dynamic response of high frequency floors is dominated by the impulsive loading of each individual footstep. Slim-Floors with wide spans and low damping are typical low frequency floors.

5 CURRENT GUIDELINES FOR THE ASSESSMENT OF FLOOR VIBRATIONS Currently sufficient data is not available on damping characteristics of slabs and human induced loads to assess the comfort of occupants in buildings with regard to vibrations in the design state. Hence vibration measurements are necessary to assess the comfort with guidelines on the evaluation of human exposure to whole body vibrations induced by mechanical vibration and shock. The relevant guidelines are shortly presented. The purpose of VDI Guideline 2057 (2002) [4] is to provide an uniform procedure for assessing the effect on human beings of mechanical whole body vibration and also to give general instructions on determining assessment variables. On the basis of measured acceleration data the rms-value aw of the frequency-weighted acceleration aw (t) is formed as characterizing variable. The frequency range extends from 0,5 Hz to 80 Hz for impairment of the sense of well-being, performance and health. The frequency weighted acceleration aw (t) is derived from the measured acceleration signal a(t), when a frequency weighting function W and a frequency band limiting function are applied. Different weighting functions are given depending on the human posture (see Fig. 9). The curve Wk is considered for the assessment of vertical vibrations regarding well-being and health for standing, seating or lying, while the curve Wm is applied when the posture of the body is undefined. The root-mean-square awT of the frequency-weighted acceleration aw (t) is the quadratic mean value, which is defined

The relationship between the root-mean-square of the frequency-weighted acceleration aw (t) and the subjective perception in the case of sinusoidal vibration is given in the Table 1. The primary purpose of ISO 2631-2:2003 “Vibration in buildings (1 Hz to 80 Hz)” [5] is to define methods of quantifying whole-body vibration in relation to comfort and annoyance of the occupants. The basic evaluation method is identical to that of VDI 2057, but a relationship between the root-mean-square of the frequency-weighted acceleration aw (t) and the subjective perception is 486

Acceleration rms [m/s2]

1 0.5

0.1 0.05

R=8 R=4

0.01

R = 14

0.005 Base Curve

0.001 1

2

4

6

8 10

20

40

60

80

Frequency [Hz]

Figure 10. Base curves and multiples acc. to “Design Guide on Vibration of Floors” [6]. Table 2. Comfort classification [6]. Environment

Reaction level A

Reaction level B

Offices Day Workshops Day Critical working areas (e.g. hospital) Residential Day Night

4 8 1 2 to 4 1,4

8 – – 4 to 8 3

missing. That confirms the difficulty of defining general reliable acceptance criteria for the human perception of vibrations. The “Design Guide on the Vibration of Floors” [6] by The Steel Construction Institute presents guidance for the design of floors in steel framed structures against vibrations caused by pedestrian traffic. It refers to a method described in BS6472:1992 [7] and ISO 2631-2:1989 to judge on the comfort of slabs for continuous vibrations. The measured building vibration should be frequency weighted and the third octave band spectrum of rms acceleration should be compared to the base curve (see Fig. 10). Satisfactory vibration magnitudes should be specified in multiples of the base curve magnitudes. Weighted acceleration values are to be evaluated with respect to the base acceleration magnitudes in the frequency band of maximum sensitivity. Reaction level A is postulated as “magnitudes below which the probability of adverse comments is low” and the values in of reaction level B “may result in adverse comments”. For the Korean requirements the walking induced vibrations of Slim-Floor slabs should be no more than easily perceptible or rather below level B for residential environments.

6 EXPERIMENTAL INVESTIGATIONS Three weeks after the systems’ erection the first vibration tests have been carried out on the pure concrete slab. After the application of the special screed of 90 mm for the floor-heating, the measurements have been repeated. The purpose has been to determine the influence of the screed on the dynamic properties of the structure. Hence 4 measurement series are conducted for the two slab systems. The measurement equipment consists of accelerometers by PCB, an Iotech Daqbook/260 and a notebook equipped with the software Diadem 8.00 by National Instruments for data acquisition. The accelerometers have been placed on various points to record the presumed mode shapes. The impact excitations have also been determined for these points. 487

Figure 11. Heel drop.

Figure 12. Hammer impact.

Figure 13. Walking test.

Table 3. Measured and calculated natural frequencies. LPM

HPM

Bare concrete

Concrete + screed

Bare concrete

Concrete + screed

Mode

Meas. [Hz]

Calc. [Hz]

Meas. [Hz]

Calc. [Hz]

Meas. [Hz]

Calc. [Hz]

Meas. [Hz]

Calc. [Hz]

1 2 3 4

8,180 9,650 15,75 16,48

8,378 9,809 12,333 16,016

7,810 9,28 – 15,38

7,975 8,626 12,498 14,131

3,050 5,740 – 7,45

3,021 6,069 6,084 7,507

2,810 5,12 – 7,080

2,772 5,048 5,858 6,928

Impact tests with heel-drops and an instrumented hammer have been carried out for the evaluation of the dynamic slab properties, i.e. natural frequencies, mode shapes and damping. A heel-drop test, see Figure 11, consists of a person who rises to the balls of his feet and then drops down impacting the floor. The advantage of a heel-drop test is the easy application and is performed quickly on different points. As even the impact of the heel drop by the same person may differ, the results are not precisely reproducible so that the test is considered controversial. Due to its simplicity and for comparison heel drop tests by the same person are performed. But it is noted that measured response of a heel drop do not allow to judge on the comfort due to walking excitation. A 10 kg hammer with a soft rubber tip and an accelerometer fixed upon is used for the impact hammer tests, see Figure 12. The hammer has an overall-length of app. 1000 mm. For the impact tests the operator drops the hammer from a height of app. 200 mm. In order to assess the floors according to current guidelines and codes and to judge on the comfort, it is necessary to monitor its response to a characteristic and realistic loading. Therefore walking tests with different number of persons are performed, see Figure 13, and the measured accelerations are taken as the input for the assessment of comfort. The natural frequencies and mode shapes are identified by peak picking of the modulus of the amplitude of the spectral density function and the sign of the appendant phase. But the change of phase due to damping is less than clear cut and the measured phase plot is very sensitive to noise. Hence the extracted mode shapes are not very precise. Consequently the mode shapes of the slabs tested have been evaluated by numerical modal analysis calibrated with the measured natural frequencies (see Table 3). The results show good compliance. The structural damping is depending on the mode and the vibration amplitude. An accurate determination of the damping is therefore difficult and a single value for the damping properties of 488

Figure 14. 1st mode shape of LPM slab.

Figure 15. 2nd mode shape of LPM slab.

Figure 16. 1st mode shape of HPM slab.

Figure 17. 2nd mode shape of HPM slab.

Table 4. Measured damping coefficient. Posco – LPM

Posco – HPM

Log. damping δ [−]

Bare concrete

Concrete + screed

Bare concrete

Concrete + screed

Decay curve Half power bandwidth

0,050 0,065

0,103 –

0,047 0,071

0,113 –

the structure cannot describe the behaviour properly. The measured decaying vibrations have been evaluated using the decay curve method and the half power bandwidth method. Walking tests consist of different load scenarios. E.g. one person walks in different directions (longitudinal to the main span, transversal to the main span, diagonal across the slab, along a circle) with different step frequencies (slow walking, normal walking and jogging); two or more persons walk in different configurations and randomly. The measured accelerations have been assessed according to VDI 2057 and the SCI-Guide on vibration of floors. It should be noted that the walking tests have been performed under laboratory conditions. The persons pace around and induce a continuous vibration over a long period. A real load scenario is when persons would just cross the slab. That will take a shorter time and hence the vibration will not build up as much as during the tests. Hence the measured and evaluated values are conservative. Additionally the support conditions are idealised ones. If the floors are implemented in an officebuilding, the structural frame has been based on continuous beams with fixed end conditions. That contributes to a significant increase of stiffness and natural damping and therefore decrease the vibration. For the assessment according to the “Design Guide on the Vibration of Floors” the fourth multiple of the base curve represents the limit of tolerable vibration for offices. For the LPM slab the vibrations due to a single person are well below that limit. But a jumping person and the walking of several persons creates higher amplitudes of vibration. For the HPM slab the vibrations due to a 489

1 0.5

0.1 0.05

Acceleration rms [m/s2]

Acceleration rms [m/s2]

1 0.5

R=8 R=4

0.01 0.005

R = 14 Base curve

0.001 0.0005 0.0001

0.1 0.05 R=8 0.01 0.005

R=4 R = 14

0.001 0.0005

Base curve

0.0001 1

2

4

6

8 10

20

40

60 80

1

2

4

Figure 18. Walking of one person on LPM slab without screed.

8 10

20

40

60 80

Figure 19. Random walking of 4 persons on LPM slab without screed. 1 0.5

0.1 0.05

Acceleration rms [m/s2]

1 0.5

Acceleration rms [m/s2]

6

Frequency [Hz]

Frequency [Hz]

R=8 R=4

0.01 0.005

R = 14 Base curve

0.001 0.0005 0.0001

0.1 0.05

R=8 R=4

0.01 0.005

R = 14 Base curve

0.001 0.0005 0.0001

1

2

4

6

8 10

20

40

60 80

Frequency [Hz]

1

2

4

6

8 10

20

40

60 80

Frequency [Hz]

Figure 20. Walking of one person on HPM slab without screed.

Figure 21. Running of one person on a circle on HPM slab without screed.

single person walking with half the first natural frequency are high above the limit curve (R = 4) as the slab is excited in resonance, while walking with another frequency is just acceptable. Running of one or more persons also excites unacceptable vibrations. The screed decreases the vibration amplitudes of both slabs slightly but do not alter the classification. The assessment according to VDI 2057 leads to similar conclusions. Walking of a person is easily perceptible on the LPM slab, while walking of several persons leads to strongly perceptible vibrations. Due to the screed the awT values decrease slightly but do not change the classification of perceptibility. For the HPM slab the vibration induced by one or two persons are classified as easily perceptible, even if the person is walking with half the natural frequency. Running or jumping leads to extremely perceptible vibrations. The vibrational behaviour induced walking loads under laboratory conditions show satisfying results for the LPM slab, while the HPM slab is acceptable for most types of walking but not for running and jumping as these activities coincide with the first natural frequency. The susceptibility to vibration of both slab system will be reduced by the stiffening effect of the real support conditions and the additional damping that is provided by friction, partition walls etc. The increase in natural frequencies are checked by means of modal analysis of FE-models of the complete slab systems.

7 CONCLUSIONS For two representative ground floor designs for high-rise residential apartments of the Korean market, one for a Low Price Market with a span of 5,2 m and another for a High Price Market 490

with a span of 9,6 m, optimised Slim-Floor systems have been developed with special regard to the Korean requirements. The most economic Slim-Floor system for the Low Price Market is the optimised HOLORIB- or ACE-deck system with a slab height of 180 mm. According to the test results it has been confirmed that this new slab system satisfies all criteria of the ULS and the SLS including crack width, deflection and comfort regarding walking-induced vibrations of single persons. As the damping will increase in the final slab system due to continuous action, partition walls etc., the vibration induced by several people walking will be acceptable. Generally the test proofed the slab’s suitability without any further modifications. For the High Price Market the HOESCH-Additive deck with a improved arrangement of the cams is recommended as the most favourable system. This slab of 280 mm height provides sufficient spans and is very economic. The test programme on one representative field has shown, that this slab meets all criteria of the ULS and crack width whereas the deflection limit is exceeded. FE-analyses for the final slab with continuous action reveal that fork end conditions for the main beams and a super-elevation for the self-weight satisfy the deflection limit criterion. Although the natural frequencies of the specimen are within a critical range for running and jumping induced loads, the vibration behaviour of the real Slim-Floor system is assumed to be satisfactory regarding comfort due to the increase in stiffness and damping by fork end conditions and partition walls. Considering these improvements this slab system is recommended as final design for the High Price Market. The investigations have shown that further improvements can be achieved by adapting the spans for the Low Price- and High Price Market to the favourable bearing system of each Slim-Floor system. The most economic span for the HOLORIB-deck is about 3,0 m and for the HOESCHAdditive-deck about 7,5 m. A modification of the ground floor designs to the most economic span would lead to another 20% of reduction on costs.

ACKNOWLEDGEMENT This research project has been carried out by order of the Studiengesellschaft Stahlanwendung e.V., Düsseldorf with a grant of Pohang Iron & Steel Co. LTD., Seoul, and the Stiftung Stahlanwendungsforschung, Essen. Sincere thanks are given to HOLORIB Germany GmbH, Offenbach and RIST, Seoul as well as HOESCH Siegerlandwerke, Siegen for providing profile sheets for the test programmes. Special thanks to the members of POSCO and RIST for close support, valuable references and contribution to the project.

REFERENCES [1] Final Report: Searching effective ways to make the steel framed residential apartment more competitive; Contract number P 534; Studiengesellschaft Stahlanwednung e.V., Düsseldorf; (2004) not published yet. [4] VDI 2057: Human exposure to mechanical vibrations – Whole body vibration, Verein Deutscher Ingenieure, 2002. [5] ISO 2631-2 (2003): Mechanical vibration and shock-Evaluation of human exposure to whole-body vibration, Part 2: Vibration in buildings. [6] SCI Publication 076: Design Guide on the Vibration of Floors, The Steel Construction Institute 1989. [7] BS 6472 (ISO 2631/2) Evaluation of human exposure to vibration in buildings.

491

Session 9: Urban design 2

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Sustainable design in construction sector L. Bragança & R. Mateus University of Minho, Department of Civil Engineering, Guimarães, Portugal

H. Koukkari VTT Building and Transport, Building Construction Technology, Espoo, Finland

ABSTRACT: Evaluation of the sustainable development – and on the other hand the effects of human activities on the future of the nature and mankind – is made on the basis of various indicators. The choice of indicators, the use of indicators and the evaluation procedure comprise both objective and subjective issues. The results of a sustainability evaluation procedure will be better interpretative and understandable when decision-making theories are applied. This concerns especially the valuation and use of weighting-factors that link the measurable and objective parameters of indicators. When the most important indicators are generally agreed, the differences between cases, countries or technological branches can be handled through weighting-factors. Further, dealing with the performance criteria of the built environment in relation to the sustainability can be advanced by application of decision-making methods.

1 INTRODUCTION Building construction affects each four topics of sustainable development that are ecological, economic, social and cultural development. The fundamental goal of sustainable development is to preserve the ecological systems that globally are the basis for human life and biodiversity of the nature. The European Commission has initiated a working group “Competitiveness of the Construction Industry”. According to this group, the challenges of the construction sector are: – – – – – – – –

environmentally friendly construction materials energy efficiency of buildings construction and demolition waste materials water conservation health in buildings building related transport aspects urban sustainability societal impacts arising from construction activities and from the built environment.

The policy-makers and decision-makers need reliable information on the current state of sustainable topics and future influences of human activities that is obtained on the basis of the best knowledge of the art. However, problems of fundamental nature still need to be solved in search of general acceptance of sustainable assessment methodologies. Arguments of scientific uncertainty have been used in objection of environmental taxes (Jenkins et al. 2002). Major reasons for difficulties are the political, technological and cultural differences of countries. They are also caused by the dependence of a subjective valuation involved in each general methods developed so far. Decision-making theories and practices have been recently used to overcome the imminent gaps between the goal of explicit results and the entanglement of basic values. Again, different approaches have been used. The novel methods do not withdraw the subject involvement but they help to understand its relation to different results. 495

2 SUSTAINABILITY EVALUATION WITH INDICATORS 2.1 Development of sustainability indicators Several countries follow the development of sustainability by the aid of indicators that cover the different topics. An indicator is usually a value derived from a combination of various parameters. A parameter is a measurable or observable property, which provides information about a phenomenon/environment/area with a significance extending beyond that directly associated with a value. Indicators have to be defined in a clear, transparent, unambiguous and right way, even before the concern whether they relate and evaluate several parameters. After this, it can be indispensable to define the areas of evaluation and the respective parameters. The indicators are in general defined according to their nature: driving force (pressure), state and response. Different indicators have been developed by administrations, organizations and industries at local, national and global levels. In addition, different parameters and their observation and assessment methods are in use. At the European level, the following ten indicators are defined by a Working Group nominated by the Commission in 1991: •

core indicators

•

supplementary indicators

– – – – – – – – – –

satisfaction of citizens impact on the climate change movability and traffic services access to services and green areas quality of air distances to school management systems of sustainable development noise sustainable land use products according to sustainable development

Use of sustainability indicators and parameters is based on definitions, rules, methods, classifications and weighting. In most of these phases, valuation and rating made by individuals are incorporated either in the development or in use of the methodology. 2.2 Sustainability indicators and eco-efficiency in the construction and real estate sector The sustainability indicators give information on the influences of the construction and real estate sector as a whole as well as on the impacts of planning, design, ownership and use phases of a building. They may be used in evaluation of a building, enterprise, sector or even a simple construction product, expressed by the aid of parameters. There exist different approaches to develop and use the indicators due to the local character of the sector and differences of societies, environment and geography. The indicators and accordingly the parameters are organized according to environmental, functional, economic and social criteria, often the two latter ones being combined. According to the European co-operative project CRISP, the Sustainable Development issues are: 1. Environmental: Natural raw materials including use of water. Bio-diversity. Energy. Environmental pollution. Land use. 2. Economic: Economic development and finance; indicators dealing with costs, productivity profitability. Production and consumption; indicators describe the quantity or quality of production or consumption. Urban and community services and responses; indicators dealing with economic responses etc. 3. Social: Access; access to buildings and built environment, barrier-free use, access to information, affordability. Safety and security; including crime, fear of crime, home safety, road safety, fire safety, industrial hazard, natural hazard, natural catastrophe. Health and comfort including sense of well-being (with regard to housing etc.). Community responses; including 496

social support, social exclusion, vitality of city/community/centre, stewardships, education for and understanding of sustainable development with regards to buildings and built environment, adaptive management ability, environmental management, spatial segregation, equity of minorities with regard to housing etc Cultural heritage. Eco-efficiency is related to the sustainable development, especially to the environmental and economic topics. According to OECD, eco-efficiency expresses the efficiency with which ecological resources are used to meet human needs. When speaking about challenges of Factor 10- or Factor 4- development, it is meant the goal to increase the eco-efficiency in industrial and development countries, respectively. Eco-efficiency is defined with terms that have different qualities. In order to get any value for the eco-efficiency, the different indicators of the input and output have to be analyzed and combined. The methods used in the LCA may be adopted. Indicators of eco-efficiency are not yet well established or practically experienced (Häkkinen et al. 2002). Proposals have been made both for principles to define the indicators and their subjects. The methodologies are debated and developed in several research institutes currently. 2.3 Environmental Life Cycle Analysis The construction and real estate sector has great responsibility for the change of the ecological systems as it: – consumes natural resources (materials, energy, water); – produces waste and emissions (e.g. greenhouse gases, sewing water); – takes land and forestry into civil use. The Life Cycle Analysis (LCA) is an evaluation procedure of the environmental effects of any kind of product, process or activity, from the cradle to grave (figure 1). The LCA is generally accepted to be applied for construction products, too. This analysis is made by identification, what has been taken from the environment and what has been brought back, by assessment of the potential harms due to these actions and by rating the significance of the impacts. The rating (or valuation) part of the LCA varies from country to country, and it is often neglected.

Raw materials extraction Energy

Product manufacturing

Use and maintenance Waste and emissions Final waste

Figure 1. Life Cycle of a product.

497

Recycling

Transport

Reutilization

Raw materials

Re-manufacturing

Material manufacturing

Objective Definition Interpretation

Inventory Analysis

Use (outside LCA)

Impact Analysis

Figure 2. Phases of the LCA.

The LCA includes three main phases that are inventory, analysis of potential effects and the interpretation (valuation) of the results (figure 2). The inventory phase includes identification and quantification of the consumption of energy and materials and the gases emitted to the environment. It can be completed by assessing the potential environmental effects that are climatic warming, acidification, eutrophication (excess thriving of aquatic flora), formation of photochemical oxidants, loss of ozone, harmfulness to health and ecotoxicity. Even though the framework of environmental analysis has achieved international unanimity, there is no detailed instruction or agreement about the method. This means that environmental analyses have been, and can be, carried out with varying boundaries and principles of valuation, which affects the result. Kaipiainen and Häkkinen (1996) emphasize that the principal endeavor of the whole methodology is the quantification of the result, even though they may also contain qualitatively described components. The major problem of the result is the multitude of figures, if the results have not been combined by valuation, and its openness to interpretation.

3 DEVELOPMENTS ON THE EVALUATION METHODS 3.1 General The relation of the various indicators shall be developed after the evaluation of each indicator in analysis. This relation is normally established giving an equal importance to all the indicators. The choice may be not the most correct one once the indicators are not expressed in the same order of magnitude and/or in the same unit. For example, the contribution of a material for the greenhouse effect is presented in the amount of carbon dioxide emitted, the acidification in equivalent of hydrogen ions, the electro fission in nitrogen equivalent, etc. On the other hand, the way that each parameter influences the sustainability is neither consensual nor unalterable along the time. So, it is difficult to express the sustainability of a solution in absolute terms, through an indicator that integrates all of the analyzed parameters and that allows the quantitative classification of a solution’s sustainability. Some systems and tools for the sustainability assessment are being implemented or in the development phase. Its application is complex and needs the previous knowledge of some data. Some of the sustainability assessment tools have datasheets that gather some of the needed data, although the data is related with the particular aspects of the country of origin, which turns its application in a different country very difficult. In most methodologies and tools of sustainability evaluation, the functionality aspects of construction solutions are forgotten. However, it is practically impossible to compare two distinct construction solutions that present exactly the same performance at all levels of the functional parameters, due to the existing limitations of the materials standard dimensions and their physical properties. It is also commonly known that with small concessions at the level of the economical and environmental performances it is possible to increase significantly the overall performance of a building, what significantly contributes for improvement of the solution’s sustainability. On the other hand, the use of the functional indicators on the sustainable assessment is an attempt 498

Table 1. Weight of environmental impacts according to EPA’s list. The categories in the Finnish application are shown with *. Impact category

Current consequences

Global warming* Acidification* Eutrofphication* Fossil fuel depletion Indoor air quality Habitat alteration Water intake Criteria air pollutants Smog Ecological toxicity Ozone depletion* Human health

Low High Medium Medium Medium Low Medium High High Medium – Low Low Medium – Low

to prevent errors from the past, where the concept “sustainable solution” has been associated to construction solutions with good environmental performance, but without fulfilling the necessary functional requirements (comfort, durability, etc.). The results of the assessment particularly depend on the analysed indicators and in the weight considered for each indicator. 3.2 Evaluating the environmental performance Interpretation (valuation) of information may be done according to the standard ISO 14042. It presents methods to analyze and assess the data concerning the emissions. The first step is to categorize the parameters identified in the inventory phase of the LCA based on their causeconsequence relations. Steps to calculate the indicator factor of each impact category includes determination of weight factors. The weight attributed to each indicator is given based on the following criteria: spatial scale of the impact, severity of the hazard, degree of exposure and risk for being wrong. With the knowledge of a qualitative evaluation it becomes necessary to convert the evaluation into a quantitative scale. Internationally, the interpretation of the results of the LCA is under rapid development. The methods of decision-making with decomposition and synthesis are in general applied in recent developments of the interpretation and valuation of the environmental indicators. In search for generally accepted indicators, it seems that the development leads to different weighting factors in different countries (Häkkinen et al. 2002). In Finland, the Decision Analysis Impact Assessment (DAIA) has been used to categorize the emission effects on atmosphere and waters. The categories to be considered are climate change, acidification, creation of ozone in the lower levels of atmosphere and eutrophication (excess thriving of aquatic flora). In the DAIA and other methods for rating, the factors also develop due to updating of knowledge on the real effects. In Portugal, the development of the interpretation and valuation phase of the LCA is an example of adoption of the EPA’s list (EPA, 2000), presented on Table 1, and application of an multi-criterion methodology of analysis that is based on the theory AHP (Analytic Hierarchy Process) presented by Saaty (1990). Through an AHP process as comparison pair to pair (Pairwise Comparison Value), the numerical comparison is attributed to each one of the possible pairs in the list of the qualitative values. Thus we can determine the number of times that the weight of a parameter must be higher than another one and establish a relation between all the parameters in study (Table 2). 499

Table 2. Pairwise comparison value. Verbal importance comparison

Pairwise comparasion values

Highest vs. Low Highest vs. Medium Highest vs. High High vs. Low High vs. Medium Medium vs. Low

6 3 1.5 4 2 2

Table 3. Evaluation of impacts. Impact category Global warming Acidification Eutrofphication Fossil fuel depletion Indoor air quality Habitat alteration Water intake Criteria air pollutants Smog Ecological toxicity Ozone depletion Human health Total

Current consequences 8 impacts (weight %)

Current consequences 12 impacts (weight %)

24 8 8 8 16 24 4 8

16 5 5 5 11 16 3 6 6 11 5 11

100

100

A quantitative evaluation for 8 and 12 parameters is presented in the following Table 3. The presentation of these two classifications is linked to the possibility of not having all the data needed in the 12 parameters evaluation, being only possible the 8 impacts evaluation. 3.3 Evaluating the functional performance The analysis and comparison of the performance of construction solutions has to be carried out at the level of each element (interior walls, exterior walls, floor, roof, etc.), therefore each one of them present distinct requirements. The first step for the evaluation is to define functional indicators and parameters. The six essential requirements and durability according to the Construction Products Directive form a regulated basis for consideration, e.g. thermal insulation, airborne and impact sound insulation, flexibility of natural illumination, structural stability, air permeability, etc. The quantification of these indicators is relatively simple according to various proven methods. However, the way each indicator influences on the performance and, therefore, the sustainability of a solution is not consensual. The evaluation involves subjective rating and depends above all on the type of use of the solution, as well as on socio-economic and cultural heritage of the subject. In a first phase it can be considered that all functional indicators have the same weight in the evaluation of the functional performance. In order to obtain more consensual values it can be made interviews to the potential users in order to identify which indicators are considered more important. Through the application of a Multi-attribute Decision Analysis, p.e. the Analytic Hierarchy Process methodology, is possible to quantify the weight of each one of the indicators. Synthesizing all the functional indicators in one number it’s possible to obtain the functional performance of the solution. 500

3.4 Evaluation of the economic performance There are several costs associated to the life cycle of a building or material/construction system: costs of the materials (it includes extraction of raw materials, production, transport for the construction place), costs of construction, costs of utilization, costs of maintenance, costs of rehabilitation, cost of demolition and costs of devolution to the natural environment, recycling or of reutilization. Measuring the economic performance of a building is more straightforward than measuring the environmental performance. Standardized methodologies and quantitative published data are readily available. Life-cycle cost analysis (LCCA) is a method for assessing the total cost of a facility ownership. It takes in account all costs of acquiring, owning, and disposing of a building or building system. LCCA is especially useful when project alternatives that fulfill the same performance requirements, but differ with respect to initial cost and operating costs, have to be compared in order to select the one that maximizes the net savings. Thus, it’s relatively easy to foresee the total costs associated to the life cycle of a building. As less will be the costs foreseen for a construction solution, better it will be the economical performance and more sustainable will be the solution. In order to have a complete economical performance, the analysis of the costs associated to the solution’s life cycle must be made, including the residual value. The residual value of a system (or component) is its remaining value at the end of the study period, or at the time it is replaced during the study period. Residual values can be based on value in place, resale value, salvage value, or scrape value, net of any selling, conversion, or disposal costs. As a rule of thumb, the residual value of a system with remaining useful life in place can be calculated by linearly prorating its initial costs. For example, for a system with an expected useful life of 15 years, which was installed 5 years before the end of the study period, the residual value would be approximately 2/3 (=15– 5/15) of its initial cost. As it’s well known, the construction solutions are very distinct at the level of the durability. It is essential to use the same study period for each alternative whose LCCs are to be compared according to the stakeholder perspective. For example, a homeowner would select a study period based on the length of time he or she expects to live in the house, whereas a long-term owner/occupant of an office building might select a study period based on the life of the building.

3.5 Balancing the environmental, functional and economic performance After comparing the solutions in each group of indicators (environmental, functional and economic) it is necessary to classify globally the sustainability of the solution. The way as each group of indicators influences the sustainability is not consensual. So, it is acceptable, in a first step, that the three groups of indicators present the same weight. The experience shows that the most compatible alternatives with the environment are generally the most expensive. However, considering that with the implementation of the concept “sustainable construction” is intended a bigger compatibility between the artificial and the natural environments, without compromising the functional performance, easily it’s understood that the weight of the environmental and functional indicators must be higher than the weight of the economic indicators in the sustainability evaluation. The Equation 1 shows how the three groups of indicators could be balanced, in order to determine an absolute value (sustainable score – SS), that expresses the sustainability of a solution (Bragança et al. 2004).

where, SS = sustainable score; W1 = weight of the environmental parameters’ group; WEnp(i) = weight of the environmental parameter (i); Enp(i) = value of the environmental parameter 501

(i); W2 = weight of the functional parameters’ group; WFp(i) = weight of the functional parameter (i); Fp(i) = value of the functional parameter (i); W3 = weight of the economic parameters’ group; WEcp(i) = weight of the economic parameter (i); Ecp(i) = value of the economic parameter (i); m = number of environmental parameters under analysis; n = number of economical parameters under analysis; and o = number of economical parameters under analysis. 4 CONCLUSIONS Sustainable design, construction and use of buildings are based on the evaluation of the environmental pressure, functional aspects (related to the users and the local building codes) and life-cycle costs. There is an environmental effect when something is taken from the environment as a resource or returned to it as waste or emissions, which weakens or threatens the availability of resources, the livable environment and the human health. The sustainable design searches a bigger compatibility between the artificial and the natural environments without compromising the functional requirements of the buildings and the costs associated. The accomplishment of a LCA provides an excellent support tool for the decisions making, however, its necessary much time and money due to the necessary amount of information. One of the limitations of the tool is that it does not guarantee that a company who uses it has its products relatively more “environmental-friendly” than others, therefore can not be used as marketing. In the international scene, the responsible entities for the technical standards, from which the world guides its production, has been extremely presented. Among them it is distinguished ISO (International Organization for Standardization), in which it detaches norm ISO 14000 that is one of the most excellent tools of environmental management. With the knowledge of the weights of each of the environmental indicators, it is possible to determine the value of a unique environmental indicator. With the knowledge of the economic indicators, functional and even the socio-cultural, and the application of the multi-criterion analysis methodology, it is possible to determine a unique global value for evaluation of the sustainability. Future developments on the sustainable assessment methodologies should bring more consensual lists of indicators to be evaluated and more consensual weighting factors. REFERENCES Bragança, L. & Mateus, R. 2004. Sustainability Assessment Datasheet. COST C12 – Final Report. Häkkinen, T. & Kaipiainen. M. 1996. Ecological criteria in building planning. Helsinki: Building Information Institute. 52 p. (In Finnish). Häkkinen T. et al. (2002) Eco-efficiency in the building and real estate sector. Helsinki: Ministry of Environment, Housing and Building Department, the Finnish Environment 580. 165 p. ISO 14042:2000 – Environmental Management – Life Cycle Assessment – Life Cycle Impact Assessment: establishes the guidelines for the Life Cycle Impact Assessment of a LCA study. Jenkins R. et al. (2002). Environmental regulation in the new global economy, The impact on industry and competitiveness. Glos, UK: Edward Elgar Publishing Ltd. 349 p. Saaty, T. L. 1990. How to make a decision: The analytic hierarchy process. In the European Journal of Operational Research, Vol. 48, No. 1, pp. 9–26. United States Environmental Agency (EPA), Science Advisory Board (SAB), 2000. Toward integrated Environmental Decision Making. EPA-SAB-EC-00-011, Washington, D.C.

502

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Aesthetics in urban design seen from the perspective of sustainability C.M. Ravesloot, L. Apon & E.M. Boelman Delft University of Technology, Delft, The Netherlands

ABSTRACT: In the paper the working of aesthetics as part of the conditions for designer to be developed in sustainable urban design and architecture, is explained. Next the possibilities of designers to make use of the special effects provided by their design are depicted from the perspective of aesthetics as part of sustainability. These designs could again cause psychological imprints of the mind, leading to behavioural change. The idea is to make a theoretical framework of design knowledge and psychological insight and knowledge of how the brain copes with aesthetic imprints. From this framework, a very easy to use guideline was elaborated to be used in designing new urban areas with stress on sustainability and aesthetics.

1 INTRODUCTION Physical objects leave visual imprints to people’s minds while they are watching urban areas and architecture. These imprints lead to psychological effects, sometimes interpreted as an aesthetic experience. The development of new urban design, with special emphasis on sustainability, might also have specific effect on the psychological effect of the brain. Sustainability, being the development of more efficient technology to preserve future potential to our next generation, consists of an improvement of technical efficiency, but also out of influence of new forms and designs on human behaviour. Not only technological innovations can improve sustainability, also human behaviour can. Designing new urban areas with stress on sustainability and aesthetics can be improved by understanding the working of aesthetics as part of the conditions a designer meet. The most powerful cultural aspect of architecture is its pleasing quality. It is the architect’s challenge to provide for artful buildings that like all art reflect society and may also express new architectural meanings and ideologies. From the beginning of this century especially the modern architects tried to design buildings that were an answer to social changes, being directly or indirectly the result of industrialization in the building process. Architects interpreted this as a road toward more wealth and comfort. By now we are well aware of the fact that industrialization also had less positive effects on architecture. This can be seen in the massive repetition of prefabricated building elements that cause dull architecture, especially in housing. However, it is possible to give sustainable architecture such significance that we might counter the destructive trend in society and in architecture in particular. The imperative argument is that beauty might contribute to the success of sustainable architecture and make it acceptable and contributing to a trend-break.

2 DEFINITION OF SUSTAINABILITY Sustainability is here defined as the creation of technology that decreases environmental pressure, thus ensuring that future generations will be able to use the same environmental qualities as we do. 503

Environment is the set of conditions with which humans, animals, plants and minerals can develop optimally. In short environment is the set of conditions for life (de Jong 1996). If the set of conditions is not optimal, minerals, plants, animals and humans can not develop optimally, there is an environmental problem. Conditions for life can be designed and organised by engineers. Aesthetics are part of the conditions for life of humans to develop optimally. To improve our future potential and to clean up environmental problems for the past we need to improve sustainable technology with a factor 20. D = P×W ×M 1/2 = 2 × 5 × 1/20 D = environmental pressure P = world population W = welfare M = efficiency of environmental technology = technical efficiency × behavioural change According to Erhlich and Speth, environmental pressure has to decrease by half, in the context of growing world-population, doubling and a fair share of welfare, meaning increasing five times, concluding a given improvement of environmental technology by factor twenty (Ehrlich & Speth 1993). The environmental efficiency improvement can be designed and organised by engineers from a combination of techniques and behavioural change of humans (Lovins L. et al. 1995). The techniques of environmental improvement can be qualified in three categories: – Energy neutral building, minimum of fossil energy with maximum of sustainable energy. – Material neutral building, closed cycles of water, waste and material-use, with a minimum of environmental damage, possibly with environmental benefits. – Healthy and comfortable building, minimum of health risks for humans, animals and plants, maximum change of survival and optimal development. 3 AESTHETICS AS CONDITION FOR LIFE Aesthetics, being part of the health and comfort conditions for life, and being a result of engineering can have a positive effect on the use of environmental techniques and on behaviour of mankind. Approximately 80% of environmental technology however is impossible to see. It is part of construction and hidden inside the building structure. This part of technical design does not have any effect on the aesthetic quality of the build up area. The visible 20% could be designed on basis of optimal mix of repetitive and varying forms. Environmental technology does not need to be perceived as aesthetically disturbing. Decision-making is done on basis of arguments in conditional order (de Jong 1996). The first condition is technical feasibility. If a technique is not working efficiently yet, it is no use on a larger scale. The second condition is economical feasibility. Economical feasibility includes organizational, legal and process barriers before financing can become a problem. If there is no finance nor return on investment, the change of realizing large scale technology for sustainable development is very small. The third condition is social feasibility. Not always a positive argument from a sound technical and economical feasibility is followed by a positive decision-making at the highest level of leadership. In practice also emotional and social–cultural arguments can have big influence on decision-making. Aesthetics in architecture can be categorized under social–cultural arguments (Berlyne 1972). If a very profitable and environmentally safe technique is ugly, so to say, it will not be used. 3.1 Effect on design Conditions for life can be put into a diagram called ecological tolerance diagram. In this diagram the probability of survival of a certain ecosystem, like for instance a group of water-plants in a 504

swamp or a social coherent group of people in a city, is set out against the quality of a condition for life which are for instance the presence of water or the aesthetic quality of a city. If conditions are too bad, to dry or to much repetition of similar architectural forms, the change of survival is limited. On the other hand if the conditions are too good, to much water in a swamp or too many explicit architectural forms in a city, survival changes are also limited. The area between too little and too high is the zone were optimal conditions for life can be found. Figure 1 shows the conditions for life related to the probability for survival according to de Jong (1996). 3.2 Modelling Assuming we can reduce the aesthetic perception of urban areas and architecture to the simple counting of repetitive and varying objects and forms, we can put this in the diagram as well. Figure 2 shows the aesthetic quality of urban design and architecture related to the probability of approval. According to design research of de Jong and Ravesloot, this way of arguing can be viable under a certain set of conditions (de Jong & Ravesloot 1995).

Figure 1. The conditions of life related to the probability for survival (de Jong).

Figure 2. Aesthetic quality of urban areas related to the probability of approval (de Jong & Ravesloot).

505

4 ANALYSIS OF AESTHETIC QUALITY 4.1 Basis of the analysis method By accepting the simplification of aesthetic impressions into counting repetition and variation, aesthetics in architecture can be put into an analysis method. This was first done in a case study by de Jong and Ravesloot in 1995 in the Amsterdam quarter De Baarsjes (Fig. 3). From the analysis of aesthetic quality by de Jong and Ravesloot in the Amsterdam quarter De Baarsjes we can conclude that indeed it is possible to count repeating and varying elements in several scale’s of urban design and architecture. We can however not make clear that visual quality can be substituted by beauty. The visual quality of ‘De Baarsjes’ has only been rough. It might be a forbidden simplification. But for the sake of the argument it can be assumed that beauty can be partly represented in its simplified form of visual quality. In the above assumed context of visual quality, the experiment to substitute visual quality for presented form in terms of repetition and variation is more familiar to the vocabulary of environmental psychologists such as Michael Stephan (1994). The argumentation for the importance of beauty for architecture could be as follows: – In ecology all physical conditions for life follow a similar rule expressed in the tolerance-curve (de Jong 1996). – In psychology variations in form, related to time and scale, do arouse potential positive and negative valuation (Berlyne 1972, Gombroich 1982, Hekkert 1994). Figure 4 shows the resemblance between the tolerance curve of de Jong (1996) and the stimulation potential curve of Berlyne (1972). This provides the only justification for the assumption of relating visual quality and beauty. 4.2 Counting repetition and variation To calculate our count the repetition and variation in buildings, it is absolutely imperative that the counting is done within each scale separately. According to de Jong and Ravesloot this is the only way to make valid conclusions. The so called scale-paradox can otherwise mess up all conclusions. As an example of a possible analysis a building is shown in the Figure 5 below. The analysis can result in the conclusion that within the scale of circle A only one form is seen, a window. However within the scale of circle B two similar windows can be seen. Within the scale

Figure 3. Urban design of the Amsterdam quarter De Baarsjes.

506

of circle C even more similar window can be counted, although this time in the same grouping of pairs like within scale B was seen. This analysis was tested in a case study in Amsterdam. The Quarter De Baarsjes was built in the period 1920–1940. The urban design and architecture is characterised by repetitions in the scales of 1000 m, 100 m, 10 m, 1 m and 10 cm. The scales in between of 300 m, 30 m, 3 m, and 30 cm have more variations. The detailing of the connection between elements and components was often elaborated by architects, craftsmen and even artists. With this knowledge, sustainable architecture can, in the long run, profit from repetition from industrialisation, leading to cost-reductions without endangering aesthetics. As long as enough variation in different scales is ensured a mix of variation and repetition, within the optimal zone, can be realised. In an example shown in Figure 6, the elevation was altered with some more variation on the scale level of circle D, one higher than C. Also the connection between two C scales was varied by bringing in some colour. The B components were changed a used to make the connexion between scales. The repetition of the window, as building components, however was not altered, leaving the potential for industrialisation in tact. Also in product-development and industrial design it is accepted that beauty of manmade things can be influenced by designing, as well as that the aesthetic perceptions of products can be anticipated by designers (Hekkert 1995; McAllister 1996).

Figure 4. Tolerance-curve by de Jong and stimulation potential curve by Berlyne.

B

B A

B A

A

B A

A

B A

Figure 5. Image of a building elevation of a multi-storey multifamily house.

507

B

C

B A

B A

A

B A

A

B A

C

Figure 6. Image of a building after demodelling according to more varying elements.

4.3 Discussion on the analysis method In the history of philosophy, art and architecture have been expressing (spiritual) knowledge and moral (Murdoch 1993). This means that architecture and the built environment are strong enough to carry some cultural significance. So I put forward the hope and the possibility that sustainable architecture, seen in this tradition, could carry the message of a trend-break toward sustainable living. According to McAllister beauty has been playing a major role in accepting new paradigm’s in science and technology (McAllister 1996). Seeing colour, movement, depth and form, is not only neurological appointed, but is also connected to our personal classifications and values and is part of our life-environment. Therefore the V4 part in the visual cortex not only measures physical information but also exchanges information with several other parts of the brain (Sacks 1995). This means that indeed, also from a neuro-psychological point of view evidence can be found for the thesis that perception is a matter of sight and interpretation. Sacks provides strong indication that visual information can arouse social and ideological significance. The brain can interpret contextual meaning from visual imprints. Cranenburgh and Lurija seem to support this thesis with their research. Micheal Stephan summarises this as the ‘possession of full existential life’ (Stephan 1994). Ben van Cranenburgh states that the right and left hemisphere do not work separately, do not oppose to each other, do work parallel and do play an equal role in communication; We can (learn to) use the mechanism of assigning attention of perceptions to the right and left hemisphere ourselves (Stephan 1994). A.R. Lurija shows that the highly structured cortex is, after reception and transportation of the stimulus, responsible for planning, regulating and controlling human visual perception and aesthetic valuation (Stephan 1994). Michael Stephan finally concludes from these statements about that images are not perceived in visual isolation, disembodied from the world to which they refer, but are possessed of full existential life. The interesting part of Stephan’s ‘Transformational Theory of Aesthetics’ is that variations in the set of known experiences would arouse some emotional reaction. Variation is a trigger for brain activity. Architecture can do this too, by providing a proper set of repeating and varying elements and would supposedly be able to arouse some emotional tension and reaction. Stephan explains this as follows: There is a perceptual paradox between the perceived image and the full existential life (equivocal reality). This gives reason for emotional arousal in the percipient, seeking for an emotional outlet. This outlet is provided by the image itself, expressed in a concentrated projection (cathetic hemisphere). The image appears to possess an emotional tension provided by the right 508

Figure 7. The housing project Le Schroupf in Geneva.

hemisphere, which is also translated by the left hemisphere, thus leading to verbal art-criticism and the perception of beauty. In other words: what I see is a variation of what I saw before, so it must be beautiful or not. At least it makes me think, according to Stephan. The challenge, then, would be to design equivocal aesthetic experience, i.e. architectural images which contain a mixture of surprising variation in a sequence of recognisable repetition. Practically this would enhance prefabrication of differing elements. Buildings would become compositions of repeating elements with surprising variations in the building details and building knots perhaps (Ravesloot & Thöne 1996). Many questions remain unanswered, but the most important one would be: Would a proper equilibrium of repeating and varying elements help people accept sustainable architecture, because it makes them think or just makes them feel at ease? The housing project Le Schroupf, in Geneva by the architect Christian Hunzicker, was designed against the machinelike architecture of modernism, but not without the benefits of building industrialization (Fig. 7). His architecture leads to variation and repetition within buildings (Scheider 1986). There is a clear touch of industrialisation, but by means of colour and craftsman-shift in forming ornamentation the buildings do not seem to be dull. There are enough surprising details to look at. 5 EXAMPLES OF APPLICATION The optimum aesthetic is biases by cultural context and by personal opinion and perception. However, the concept of contemporary architecture is often characterized by a high number of repetitive elements in different parts of the buildings (Fig. 8). Within scientific acceptable tolerances it can be put that the excessive use of repeating form elements a boring not appealing image might occur that might in most cases not be seen as aesthetic pleasing (Gombrich 1982). 6 CONCLUSIONS The introduction of sustainable building techniques in the practice of architecture is a daring enterprise. On the one hand we see a strong focus on technical feasibility and the challenge to 509

Figure 8. Examples of an analysis of the skyline of down-town New York and a high-rise building in Berlin.

improve efficiency, on the other we have to take into account the tendency in our society to obstruct or slow down innovations. Improving the efficiency in architecture is not only a technical problem. Technically we have the answer, but the legal, financial and social structure of our society does not support the introduction of sustainable architecture. The most important question would be: Would a proper equilibrium of repeating and varying elements help people accept sustainable architecture, because it makes them think or just makes them feel at ease? REFERENCES Berlyne D. E. 1971. Aesthetics and Psychology, Appleton Century Crofts, New York. Ehrlich & Speth, 1993. Can the world be saved, Ecological Economics Volume 2. Gombrich Ernst H. 1982. Ornament und Kunst, Schmucktrieb und Ordnungssinn in der Psychologie des dekorativen Schaffens, Klett Cotta Stuttgart. Hekkert P. 1995. Artful Judgements, a psychological inquiry into aesthetic preference for visual patterns, Delft University of Technology. Jong T. M. de & Ravesloot C. M. 1995. Beeldkwaliteitsplan De Baarsjes, Stadsdeel De Baarsjes, Amsterdam. Jong T. M. de 1996. Essays over Variatie, Publicatiebureau Bouwkunde, Delft University of Technology. Lovins L. et al. 1995. Faktor 4, doppelter Wohlstand, halbierter Naturverbrauch, Droemer Knaur, München. McAllister J. W. 1996. Beauty and Revolution in Science, Cornell University Press, Ithaca. Ravesloot C. M. & Thöne V. 1996. Modularer Holzbau unter esthetischen, ökologischen und ökonomischen Aspekten, Das Bauzentrum, Verlag Das Beispiel, Darmstadt. Schneider R. 1986. Architektur die sich fühlt, Christian Hunzicker, Le Schtroumpf in Genf, Edition Fricke Köln. Stephan M. 1994. Transformational Theory of Aesthetics, New York-London. Thöne V. & Ravesloot C. M. 1996. Method for designing connections between components, Conference, Detail Design in Architecture, Nene College Northampton.

510

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Social demands and stakeholders participation in Dutch sustainable housing policy C.M. Ravesloot Delft University of Technology, Delft, The Netherlands

ABSTRACT: The paper describes a new tactical management tool developed for Dutch municipalities to incorporate all stakeholders in the development process of sustainable housing. By the aid of this tool municipalities can aim at higher technical ambitions and solve technical problems considering sustainable technology in housing and still create social feasibility with all the stakeholders. Especially the social demands of the future inhabitants of the houses can be fully addressed. The paper shows the results of a four year case study in the city of The Hague were the tactical tool was developed. The paper also shows the theoretical and scientific background of the creative problem solving capacity in the tool as well as of the theoretical and scientific background of the process management tool that creates economical and social feasibility.

1 INTRODUCTION The direct link between environmental problems and the way we have to solve them is easier to understand when one considers the environment as “the total set of conditions for life, plants, animals, human beings” (de Jong 1992). The direction that we have to follow is quite clearly explained: we have to start analyzing the conditions in which humans and life can optimally develop, while considering human needs and looking for them. In the Netherlands the concept of “sustainable building” was officially introduced in 1990 by the National Environmental Policy Plan-Plus to minimise environmental problems during the whole life cycle of a building, a district or a town (VROM, 1990). It is a means of improving the quality of buildings and their environment.

2 DUTCH POLICY ON SUSTAINABLE BUILDING Sustainable building was primarily aimed at improving the environmental efficiency of buildings, but nowadays the objectives have diversified. Today it is also about the sustainable development, maintenance and renovation of buildings and their environment as a whole, including the demolition and removal of buildings. Sustainable building as a standard part of the planning, of design, of construction and of management process is overriding importance when it comes to improving our living and working environment. It therefore means building with a clear view to a healthy future. In its ultimate state of equilibrium sustainable building leads to: – Energy neutral building: the absence of fossil energy in favour of durable energy like solar heat, heat pumps, geothermal heat and photovoltaic electricity. In the Netherlands the goal is to reach energy neutrality by the year 2030. – Material neutral building: the building with balanced locally administratively closed systems of building materials, waste treatment and water to avoid environmental hazards, damage and pollution. The goal is to build with industrial, flexible and de-mountable components in housing. Material neutral would be a situation of being able to return all building materials, components and elements to some kind of recycling process. 511

– Healthy building: the building of safe, healthy and comfortable buildings, cities and landscapes for people, animals and plants, with a maximum of bio-diversity. 2.1 Role of Municipalities Municipalities have to aim at high ambitions to achieve energy neutral housing, low environmental effects on water, material and waste during the life cycle of the houses and still have to achieve healthy and comfortable accommodations. However, the existing laws and building code only provide for minimal demands on sustainability. The problem of diffusion of energy-neutral housing and sustainable building seems to be a matter of market imperfection. For this reason, it seems inevitable to incorporate the stakeholders, like future inhabitants, architects, housing developers and the municipality itself to achieve more sustainability. Especially future inhabitants can have an important influence on the feasibility of sustainable housing, because they can balance out market-imperfections. The hindering obstacles all originate from an immature organisation structure for sustainable development. For sustainable development people have to be moved (practically and emotionally) to make new paths for sustainable development. At the beginning of new processes for sustainable building they are not interested (Schot et al. 1996-a). Unfortunately in common processes the responsible engineers and designers just do not have the experience with these new possibilities in environmental spatial design. The challenge of renovating the existing buildings and designing new sustainable houses lies in the enormous technical and organisational complexity. The approach to this task must be interdisciplinary and based on newest management tools. Every step of the process must be newly designed and newly trained, because it has never been done before. This especially refers to the facts that there is no communication about technical possibilities or about feasibility; there is not enough problem solving activity and action; there is no clear decision-making; there is no learning from previous projects; there is no possibility to find out who or what is obstructing the process, because the process itself is not always transparent. 2.2 Covenants with stakeholders from the building sector Legislation can be considered as a framework for negotiating operating conditions (subject to standards being met), but the legislative process is slow and compliance and enforcement can conflict with business interests. Both Government and industry have recognized the need for a complementary approach which would allow greater speed, flexibility and efficiency at every level and deliver real environmental improvements: Covenants. They involve co-operation between government, all the relevant parties in the construction process and target groups, so that everyone can make a significant contribution to a better environment (for example by switching to more environmentally friendly materials, by using sustainable building techniques or taking measures to conserve energy, etc.). By means of these covenants, the parties concerned specify the measures that they will voluntarily take to promote sustainable building in new construction projects and in the management of residential and non-residential buildings. The fact that most of the government’s covenants are addressed prevalently to the building industry is not that surprising. The exploitation of natural resources, the manufacture of building materials, transport in the construction phase, energy and water consumption in buildings, and demolition waste: all have a detrimental effect on the environment. At the beginning of 1999, for example, half of the raw materials consumed (for some raw materials this percentage was even higher) and about 40% of total energy consumption could be laid at the construction industry’s door. Covenants are being used within industry as implementation instruments in areas where legislation already exists and government can exercise control (for example through issuing licences). In such cases, covenants serve as a management tool by providing a concrete implementation program within a more general legal framework; they are not an alternative to regulation and they do not take precedence over existing law. Covenants represent a commitment by industry sectors to play their part in meeting the environmental objectives established in the National Environmental Policy Plans. Also municipalities fulfil an important stimulating role in achieving sustainable building in practice, 512

both by fixing a level of ambition as well as by maintaining it. From this point of view covenants are an important aid, in fact, only in one fifth of the councils is there not (yet) a sustainable building covenant. For example, the development of sustainable building in the existing housing stock has received a strong stimulus through a covenant with the social sector and the “Temporary Stimulating Measure on Sustainable Building”. On the basis of this measure around 56 000 existing dwellings have been renovated in a sustainable way with an average contribution of about a910 per dwelling. However, after evaluation of several sorts of covenants regulating environmental problems, the working of the covenants is not overall positive. The covenants do have a limited effect on the behaviour of stake-holders and do have the habit of getting worn out. This conclusion introduced the need for a stronger way to bind stake-holders to a common task in environmental protection. 3 NEED FOR STAKEHOLDER PARTICIPATION Especially future inhabitants can have an important influence on the feasibility of sustainable housing. Investments in housing have a direct influence on the sales prices and rent of the house. As long as the house can be sold and rented, any investment in comfort and environmental techniques can be allowed. All stake-holders, but the eventual inhabitant, being owner or renting the place, benefit from higher investments because their financial margins also go up. Many investments in environmental technology can be defended against the background of the benefits that the inhabitants have during the use of the house. Investments in energy saving and in sustainable energy reduce the energy costs and improve the indoor comfort. Inhabitants will not object against these costs. Van El (1992) estimated that from all technical measures that can be taken to build sustainable houses more than 70% will not provoke any objection from inhabitants at all. The remaining 30% of the measures will only for half of the cases present a problem because the costs somehow are higher than the benefits in energy saving or improvement of comfort. Sustainable building is not static: manufacturers regularly launch new sustainable products into the market, and legislation and regulations, relating to building and environment, are amended over and over again. As a consequence, knowledge about sustainable building has to be able to develop apace. The history of sustainable building in the Netherlands shows us that the technology is not the problem, the sticking point is the necessary change in culture, attitudes and working practices. Sustainability has to be perceived as unremarkable “as the most normal thing in the world”, and this requires long-term dedication from all those concerned. Today, more and more parties that are involved in the design of our built-up environment are coming into contact with sustainable building. The level of knowledge of the building parties is rising, even though not all aspects of sustainable building are equally well known and the actors do not all have the same level of experience and knowledge. Researches have underlined the importance of inter-personal information transfer and the importance of accessible information also for professional groups and households. 4 METHODOLOGY FOR UN URBAN PROJECT: ROMBO TACTICS The Rombo tactics reflects a new awareness of quality assessment to answer all the questions of the municipality in one single question: how to produce an organization that can work effectively on the high ambitions. The Rombo tactic is used in urban projects on the scale of city quarters and smaller (re) development areas, but is also appropriate for the design of individual buildings and for policymaking. From earlier research it has been known that user involvement in sustainable building is an important condition for a successful innovation process. For that reason users are present and co-operate in this egalitarian approach. The Rombo tactic tries to combine the ideals of appropriate technology and the management tools of construction technology assessment with the powerful decision making tool of consent. The quality circle is maintained by delegating three tasks: leading the process, execution of actions to run the process and monitoring the results, according to the tactical approach developed by Endenburg (Endenburg 1997). 513

4.1 Processing decision-making To speed up the process and to follow the highest ambitions, it is important to realise that Rombo is a “consent” conference, not at all a consensus conference. According to Endenburg’s theories consent means that the point of decision is reached if no party present has any substantial objection against a proposal. This proposal will be executed. Consensus would mean that every party present agrees on the proposal, or that the majority can implement a decision against the will of a large minority. Consent is the bottom line of consensus and easier to achieve. In the consent process, the opinion of the minority asks for solutions, without delaying the process too much. The process is arranged for a decision making structure, built from a circle, in which consent governs. This decision making structure includes all members of the organisation and makes shared decisions on policy and work preparation issues. After the decision has been made, the authority of execution is delegated to the leader of the process (that generally is the chairman of the workshop), assisted by an extra appointed person who’s only task is to monitor the results of the execution of tasks. A typical ROMBO-tactics works in three sessions on nine steps. Leading/Direction Execution Monitoring Building a vision

Discuss feasibility Collect arguments Decision making

Municipality or principal of project All participants, especially architects and engineers Subsidising parties and the Municipality Build 1 results of earlier session 2 inventory of alternatives 3 consent on agenda 4 problem definition 5 economical feasibility 6 consent on strategies 7 presentation alternatives 8 selecting priorities 9 consent on proposals

4.2 Building a vision The spatial and environmental quality of the existing situation will be studied at first on every scale, from landscaping to interior architecture and building services. This exercise can be performed for example on three themes of environmental and spatial planning, according to the strategic Factor 20 policy. Every theme has to include the subjects in the following way (subjects in the brackets): Energy (energy saving, production of renewal energy, transportation), Material (integral water management, integral waste management, integral building life cycle and material management) and Health (green areas, vegetation and urban climate, safety, comfort). The result of the analysis on theme and scale can be put in one single matrix on one piece of paper size A4. By the aid of scheduling, the Rombo procedure changes a normal design sequence in favour of environmental technology. The existing spatial and environmental quality is written in a special matrix in the left lower half. From here you can move on by designing and generating new idea’s to improve the existing situation. The results of this brainstorm have to be put in the right upper half of the matrix. Then, finished the matrix, it becomes possible to see in one overview the tension between minimum performance and maximum goals. That is what has to be organised to improve the environmental and spatial quality in this special case. During this exercise economical or social feasibility are not to be considered, because these are for the next step to come. Next the same exercise has to be repeated by looking for a further level of ambition. After having repeated this exercise four or five times (from the frame to the grain), the result is a pile of papers. The matrices can be connected vertically now, thus forming the relationship of possible technical improvements in one theme through several scales. By now you have enough information to check if design features on a later scale have to be organised earlier in 514

order to make them more feasible. The leftovers from this last check are the techniques difficult to realise and they are to be left alone for later. 4.3 Collecting arguments As a result of the first phase in the ROMBO strategy, you have a pile of promising and a pile of more difficult strategies or practical measures. Now it has to be made an inventory of conditions under which it is possible to organise the execution. For the promising measures you might not have to do anything, but for the difficult strategies, technical and scientific knowledge and price information might have to be collected. It could be necessary that the members of the workshops need to make research by themselves (in order to find the necessary information and proofs for the further development of the analysis). In these cases there could be more workshops during this phase. The result of this second phase is a collection of arguments in favour of certain environmental and spatial objectives. In some cases there are sort of checklists to supports this exercise. Some of the questions that have to find an answer in this phase are: How do I execute and organise the maximum quality in environmental and spatial design as efficient and as effective as possible? What are the expected investment costs, the process and maintenance costs? What environmental, social and financial profits are feasible and who will benefit? What subsidies and tax reductions can be obtained? Which social revenues can be expected? 4.4 Decision-making When you have collected all the arguments, a detailed proposal can be put forward to the city council for a final approval. In this way it is possible to show the governors that the best job possible within this context has been done, the goals have been set and the means for execution have been organised. Now it is up to the city council to take a wise decision. All participants have been forced to design simultaneously and interdisciplinary through several layers of scale (landscaping, town planning, urban design, architecture and building technology) considering both spatial and environmental aspects in this process, during several workshops. The final result is always presented in compact sheets. 4.5 Creative problem solving instrument A tactic like Rombo can be seen as a memory aid to plan design steps. During every next step to be taken Rombo reminds you of environmental design decisions not to be forgotten and it shows technical possibilities in spatial planning. Rombo provides the design- and decision-making tools on time and within the specified technical and socioeconomic context. With this method of interdisciplinary collaboration of planners, city clerks, designers, urban designers, architects and engineers, we can prevent design decisions on higher scales blocking design decisions on lower design scales’. The ingredients in the Rombo tactic that are meant to help solving the technical problem are known from the work of de Jong (de Jong 1992; de Jong 2000) from niche technology management (Schot et al. 1996-a). The problem-solving has the potential to be successful because the designers and all stake-holders can work together on the basis of equality. Equality is provoked by the use of the consent decision-making. The equality provides general access to all knowledge in the group. The free access to knowledge enlarges the group knowledge, but is also allows interdisciplinary collaboration in the group. The combination of problem solving and process instrument guarantees that the problem either is solved and everybody has contributed to the solution or the group has decided that given the circumstances that the problem can not or only partly be solved. 5 CASE-STUDY THE HAGUE The city The Hague activities started before 1990 during the first years of national awareness in sustainable development and the first national environmental policy. From 1990 on, the city produces 515

a continuous number of experiments and pilot projects in the field of sustainable development (Ravesloot et al. 1999). However from 1998 on the increase in amount of project and complexity of projects stopped. The city policy lost effect. During that time also the national government stepped back in its direct implementation of sustainable building policies. The further development was left to the stake-holders in an open planning process. The city council of The Hague however wanted to continue a strong and directive influence on the building industry, despite the decreasing direct support from the national government. The questions the municipality wondered about were: 1. Can participative design processes improve the quality of innovations from the perspective of users? 2. Do users have design-relevant knowledge and how can this knowledge be incorporated into technological development? 3. Is it possible to organise user involvement effectively or do such strategies belong to the realm of symbolic policy-making?

5.1 Social participation instrument Within the Rombo tactic, seven different project-types, recognised by the economical relationship between participant like the public authority, the principal and the end-user. Within every one of these types at least one sort of participation for end-consumers was developed. The participation can be direct, because the renting or buying future inhabitant is already present and can involve him or herself in the Rombo tactic process or if the future inhabitant is not present, he or she can be replaced by consumer organisations or by stake-holders representing them. In that case we call it a indirect participation. There are five general conditions that have to be fulfilled to make direct or indirect participation valuable: The planning-process is clearly depicted in separate steps and is clearly planned in time and time-usage. The planning-process will be clearly visible ended with a decision that also reflects the motivations and arguments of the participating future inhabitants and other participating stakeholders. Inhabitants should be able to judge the terms and conditions for newly developed projects. Inhabitants should be able to count on professional support to participate in developing processes and in design workshops. The participation of inhabitants should be formalised in the the starting documents. The Rombo procedure knows to types of social participation. In the direct participation, when the inhabitants are known, like in retrofitting projects or the process of buying and renting is reversed, the following conditions are attached: 1. The representatives have been chosen decoratively, if possible. 2. The representatives are being educated in the use of consent decision-making and matrices of Rombo tactics. 3. The representatives have support and mandates from their group. 4. The representatives must have free access to all information to prepare their discussion, they can expect support from professionals if necessary. 5. The representatives do have enough knowledge and experience in communication in public debates, they have studied the planning process and do comply with the terms and conditions from Rombo tactics as well as usual legal constrains from the municipality. The indirect type of participation is supported by a simulation of the input from future inhabitants, if they can not be involved directly, there are also special conditions attached to this indirectly type of participation: 1. The representatives are aware of their right to bring arguments into the process of decision-making on behalf of future consumers. 516

2. The representatives are familiar with the way of the usual form of consumer participation organised by the municipality or the housing corporation. They will try to gain insight in the needs of the groups of inhabitants they represent. In the Rombo tactic seven possible participation processes can be distinguished. The diversification is based on the relationship between the stake-holders who is, or who are, investing in the process and the final consumer. 5.2 User participation in the quarter of Duindorp The Duindorp area in the Hague consists of about 6000 houses, built in the first 20 years of the 20th century. A total of 1100 houses will be demolished in the next 10 years. Only 750 houses will return in their place. This idea caused a revolt in the neighbourhood, with a lot of social action. The discussions brought mistrust between the municipality and the tenants (60%) and home owners (40%) of Duindorp. After some time of negotiation a “social covenant” was signed, where the right of re-housing, the financial margins and the participative decision making process was agreed upon. Two different community organisations had spokespersons and after ups and downs the co-operation of these (informal) representative bodies improved. At the beginning of the planning process some 35 persons were in different ways involved in the community. They were invited for discussion sessions on sustainable housing. Twenty four people responded to the invitation. The wide limits of sustainable housing were illustrated with an introductory “fun” session. Topics were: Demographic changes in the near future will create new housing needs; How will old people live, feel secure and how do we take care of them? New democratic principles will prevail: the social involvement of citizens is needed, because new liberalism produces low administrative support for solving community problems; Do you like to live “wild”, to express yourself in new housing designs? Thinking in terms of financial cost-benefit is past time. Quality of life is the issue. Sustainability has its price, even if it does not increase comfort of living; Treat water as gold; Air as is your daily food. During the more serious sessions, everyone worked in small groups on design issues. The group decided on the main issues: flexibility, healthy indoor environment, high standard of materials and construction details, improvement of the total area and retrofitting the existing houses included. The results were integrated in conceptual solutions. Some sparkling new ideas showed how much fun the participants had while talking about their environment. There was great commitment to promoting the quality of the neighbourhood. The ideas were wrapped into three metaphors: the ecological house: energy efficient and healthy; the rubber house: flexible and accessible; salt, sand and wind: low in maintenance and environmental friendly. The planned new constructions and retrofits were to be developed according to these three metaphors. The architect was given both the freedom and the responsibility to think of solutions that work and that respond to the metaphors. For the outdoor environment and the existing housing stock a number of adjusted measures were proposed. In general the users agree with the national policy towards sustainable building, but in the case of Duindorp, they refuse to act as guinea pigs. New materials must have a certified quality, delivery of heat must be monitored individually, solar installations must be well integrated in the roof structure. The experience as user is clearly evident in the demand for sound insulation, also with regard to building services. Easy and understandable controls are wanted. Users pay much attention to the outdoor environment: new buildings must increase the visual quality of existing buildings and existing buildings deserve a retrofit. In terms of sustainable development, the goals are set at Duindorp: low energy demand for housing (envelope Rc = 4–5, glazing U < 1,2, boiler efficiency 107%); heat pump and sea water as collective heating source; wind mills in the sea, PV systems on roofs; high quality building materials and re-use of recycled materials and components (components for pavements and landscaping, recycled materials for construction purposes); keep rainwater in the area (sandy soil along the coast) and make the process of re-use visible; design with natural elements storm and sand-dust, salt and water: special selection of materials, details, wind screens; apply more wood as building material. 517

The conditions above would lead to the first Dutch existing quarter that would be converted to an energy-neutral quarter. In spring 2004 the housing corporation Vestia signed the contract for the realisation of the sea-water heat-pump installation. The investment was supported by the municipality of the Hague with a considerable amount of money (Ravesloot, C.M. 2003). The ideas and position of the inhabitants of Duindorp are in itself consent. They could not think of any substantial objection against the proposed sustainable measures: do what you want with environmental technology, as long as we keep comfortable and healthy.

6 CONCLUSION In the case of the renovation and replacing of houses in theThe Hague quarter Duindorp participation was an important part of the Rombo procedure. The problem was solved fast because all participants could be free to access the Rombo group and could influence the process as well as the problem solving. The innovative approach of Rombo tactics is the combination of a process instrument, a problem solving instrument and the participation of all stake-holders on a basis of equality. This new approach to sustainable development is called Constructive Appropriate Technology (CAT) because the roots of the Rombo tactic is found in appropriate technology, technology assessment and in sociocracy and niche technology management. REFERENCES Endenburg, G. 1981. Sociocracy; the organisation of decision making. Rotterdam: Sociocratic Centre The Netherlands. Endenburg, G. 1997. Sociocratie als sociaal ontwerp. Delft: Eburon. Grin, J. & van de Graaf, H. 1996. Technology Assessment as Learning, Science. Technology and Human values 20(1): 72–99. Hasselaar, E. & Ravesloot C.M. 2001. User Involvement in Innovation in Sustainable Architecture, The case of Duindorp, The Hague, The Netherlands. In, User Involvement In Technological Innovation, Deutschlandsberg, Austria, July 8 – 133rd International Summer Academy on Technology Studies; Jong T.M. de, 1992. Kleine Methodologie van Ontwerpend. Onderzoek, Boom Meppel, 1992; Jong T.M. de & Cuperus Voordt van der. 2000. Ways to Study, architectural, urban and technical design. Delft: Delft University Press. Ravesloot C.M. 1994. Invloed bewonersgroepen op woningbouwprojecten; in Schmid P. (ed)Gezond Bouwen & Wonen, Uitgeverij van Westering Baarn Ravesloot C.M., Bakker H.E. et al. 1999. Duurzaam Bouwen in Den Haag 1990–1999. The Hague: Municipality of The Hague, Vilaricca Publishing Baarn. Ravesloot, C.M. & Bakker H.E. 2000. Rombo, Spatial Planning and Environment Policy Design Strategy. The Hague: Gemeente Den Haag Wellicht, News Bulletin on Energy Saving in The Hague, nr 10. Ravesloot, C.M. 2003. Energieneutrale nieuwbouw met zeewater in Scheveningen-Duindorp, Marktberichten. In Ravesloot, C.M. (ed.) Gezond Bouwen en Wonen 2003–4, uitgeverij van Westering Baarn; Riedijk, W. 1989. Appropriate Technology in Industrialized Countries. Delft: Delft University Press. Schot, J., Rip, A. & Misa T.J. Ed. 1996-a. Managing Technology in Society. The Approach of Constructive Technology Assessment. New York: Pinter Publishers. Schot, J., Slob, A. Hoogma R. 1996-b. De implementatie van duurzame technologie als een strategisch niche-management problem. Rapport voor het interdepartementale onderzoeksprogramma Duurzame Technologische Ontwikkeling. Enschede: Twente University. Serini, M. 2002. Architettura e Sostenibilita, Strategie Politiche, Economiche Ed Ambientali Nella Programmazione E Gestione Del Patrimonio Edilizio, Torino: Master Thesis Politechnico di Torino, Carlo Ostorero & Delft University of Technology, Christoph Maria Ravesloot. VROM 1990. Nationaal MilieubeleidsPlan plus, supplement duurzaam bouwen, Ministry of VROM The Hague.

518

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Energy neutral retrofitting of apartment flats – modelling and detailing with consent of inhabitants C.M. Ravesloot, L. Apon & E.M. Boelman Delft University of Technology, Delft, The Netherlands

ABSTRACT: The Dutch policy for sustainable housing aims at refurbishment of existing housing complexes. The economical conditions for a successful redevelopment are lowest were the income of the inhabitants is lowest. However, the multi-storey buildings are technically easy to adjust to higher standards of insulation, glazing and ventilation. The energy consumption for heating drops drastically because of the optimal ratio between the surface of the walls and roofs adjacent to the outdoor climate and the heated volume. Calculations show that zero-energy for heating is technically feasible. If the social demands of the inhabitants are incorporated, studies show that inhabitants will not have any objections towards most of the techniques involved to reduce energy consumption due to transmission and ventilation. On the contrary, it will save them money yet it will increase their indoor air quality and thermal comfort.

1 INTRODUCTION In The Netherlands there are many post-war quarters, built between 1945 and 1973. In order to satisfy the market demand for dwellings, many houses had to be build quickly just after the second world war. New industrialised building systems were developed to increase the speed of building. Especially in the sixties it became possible to produce high rise prefabricated apartment flats. The post-war high rise apartment flats do not meet housing standards of contemporary inhabitants anymore. People with higher incomes move out and are replaced by people with lower incomes. The social decay is in progress as is the economical decay in the quarter, also affecting the financial situation of the housing corporation. In order to turn the tide and to give the apartment flats a new strong position in the market of renting, severe technical and social measures have to be taken. The Dutch government and the housing corporations are challenged to renovate thousands of apartment flats within a relatively short time span. High-rise apartment buildings do have a favourable quotient between volume and outer surface, which in theory can keep the energy use for heating relatively low. However the past war apartment buildings are mostly not insulated and the energy use for heating is very high. As a consequence the thermal comfort of the apartments is very low. In order to improve the energy-performance of the apartment-flat building two options are considered: Firstly, decrease the use of fossil energy for heating by minimising heat losses from transmission and ventilation and secondly, increase gains from sustainable sources like directly or indirectly from solar radiation. In order to improve the energy-performance of the apartment-falt buildings, a design approach was developed, targeting the improvement of energy performance as well as thermal comfort in the Delft post-war quarter “Poptahof” (Fig. 1). Although the research specifically aims at zero-energy design, the results as well as the method can also be used an any general design process, that tries to integrate building design, energy conservation and comfort criteria. 519

Figure 1. The appartment flat quarter “Poptahof ” in Delft The Netherlands.

Figure 2. The ground-floor plan of the case study apartment-flat Poptahof.

2 METHODOLOGY The goal of this research is the development of tools to design a zero-energy building with a comfortable indoor climate. Next to the technical results, the research also generates a new method to design such a building. Contrary to normal design procedures, in this research a case study was used to reverse the spatial design process. In the new method first data were collected and analysed, then a design was generated to fulfil the energy requirements that will meet energy standards of zero-energy building within parameters for comfortable indoor climate. Knowledge about the existing situation was essential for a solid basis and realistic goal. From the case study was concluded that, if properly analysed, design aspects regarding energy and comfort, will be integrated efficiently in a (re)design for a zero-energy building. The first step in the research was to study the general energy household of a genuine post-war apartment-flat building. Then a specific case, the Poptahof apartment-flat in Delft was analysed, specifically for the existing building conditions (Fig. 2). To get a complete overview on the influence of several technical parameters influencing the design of the elevation and influencing the energy use as well as the indoor climate of the dwelling, a parameter research was done. A computer programme for the dynamic simulation of the building and its heat exchange was used. Especially low- and high temperature, heat use and energy use for each room can be simulated. The results of these parameter calculations were used as basis for the appointment of conditions and priorities for the building redesign towards zero-energy. 520

Figure 3. Energy balance of post-war apartment-flats in general (Apon, 2003).

2.1 Energy use for heating and ventilation in post-war flats The calculated consumption of an apartment is as follows: Internal heat production, solar radiation ventilation losses, transmission losses, heating,

ap. 2000 kWh ap. 2000 kWh ap. 1700 kWh ap. 6000 kWh ap. 19000 kWh

The energy use for ventilation and heating of an apartment flat is influenced by several factors (Fig. 3). On one side there is input of energy from the sun and internal sources. On the other side there are losses from ventilation and transmission of heat. If in summer season the dwelling gets to warm, the extra ventilation to cool the house also has to be taken into account. 2.2 Current situation in the apartment-flats The current situation of the flats was analysed based on information gathered by visiting the flats and from drawings. The analysis shows the following characteristics: – – – –

single glazing, big windows, no sunscreens very little insulation of the construction many heat-losses from not insulated detailing bad ventilation-system (with problems like insufficient ventilation, smell, dirty intake) shafts, unreliable functioning, not controlled energy-losses – bad heating installation (no radiators in some rooms, noise) The heat loss from transmission are relatively big. This is due to the inferior insulation of the elevation. Also the heat losses from ventilation form a considerable amount of energy in the balance. At the side of the heat gains the interior heat and the energy gains from sun radiation are not big at all. The balance has to be completed with use of heating from fossil energy to keep the apartment-flats warm in winter. There is no contribution from sustainable energy. 2.3 Problem definition The possibilities of new building technology, especially in the outer skin (envelope) of the buildings to enhance the energy-performance and to improve the indoor air-quality and thermal comfort are essential means to generate a future for the existing flat-buildings. After the research data were formed and processed, the conclusions were used to redesign the building with equal influence on the new design as the architectural and functional aspects. Later 521

these aspects were matched with the research in the other case study were the influence of inhabitants on the decision-making was researched. 3 PARAMETER-STUDY (MODEL DEFINITION) In this study the energy use, usually on basis of one year on the scale of the total apartment-flat building, as well as the thermal comfort, usually on basis of hourly calculations and in one room within an apartment-flat, were calculated. The influence of the following building parameters have been studied: U-value of the glazing, U-value of the construction and sun-blocking and shading from building parts. 3.1 Computer programme and model The computer-programme VA114 was used for the parameter study. This dynamic building simulation programme was developed by the organisations VABI and TNO. With this model indoor temperatures, transgressions of indoor comfort, heat use and cooling need of separate rooms can be calculated. For the calculations in this case study the climate data from the year 1964 were used. Normally this dynamic simulation programme is used to design and fine tune cooling and heating installations in existing buildings. In this specific case study the programme was used to make preliminary calculations advancing changes in the design of the building in order to achieve zero-energy heating balance. Within the programme a maximum of three rooms can be modelled. The construction can be modelled up to a maximum of 10 differently materialized layers. Because of the some complex constructions with more than 7 materials had to be abstracted before the input would fit into de programming. Dimensions and geometrical form of rooms, situating of doors and windows, shading and other building parts can also be modelled. Because of the fact that there is only three room modelling available in the programme, a selection had to be made of the most critical side of the apartment-flat. The room oriented to the west side of the floor-plan was chosen to be most representative, because of the possible heat problems during summer season. At this side of the apartment-flat the living room and an extra room are situated. Another reason for this choice is the absence of shading from the outside. Apart from the simulation model for the calculation of heat losses, a calculation for the possible gain of solar energy through solar collectors and solar cells was made. 3.2 Parameters To be able to combine these two items in this case study, the basis of the calculations were set for energy use and comfort within a year, a season and on hourly basis. To be able to analyse the influence of the selected parameter objectively, all the others possible aspects of any influence on the energy behaviour have been set to a fixed value or to zero. There is no changed input for the internal heat gains, nor for the ventilation-rate, the temperature-settings in the room nor to the heating system. Three different kinds of glass were used in the calculations. These glass types were calculated in combination with a U-value of the closed construction elements of 0,5 W/m2 K. Three different closed construction elements were calculated. The different U-values are caused by the addition of three different thickness of cellulose insulation board (Lambda = 0,036 W/mK). All variations in closed construction U values have been combined with double glass (U-value = 2,0 W/m2 K) and so called HR++ glass (U-value = 1,0 W/m2 K). The effect of the form and position of a window was studied by using different percentages of glass. The calculations were performed for the glass percentages of 25, 50 and 75%. There were also two variations with 50% glass and one with 25% glass. The 50% glass variation was modelled horizontally, the other one of 25% vertically. The influence of the balcony form was calculated in comparison with a building with no balconies at all and with a small and deep balcony or a long and wide balcony. After that, the influence of 522

shading outside the windows was studied. The first calculation describes the comfort and energy use with the shading always down at a maximum. The second one is programmed to go down if the sun-radiation exceeds the level of 250 W/m2 . The estimation of 250 W/m2 should be comparable to genuine behaviour of inhabitants. To improve the overall impression from the different parameters settings and its influence on the thermal conditions in the modelled room, all varying calculations were put together in one diagram. 3.3 Results of calculations for energy use The yearly energy use for the heating of room with respect to different calculated variables can be presented visually and numerically. The results are presented in Figure 4. The influence of different parameters can be described as in Table 1. The computer-programme can not distinguish heat losses due to transmission from ventilation. Because of the constant ventilation with all calculations, it can be reasoned that the transmission losses (also at night) will increase with the increase of window area. The results also show that the use of better insulated glass has effect on the energy use during the year. Moreover it can be concluded, that the automatic regulation of the shading also causes a decrease in yearly energy use. The amount of energy saved is as big as the amount of energy saved by decreasing the u value of the glass from 2 W/m2 K to 1 W/m2 K.

Figure 4. The calculated results for influence of different parameters on the energy use. Table 1. The influences of different parameters to the energy use. Parameter

Description of results

U glass

At a big glass percentage (75% glass) the type of glass has a big influence on the energy use. By decreasing the U value, also the yearly energy use will decrease. Energy use decreases in relationship to the insulation value of the closed construction elements. However, this influence is considerably smaller than the influence of the glass-percentage. The gain of energy from solar radiation decreases, when the area of window decreases. Because of the balcony the energy use increases a little. This small raise can be due to the small decrease in solar gains caused by the shading from the balcony. When the shading is always completely down, the energy use is much bigger than when the shading is not calculated. The absence of solar gains has to be compensated by the use of extra fossil energy. The rate of decrease of solar gains is hardly of any influence on the yearly energy use.

U closed Glass % Balcony Shading

523

3.4 Results of calculations for over-heating The results of calculations for over-heating are presented in Figure 5. In Table 2, the results of the calculations for indoor comfort (over-heating) during the summerseason are presented. 3.5 Recommendations for further research on energy saving The calculations showed the relative influence of building design parameters (e.g. thermal insulation, sun-shading) on energy needs for space heating and thermal comfort. The results also clearly indicated the need to consider space heating energy needs together with thermal comfort and overheating risk. In this research the influence of some parameters was researched calculating by a limited number of parameters, given a certain reference climate. The validity of the research therefore also is limited to situations were the actual situation matches the reference almost completely.

Figure 5. The risk of over-heating. Table 2. The influences of different parameters to risks of over-heating. Parameter

Description of the influence

U glass

With a big glass-surface, the type of glazing has some influence on the maximum number of hours that the comfort temperature of 25◦ C and 28◦ C have been exceeded. However, if this happens for instance with better insulating glass-types the LTA and ZTa values also get lower. This will eventually result in a longer duration of the period that the indoor-temperature exceeds 25◦ C and 28◦ C. The influence of the insulation of the closed construction-elements, at a smaller glass-percentage, is small. The glass-percentage plays a major role during summer. The graph shows that the maximum temperature decreases fast and the hours exceeding comfort-temperature also decrease, when the glass percentage gets smaller. By making a balcony the hours of discomfort due to overheating will be cut half. A broad balcony even decreases exceeding comfort-hours to 40%. The use of shading is the most effective way of protection against overheating. If the shading is always down, the overheating is also at a minimum risk. This will of course make the dwelling practically useless. Therefore the next best solution will be an automatic shading, which can also be adjusted to gain solar energy in winter.

U closed Glass %

Balcony Shading

524

The influences of parameters on the indoor climate must be seen as indicators of a trend. Only a small number of possible variations were modelled and calculated. To increase the validity of the research, many more variants and changing parameters should be calculated. Especially for the relation between the u value of the closed construction elements and the glass percentage, such an in-deep research could show valuable. Besides that, more research could be done on the influence of climate control installations and possible alternative installations. In such a research the influence of inhabitants could be worth wile to be modelled as well. The influence of decreasing night temperature in the rooms, was found of significant influence on the energy-use. Further research would be recommendable. 4 PRODUCTION OF RENEWABLE ENERGY Although the energy use of apartment-flat buildings can be minimised, there still is a need for heat and electricity. In general houses can be built and modulated in a technical way that these energy needs for heating and electricity can be fulfilled by putting solar collectors and solar cells on the roof (Ravesloot, 2003). Solar collectors produce warm water out of sun-radiation. Solar cells produce electricity out of solar radiation. In typical state of the art houses that were retrofitted or newly build according to the concept of energy-neutral building, the roof always was big enough to cover enough collectors and cells to produce renewable energy. If this heat and electricity was stored efficiently, a house can be energy neutral on a year time basis. However, considering the volume of apartment-flat buildings related to the outside surface of roofs and elevation, in some cases the roofs and elevations are just not big enough to produce enough renewable energy on the building. Renewable heat and electricity have to come from outside the building in those cases. Research of models with apartment-flat buildings show that up to four storeys most buildings can be energy neutral. The roof provides enough space to place solar modules for heat and electricity. From four to eight storeys, the gains from solar energy has to compared to the heat losses due to the bigger transportation distances from roof to storage and from there to the users in the dwellings. But by using elevations for production of renewable energy, the balance can be energy-neutral in most cases. From eight storeys up the production of renewable energy can not keep pace with energy losses and transport-losses (Felsch, 2001). That is unless, the energy losses are not cut down again, by more energy saving measures. 5 PRELIMINARY CONCLUSIONS The idea of the research was not only to find answers to the design problem how to find the most influential parameters to the indoor climate and the energy use. The second part of the research had the goal of investigating the usefulness of such an approach of calculating first and then designing according to the findings. It was expected that by calculating first, the redesign of the energy balance of the apartment-flat would be made more efficient. The results of the research did indeed confirm the expectation. The new approach can be repeated with some minor modifications. After that an architect can use the calculations to create more grip on his design. Less design variations have to be elaborated before the most convenient and effective design is found. If many calculations could be found and many different types of apartmentflat building would be modelled and calculated, even generally valid conclusions can be drawn. These general conclusions could lead to simplified design directions, eventually making energy calculations of this kind obsolete. 6 CONSENT FROM INHABITANTS Parallel to the research in techniques to reduce energy use in apartment-flat buildings, a survey was executed to find the limits of social economic feasibility for the inhabitants of these kinds of 525

buildings. Participation and acceptation of inhabitants has a long tradition in Dutch social housing. The case-study was done in the city of Soest in the quarter Smitsveen (El van 2001). Basis for the categorisation of the technical measures that can be taken in these kind of buildings was the national package of sustainable building for retrofitting. The conclusion is that from al the possible measures to be taken, half are according to wishes and needs of existing inhabitants. Only 20% can count on objections from inhabitants because they cause high investments in combination with long terms of repaying. Especially in the range of measures contributing to energy-neutral retrofitting, only the technique of recovering heat ventilation air has to be considered possibly unacceptable, because of the changing of filters on a regular basis. If the filters are not replaced the indoor air can become hazardous to the human health. In general, energy-neutral retrofitting of apartment-flat buildings can, if communicated properly, be supported by inhabitants.

7 CONCLUSIONS The retrofitting of apartment-flat buildings up to eight storeys high can be technically feasible. The production of renewable energy on the building is the obstructing factor. In most cases the techniques also contribute to the wishes of existing inhabitants to improve there level of comfort. In winter comfort levels will rise. In summer the indoor temperature needs not to be compromised by over-heating problems. The inhabitants can give consent to almost all changes, if the economical side effects are communicated clear and can be compensated if necessary. The method to model energy and comfort can possibly be used in other types of buildings for housing as well. The Dutch context might diver from the given preconditions in other European countries, but as it seems at least the technical method to research the energy behaviour of apartmentflat buildings can be used Europe wide. The successful application of energy-neutral retrofitting has to be adjusted to local practice and tradition in participation of inhabitants. REFERENCES Apon L., 2003, Energy neutral renovation of an apartment-flat building, graduation thesis, Delft University of Technology, faculty of Architecture, Building Technology department. BAK, 2001. Basisonderzoek Aardgasverbruik Kleinverbruikers 2000, Resultaten van een enquête onder ruim 3000 huishoudens, EnergieNed. BEK, 2001, Basisonderzoek Elektriciteitsverbruik Kleinverbruikers 2000, Resultaten van een enquête onder ruim 3000 huishoudens, EnergieNed. El J. van, Leven in Lagen, Living in Layers, towards a livebla and sustainable apartmentflat-building, graduation thesis, Delft University of Technology, Faculty of Architecture, Housing department. Felsch J., 2001, Maximalisatiemodel voor PV panelen, Maximalisationmodel for PV panels on a apartmentflatbuilding with variable height, N.Sc. graduation thesis, Delft University of Technology, faculty of Architecture, Building Technology department. Hulsbosch S. Boelman, E. Dijk van. E. Ravesloot C.M., 2002. Flexible buildings and cellulose insulation, Sustainable Building 2002 International Conference The Challenge • The Knowledge • The Solutions, September 23–25 2002 Oslo. Novem 1984. Energiebesparing bij woningverbetering, Projectbureau Energie, Novem Arnhem/Utrecht. Ravesloot C.M., 2003, Technical and economical feasibility of energy neutral housing, COST C12 EU working group, Lisbon meeting 2003. Schalkoort T.A.J., 2001. Klimaatinstallaties handboek, Delft University of Technology.

526

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Sustainability assessment of new construction technologies: a comparative case study H. Gervásio GIPAMB Ltd., Lisbon, Portugal

L. Simões da Silva University of Coimbra, Coimbra, Portugal

L. Bragança University of Minho, Guimarães, Portugal

ABSTRACT: In the framework of mixed building technology, different construction methods and materials are combined in the same building, allowing structures to achieve higher quality, safety, economy and to improve sustainability. Buildings are one of the major concerns regarding the sustainability of our habitat. Construction industry consumes materials and energy and is one of the main causes of pollution and resource depletion. The use of environmentally friendly materials and cleaner construction methodologies are a major step towards the well being of the environment and foster sustainable development. In this paper the results of a comparative analysis, in view of a sustainable development, conducted upon a typical residential floor plan of a family-house located in Algarve, south of Portugal, is presented. 1 INTRODUCTION Mixed building technology integrates different construction methods into a single building. Mixed construction allows to maximize structural and architectural advantages in combining components made of different materials. In this new way of construction the different materials may work together or independently, but will always provide advantages over the use of a single material. To illustrate this statement a comparative study between two different construction solutions of a residential house is performed. Special emphasis is given to the sustainability of each solution. The construction methods depend on the framing system considered for the house and of the type of external and internal walls. The comparative analysis will evaluate the different solutions in terms of building physics and sustainability, beyond the technical performance that each and every solution must fulfill. In building physics the thermal and the acoustic behaviors of each system are analyzed. In this stage of the study all the solutions will be balanced, so that there will not be a “favored” solution. Subsequently, a sustainability analysis is carried out. This analysis will evaluate which system achieves the most appropriate balance between environmental and economic performance, given all the previous requirements. 2 PROJECT OVERVIEW 2.1 Plan and location A typical residential floor plan representing a family house in Portugal is selected for the analysis and it is represented in Fig. 1. This house is located in the Algarve with the main façade facing west. The floor plan has a total area of 141.5 m2 . 527

N Bedroom 18.72m2

Master Bedroom 22.78m2

Bedroom 10.40m2

3.28m2 I.S.

CL.

6.30m2 I.S. 6.66m2 CL.

CL.

CL.

CL.

CL.

CL. 2.80m2

Kitchen 20.03m2

Living room 36.87m2

Figure 1. Floor plan of a family house. Table 1. Alternative solutions. Case

A) Concrete

B) Composite

Columns Top slab External walls Internal walls

Reinforced concrete Reinforced concrete Double clay brick wall Simple clay brick wall

Hot-rolled steel Steel and concrete Light gauge steel Light gauge steel

2.2 Selected alternatives Two alternatives are analyzed in this paper and summarized in table 1. Case A represents the most common constructive solution in Portugal. The structural frame is made of reinforced concrete grade C25/30, composed of exterior columns with a cross-section 30 × 50 cm, interior columns with 30 × 30 cm, beams with 30 × 40 cm and top slab with a thickness of 15 cm. The exterior walls are made of a double clay brick layer, 15 cm on the outside and 11 cm inside, and an air space with 4 cm width, filled with mineral wood for insulation. The interior walls are made of a single clay brick layer. A 5 cm layer of mineral wood covers the top slab. The wall surfaces are covered, on both faces, with stucco and latex paint. Case B is a composite solution, made of a steel frame grade S355, with columns made of hot-rolled profiles HEB200/240, beams IPE200/220, and a composite slab with a total thickness of 15 cm. The exterior walls are made of OSB panels, light gauge steel frame, plasterboards and insulation (mineral wool), with a total thickness of 30 cm. The interior walls are made of OSB panels, light gauge steel frame and plasterboards, with a total thickness of 25 cm. The top slab is also covered by a layer of mineral wood and by plasterboards underneath. The wall surfaces are covered, on both faces, with latex paint. In table 1, only the elements that are different in both cases are shown, the foundations, the ground slab and the roof being kept constant in both cases, so that these elements will not be considered in the following study. Both structures were design to satisfy the requirements of the relevant Eurocodes. The Algarve is located in the most potential seismic area in Portugal, therefore the structural analysis was made taking into consideration the seismic load and a dynamic analysis was performed accordingly. Both structures were defined to have exactly the same structural performance and safety factors. 528

Figure 2. “Classic” concrete solution (a) External wall (b) Internal wall.

Figure 3. Composite solution (a) External wall (b) Internal wall.

The construction details for Case A are shown in Figure 2. The construction details for Case B are shown in Figure 3. Because of the inherent complexity of achieving unbiased comparable designs, the functional requirements such as thermal behavior and acoustic behavior, were balanced for both alternative solutions. A life cycle environmental analysis and life cycle cost analysis are performed for the different solutions. In those analyses only the components that differ across the different solutions are considered (see table 1). 3 BUILDING PHYSICS Algarve has a temperate climate, with an average temperature in winter of 14◦ C and in summer of 24◦ C, rarely going below 12◦ C during winter and reaching as high as 28◦ C or 30◦ C in July 529

Table 2. Global thermal analysis. Case [kW.h/m2 .year] Concrete Composite

Winter

Summer

Nic

Ni

Nvc

Nv

24.61 24.56

43.00 42.90

5.21 5.79

19.73 19.14

where Ni = maximum admissible value for heating and Ni = maximum admissible value for cooling.

and August. The thermal and acoustic behavior of both solutions are analyzed taking into account natural environmental conditions, and supposing that no other system for heating or cooling is installed. The structural elements (concrete and steel) and wall systems have different thermal mass and thermal and acoustic conductivity implications leading to distinctive behaviors. However the aim of this study was to have two alternative solutions with similar functional behavior, so that there would not be a “favored” solution. Therefore the amount of insulation needed in each case was determined in order to have two balanced solutions. The quantities of material insulation needed in each case will be taken into account in the life cycle analysis. 3.1 Thermal behavior According to the Portuguese regulation for building thermal behavior, the thermal behavior of a building is characterized by the following basic thermal indices: – Nominal needs of useful energy for heating (Nic ); – Nominal needs of useful energy for cooling (Nvc ). The global thermal analysis leads to the results of table 2, indicating that both construction solutions are in agreement with the prescribed thermal requirements, in a very similar way. 3.2 Acoustic behavior The Portuguese regulation that deals with the acoustic behavior of buildings defines two types of areas in terms of sound level of environmental noise: – sensitive area – defined in the land use planning system as suitable for residential purposes, existing or foreseen, as well as for schools, leisure and recreation spaces and others used by the locals for retirement, existing or foreseen; – mixed area – the areas, existing or foreseen, defined in the land use planning system as suitable for other purposes, that not the ones referred above, such as commerce and services. According to this regulation, the sensitive areas can not be exposed to a continuous equivalent sound level of external environmental noise, greater then 55 dB(A) at day time and 45 dB(A) at night. In the current analysis it was assumed that the building was implanted in such an area, so that it should meet the following requirement:

where D2m,n,w = normalized sound insulation of external walls. The analysis was made for each building façade, taking into consideration the existing openings in each one, and the results are indicated in the following tables. 530

Table 3. Double clay brick wall.

Table 4. OSB Panels, light gauge steel supports and plaster board.

Façade

D2m,n,w

Façade

D2m,n,w

West South East North

41.23 54.25 42.79 48.59

West South East North

33.82 33.17 33.86 33.30

Hence, all the façades fulfill the minimum admissible value.

Hence, all the façades fulfill the minimum admissible value.

4 LIFE CYCLE ASSESSMENT In this paper environmental performance is measured using the evolving, multi-disciplinary approach known as environmental life-cycle assessment (LCA), following the guidance on the International Standards Organization 14040 series of standards and the methodology and computer program developed by BEES (BEES, 2002). According to the ISO methodology, LCA involves four steps. The goal and scope definition step spells out the purpose of the study and its breath and depth. The inventory analysis step identifies and quantifies the environmental inputs and outputs (inventory flows) associated with a product over its entire life cycle. The impact assessment step characterizes the inventory flows in relation to a set of environmental impacts, and finally the interpretation step combines the environmental impacts in accordance with the goals of the LCA study. 4.1 Goals and scope of the study The goal of this study was to generate relative environmental performance scores for structural alternatives based on available average data. The scoping phase of any LCA involves defining the boundaries of the system understudy and determining which inventory flows are tracked for inbounds unit processes. 4.2 Inventory analysis and impact assessment Inventory analysis entails quantifying the inventory flows for a product system. Inventory flows include inputs of water, energy, raw materials, and releases to air, land and water. In this study manufacturing data for concrete was based in BEES products, which was collected by the Portland Cement Association LCA database (BEES, 2002). Data for steel was taken from IISI (IISI, 2002). In this case data covers all the production steps from raw materials acquisition to finished products ready to be shipped from the steelworks. However it does not include the manufacture of downstream products, their use, end of life and scrap recovery schemes. The impact assessment step of LCA quantifies the potential contribution of a product’s inventory flows to a range of environmental impacts. The LCA impact assessment method used in this study is the Environmental Problems approach developed within the Society for Environmental Toxicology and Chemistry (SETAC). It involves a two-step process: – Classification of inventory flows that contribute to specific environmental impacts; – Characterization of the potential contribution of each classified inventory flow to the corresponding environmental impact. This results in a set of indices, one for each impact, that are obtained by weighting each classified inventory flow by its relative contribution to the impact (e.g., the global warming potential index is derived by expressing each contributing inventory flow in terms of its equivalent amount of carbon dioxide). 531

Water Intake Smog Ozone Depletion Indoor Air Quality

120.000

Human Health

100.000

Habitat Alteration 80.000 Global Warming 60.000

Fossil Fuel Depletion

40.000

Eutrophication

20.000

Ecological Toxicity

0.000

Criteria Air Pollutants Concrete

Composite

Acidification

Figure 4. Environmental performance by alternative.

In this analysis, twelve impacts were considered: i) Global warming potential; ii) Acidification potential; iii) Eutrophication; iv) Fossil fuel depletion; v) Habitat; vi) Criteria air pollutants; vii) Human health; viii) Smog formation; ix) Ozone depletion; x) Ecological; xi) Water intake; xii) Indoor air quality. Every impact is evaluated by an index derived from the following expression:

where IAjk = characterized score for alternative j with respect to impact i; Iij = inventory flow quantity for alternative j with respect to impact i; IAfactor i = impact assessment characterization factor for inventory flow i; i = inventory flow quantity for alternative j with respect to impact i; and n = number of inventory flows in impact category k. The measures of each impact category performance are expressed in non-commensurate units. In order to perform the next step of LCA, interpretation, performance measures are placed on the same scale through normalization. The normalized data used in this case was developed by the U.S. EPA Office of Research and Development (BEES, 2002). Normalization is accomplished by dividing each impact by the fixed U.S.-scale impacts, yielding an impact category performance measure that has been placed in the context of all U.S. activity contributing to that impact. 4.3 Interpretation of results To compare the overall environmental performance, the performance scores, for all impact categories, may be synthesized by weighing each impact category by its relative importance to overall environmental performance and then computing the weighted average impact score. The weights sets used in this case study are based in a study from University of Harvard (BEES, 2002). The study developed separate assessments for the United States, The Netherlands, India, and Kenya. In addition, separate assessments were made for “current consequences” and “future consequences” in each country. For current consequences, more importance is placed on impacts of prime concern today. Future consequences place more importance on impacts that are expected to become significantly worse in the next 25 years. Sets of relative importance weights are derived for current and future consequences, and then combined by weighing future consequences as twice as important as current consequences. 532

Water Intake Smog Ozone Depletion Indoor Air Quality Human Health Habitat Alteration Global Warming

Composite solution Concrete solution

Fossil Fuel Depletion Eutrophication Ecological Toxicity Criteria Air Pollutants Acidification 0.000

5.000

10.000

15.000

20.000

25.000

30.000

Figure 5. Environmental performance by flow.

The graph in Fig. 4 represents final environmental performance results. According to the methodology described in the previous paragraphs, the final score is a normalized sum of each weighted environmental impact category score. In Fig. 5 the bar graph represents the environmental performance result by environmental flow. In this graph the most important impact flow categories are clearly indicated. From the previous graphs, the composite solution is better than the concrete solution, from the environmental point of view. However, emphasis should be given to the fact that the steel was assumed to be produced by the electric arc furnace (EAF) process. The EAF process uses about 95 percent recycled steel while the basic oxygen furnace (BOF) process uses 25 to 35 percent recycled steel. The energy intensity of BOF is about 26 GJ/ton of steel, while that of EAF is about 11.8 GJ/ton of steel (Chaturvedi and Ochsendorf, 2002). Carbon and other emissions are correspondingly lower for EAF process also making it better environmentally. 5 LIFE CYCLE COST In a life cycle cost analysis all costs arising from owning, operating, maintaining, and ultimately disposing of a project are taken into consideration. In this case study, all project-related costs occurring at the different stages over the life-time of the structure are discounted to their present value as of the base year (i.e. first year of study period), before being combined into the LCC estimate of each alternative. Equation (3) was used to convert future costs to present value and sum them into a single life-cycle cost member.

where Ct = sum of all costs incurred in year t, value in base-year euros; d = real discount rate for converting time t costs to time 0; and T = number of years in the study period. The compilation of initial construction costs is based on the quantities of materials needed in each alternative solution. The future costs are estimated upon the maintenance expenses expected to occur during the study period. In this case, the study period was 50 years and a real discount rate of 3.9% was used. The results obtained in each case are represented in the graph of Fig. 6. In this example the composite solution has a slightly better economic performance than the concrete solution. However it should be noticed that the future costs for the composite solution 533

35000.000 30000.000

Future Cost (€)

25000.000

First Cost (€)

20000.000 15000.000 10000.000 5000.000 0.000 Concrete

Composite

Figure 6. Economic performance.

60.00 50.00 Econ. Wt (50%) 40.00

Environ. Wt (50%)

30.00 20.00 10.00 0.00 Concrete

Composite

Figure 7. Overall performance.

are higher than the first case, and this is due to the maintenance requirements, which are more demanding in the case of the composite solution. 6 OVERALL PERFORMANCE The overall performance summarizes the environmental and economic performances into a single score. According to the methodology used in this study (BEES, 2002) the overall performance score for each alternative is derived from the following expression:

where Sj = overall performance score for alternative j; EnvW t = environmental performance weight; EconW t = economic performance weight; EnvSj = environmental performance score for alternative j; and LCC j = total life-cycle cost in present value euros for alternative j. The total score is a weighted average between environmental and economic performances, and the weights are freely chosen by the user, so that EnvW t + EconW t = 1. In this case it was chosen to have EnvW t = EconW t . The final result is represented in Fig. 7, giving clearly a better overall performance for the composite solution. 534

7 CONCLUSIONS In this paper a comparative analysis between a classic concrete constructive solution and a composite solution was performed, from a sustainable point of view. The final balanced score between environmental and economic performances showed that, in this case study, the composite solution had a better overall performance. The first aim of this work was to present and apply a methodology for a sustainable analysis between two alternatives, therefore the final result should not be taken as a general rule between these two design alternative construction methods. REFERENCES Chaturvedi, S. & Ochsendorf, J. 2004. Global Environmental Impacts due to Cement and Steel, Structural Engineering International, Volume 14, Number 3, pp. 198–200. International Iron and Steel Institute (IISI). 2002. World Steel Life Cycle Inventory – Methodology Report 1999/2000. Committee on Environmental Affairs. International Standard 14040. 1997. Environmental Management – LCA – Principles and Framework. International Standard 14041. 1998. Environmental Management – LCA – Goal and Scope Definition and Inventory Analysis. International Standard 14042, 2000. Environmental Management – LCA – Life Cycle Impact Assessment. U.S. National Institute of Standards and Technology. 2002. Building for Environmental and Economic Sustainability (BEES) 3.0, NISTIR 6916.

535

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Aluminium – a sustainable building material? C. Radlbeck & D. Kosteas Section for Light Metal Structures and Fatigue, Technische Universität München, Germany

M. Schlinz Department for Steel, Technische Universität München, Germany

ABSTRACT: Realizing the concept of sustainability in the building sector presents a challenge for architects, planners and engineers. Structures have to be functional and efficient during their lifetime while fulfilling economic, social, technical and ecological requirements. A significant factor in this process is the selection of the building material itself, which has to be of highquality while, durable, recyclable as well as cost and energy efficient. In these aspects aluminum has a high potential. To evaluate sustainable performance, the application of respective indicators is necessary. Thereby aspects of design, utilization, maintenance, disposal, recycling as well as interaction effects are to be taken into account. To put indicators into practice tools such as Life Cycle Cost Analysis (LCCA), Life CycleAssessment (LCA) procedures and appropriate design methods have to be used. A first step in this direction is presented by the ADT-Program combined with a LCA software tool enabling the analysis of correlations between design measures and ecological effects already in an early planning phase. The procedure is presented by simulating a façade.

1 INTRODUCTION Increasing demand for adequate and sustainable performance offers an opportunity for aluminum as building material since its advantages become even more evident considering the whole lifecycle of a structure. However, there is no common method yet, to evaluate the performance of structures or building materials in respect to sustainability. As there are many aspects to consider, some of them even linked with each other, it is obvious to apply a set of indicators. By that, it is made possible to accurately grasp the scope of impacts and to access the outcome of measures. Indicators can be realized by using tools such as Life Cycle Cost Analysis (LCCA), Life Cycle Assessment (LCA) and appropriate design methods. Thereby the latter plays a decisive role, as it presents the basis for any construction and influences the economical as well as the ecological performance over the whole life cycle. In general, the earlier indicators are applied and realized in the life cycle of a structure, the higher the impacts they have. As a consequence of that, architects, planners and engineers should bring to mind and optimize ecological, economical and social aspects during building activities. The objective is to combine LCCA, LCA and design by means of respective software utensils, make these available to the engineering community and thus facilitate optimal building structures already in an early planning phase. These tasks are undertaken within a new research project supported by the Bavarian Government. Before this background, the authors present indicators to help architects, planners and engineers evaluate their aluminum structures in respect to sustainability. Based on the indicators, an aluminum-glass-façade is modeled. Thereby several variations are simulated and evaluated. For this purpose the ADT program is used together with an LCA software tool. Thereby it is made possible to combine design features with an ecological data bank. With the results correlations 537

between technical measures and ecological effects are derived and thus optimal building structures are facilitated already in the planning and design phase. 2 FACTS ABOUT ALUMINUM 2.1 Production The production of one ton aluminum requires four tons of bauxite as base material. Bauxite is extracted by open cast mining, mainly in Australia, South America and Africa, defacing large land districts. After use, 80% of the mining are vegetated with the original vegetation and 20% are used for agriculture and forestry (Glimm 2001). Alumina is extracted from bauxite with caustic soda. Thereby slag is produced as secondary product (0,9t/tAL (Krone 2000)). Since no efficient method for further use has been developed yet, all the slag has to be disposed. In contrast to that, slag accumulated during the steel production is further used in the building industry, e.g. as sand for cement. Primary aluminium is finally produced by separating oxygen from alumina by electrolysis. By additional forming or rolling semi-finished products such as extruded profiles and rolled products, are manufactured. The production of aluminium consumes 174 GJ/t which is equal to 48,33 MWh/tAL primary energy1 . Thereof the electrolysis needs approximately 74,7% (36,12 MWh/tAL ) (Krone 2000). The input of grey energy2 for the aluminium production accounts for 141 GJ/tAL (EAA 2000). The terminative energies3 are composed of 15,5 MWhel /tAL for electricity and 15,33 MWhBS /tAL for combustibles. Most electricity thereof (15 MWhel /tAL ) is consumed by the electrolysis. However, this value is dependent on the respective technology and varies worldwide between 12,9 and 18,3 MWhel /tAL (Martens 2002). Values for electricity are based on power productions in the Western World. Thereby the degree of efficiency is estimated as 53,7%. The latter value varies in dependence of the quantity of hydraulic power. Worldwide, 61,3% (Krone 2000) of the electricity, used for electrolysis processes to extract aluminium, is produced by hydraulic power. The production of semi-finished products consumes 17,9 GJ/tProfile of grey energy for extruded profiles and 9 GJ/tRolled product for rolled products (EAA 2000). 2.2 Recycling The energy-intensive production can be compensated by the excellent recycling characteristics of aluminium. In general, aluminium can be recycled without loss of quality and as often as necessary. In contrast to the production the recycling process only consumes 20 GJ/tAl primary energy. Thereby 0,47 MWh/tAl electricity is needed which is 97% less than for the production (Krone 2000). The estimation of the grey energy demand for recycling has to include the disassembly of the aluminium structure. Thereby, influencing variables such as the size of the component parts, composites, joining techniques, alloys and coatings have to be taken into account. e.g. large components, like façades and windows, can be collected and remelted directly. The grey energy consumption for these processes is equal to 6 MJ/t recycled metal whereby transport is included. Small components such as foils, knobs etc., have to be collected together with mixed scrap. The remelting process for mixed scrap, transport included, consumes 10 MJ/kg recycled metal and therefore is clearly higher than for pure scrap. Besides small components easily get lost in the building rubble. The separation of composites might also lead to material losses. This is also the case during the remelting of high alloyed scrap and components with fine coverings. Non-magnetic coverings, such as zinc-coatings, can not be split by magnetic separators but have to be treated with 1 Primary energy: amount of renewable and non-renewable energy. 2 Grey energy: fossil energy (non-renewable). 3 Terminative energy: dependent only on the process and therefore can not be influenced by the way of energy

allocation.

538

swim-sink-procedures. This results in 15% material losses and 10% quality losses. The recycling of aluminum with coatings, consisting of bitumen, produces a significant amount of hazardous waste which has to be taken into account. 2.3 Aluminum as building material The advantages of aluminum as building material are not known sufficiently, not surprising in view of the lack of respective educational courses and tools until recently. Combined with high energy consumption and production cost in the beginning of the material life cycle and little general experience on the side of planners this has led to frequent prejudice and restrictive policies over the past years. Indeed, aluminum structures as primary load carrying as well as secondary or decorative elements offer excellent sustainable properties. As main advantage, there is to mention its lightweight (1/3 steel) together with high strength values. According to the type of alloy, strength values within the range of steel are yielded. Other advantages are presented by its functionality because of various alloys and extruded profiles, its natural corrosion resistance, formability, workability and flexibility. In addition aluminum is non-toxic, does not burn, provides high reflecting power and excellent conductivity of heat and electricity. The benefits of aluminum get even more obvious by considering the whole life cycle of the product. Thereby durability, low if any maintenance cost and no loss of quality during recycling (see 1.2) are the main characteristics. However, for the effective application of aluminium also the problem areas, such as susceptibility to deformations, the reduction of strength as a consequence of welding, and the risk of thermal bridges, have to be considered. In addition the high energy consumption during production and the higher starting costs in respect to other materials have to be compensated by an efficient use phase and an optimized deconstruction of the structure. By that it is possible to put the unfavourable starting positions into perspective. As a consequence of that, aluminium might even be a better choice than other building materials considering the whole life cycle. 2.4 Building applications According to a recent survey in Austria (Lins 2003), aluminum building applications are used ‘frequently’ by 36% of the architects and ‘sometimes’ by 27% of the architects. Thereby aluminum is applied commonly in the building envelope, e.g. for windows, façades and roofing systems, but also in portals, prefabricated structures, doors, winter gardens and small parts such as fittings, heating/air-conditioning systems, solar elements, knobs, etc. Finally, aluminum is also a relevant material in rehabilitation and renovation of existing structures. Especially in the building envelope, aluminum is often combined with glass. Extruded profiles, providing various possibilities to design terminal strips, together with significant developments in respect to point fittings make aluminum a very attractive material to combine with glass panels. The light weight of aluminum underlines the transparency of glass so that at the same time high demands of modern architecture can be fulfilled. In addition, the excellent thermal characteristics of low energy glass compensate the high thermal conductivity from aluminum. 3 INDICATORS 3.1 General definition Indicators increasingly continue to play a crucial role in the corporate development towards sustainability. The challenge is to develop commonly accepted, harmonized and practicable systems, which enable comparisons between different structures. By that it is made possible to accurately grasp the scope of impacts and to access the outcome of applied measures. In respect to building structures the following indicators (SIA 2000) have been developed: – To cover economic aspects indicators include the minimization of life-cycle-costs whereby the ratio of effort to profit has to be optimized. This involves great demands on durability and 539

functionality. Especially the latter becomes important in the area of reconstruction, renovation or changes in utilization. – Relating to ecology, indicators comprise as a first objective the closed life-cycle for materials to protect resources and minimize waste. Also included are the reduction of (CO− 2 ) emissions, land, water and energy consumption. – As for the social area, the combination of safety aspects with aesthetic demands by appropriate design has to be mentioned. In addition, structures are to ensure a comfortable room climate, optimize acoustic design and contribute to healthy living conditions. The availability and transfer of data and knowledge presents a final social indicator.

Table 1. Sustainability indicators vs. characteristics of aluminum as building material. Indicator Economy Durability

Life-CycleCosts

Maintenance of capital

Functionality

Ecology Waste Emissions Resources Land Water Energy

Social issues Safety

Aesthetics

Room climate Health/Comfort Acoustics Know-howavailability

Contribution

Tool

Problem

Light weight, various alloys, strength, low maintenance, corrosion resistance, non-combustible, guidelines Light weight, various alloys, functionality, workability, strength, low maintenance, corrosion resistance, non-combustible, non toxic, easy surface treatment, recycling, high scrap value Strength, low maintenance, corrosion resistance, non-combustible, non-toxic, easy surface treatment, recycling, high scrap value Light weight, various alloys, functionality, workability, high strength, corrosion resistance, non-combustible, easy surface treatment, coloring

Design

–

Design, LCCA

Initial costs

Design, LCCA

Initial costs

Design

–

Functionality, recycling, high scrap value Light weight, workability, low maintenance, non toxic Recycling, high scrap value Light weight, workability – Light weight, functionality, workability, low maintenance, corrosion resistance, recycling

Design, LCA LCA

– Production

LCA LCA LCA Design, LCA

Production Production Production Production, conductivity

Light weight, various alloys, functionality, workability, strength, non-combustible, corrosion resistance, easy surface treatment, guidelines Brightness, appearance, light weight, functionality, workability, corrosion resistance, easy surface treatment, coloring Reflectivity, non-toxic, guidelines Reflectivity, non-toxic Existent guidelines Existent guidelines

Design

Deformation, heat affected zone

Design

–

Design Design Design Publishing, workshops

Conductivity Conductivity Reflectivity Limited number of experts

540

3.2 Indicators for aluminum For investigating the sustainable performance of aluminium structures, their characteristics have to be checked with the above described indicators. Thereby it has to be differentiated between contribution possibilities, respective tools for realization and problem areas, Table 1. Except for water consumption, aluminium structures provide positive contributions to all defined indicators. The economic indicator ‘life-cycle-costs’ and the social indicator ‘safety’ yield the majority of contributions. They are followed by the indicators ‘durability’, ‘maintenance of capital’, ‘functionality’ and ‘aesthetics’. However, also some problem areas are derived. These are mainly ‘production’ and ‘initial costs’ followed by ‘heat conductivity’, ‘reflectivity’, ‘deformations’, ‘heat affected zone’ and ‘limited number of experts’. High energy consumption during ‘production’ and thus high ‘initial material costs’, can not be influenced by planners but have to be considered and compensated by appropriate design and optimized recycling possibilities. Other problem areas such as ‘heat conductivity’, ‘reflectivity’, ‘deformations’ and ‘heat affected zone’ are extensively treated in respective guidelines such as the European Norm EN 1999 (CEN 2004). In respect to ‘limited number of experts’ the research institutes and the aluminum industry are called upon intensifying data exchange and information politics. In general it can be stated that design is the basis for realizing nearly all the indicators and hence is the most important tool to realize indicators and thus sustainable building structures. 4 ‘BAYFORREST’ RESEARCH PROJECT 4.1 Objective In order to explore the influence of design measures on the building performance in a quantitative way, the Bavarian Government initiated the research project ‘Material Flows of Buildings’. Thereby correlations between design, economical and ecological aspects are investigated and evaluated. The objective is to derive practice-oriented data and procedures enabling architects and engineers to realize the concept of sustainability. Within this project also investigations of aluminium and aluminium-glass-structures are included. 4.2 Architectural-Desktop-Software tool The basis for these investigations is presented by an Architectural-Desktop-Software tool (BayForrest 2003) connecting design data with LCCA- and LCA-data, Figure 1. The software is based on the CAD-program Architectural Desktop from AutoDesk and is connected via internet to an ecological data bank called Gemis (Institute for Applied Ecology 2003),

Data of Aluminium/Glass Structures Ecology Databanks

Economy Companies

Implementing

Design LME

Programming

Software LCCA & LCA & Design Program (ADT) Modelling

Simulations

Examples, Scenarios, Break-Even-Analysis Analysis

Evaluation

Correlation, Optimization

Figure 1. Sustainable design of building structures with ADT software.

541

100 kg PA →

80 kg SA (100% PA) 20 kg Loss

EL: 20 kg PA

100 kg primary aluminium (PA) are used for a product. After remelting, 80 kg secondary aluminium (SA) replace 80 kg PA. The environmental loads (EL) have to be calculated for 20 kg PA and for the recycling process.

Figure 2. Substitution Method according to ISO 14040 (EAA 2004). 100 kg PA →

80 kg SA (90% PA) 20 kg Loss

EL: 20 kg + 0,1*80 kg = 28 kg PA

100 kg primary aluminium (PA) are used for a product. The output are 80 kg secondary aluminium (SA) with 90% quality of primary aluminium. Consequently, 80 kg SA replace 80 kg  0, 9 = 72 kg PA. The environmental loads (EL) have to be calculated for 28 kg PA and for the recycling process.

Figure 3. Value-Corrected Substitution Method according to ISO 14040 (EAA 2004).

including life cycle inventory data for respective materials, processes and scenarios. Thereby Germany presents the reference area. For the purpose of impact assessment the methods KEA/KSA, CML, ECO 95 and UBP are implemented. The KEA defines the cumulated energy demand of products/processes whereby it is differentiated between the different energy sources, e.g. renewable and non-renewable. The same is done in respect to the cumulated material demand (KSA). The CML method does not evaluate all single substances but combines them in categories according to their effect. This is done by equivalence factors. The categories themselves are not weighted among each other. The same procedure is implemented in the ECO 95 method. However here, the categories are weighted and aggregated to one ‘Ecoindicator’. Also the UBP calculates one indicator which is based on the principle of ecological scarcity. Thereby the ratio of current environmental loads to critical environmental loads is calculated. Within the research project, as a first step structures and buildings are modelled in 3D. Then materials are assigned to the respective elements. After that, the LCA-tool is applied. Thereby life cycle inventory data are calculated, evaluated according to the impact assessment methods mentioned above, and optimized. The same can be done in respect to LCCA data, although they are not implemented in the software yet. A further task will be to arise economical data of aluminium structures, integrate them in the program and do simulations considering all three areas of sustainability, economy, ecology and social affairs. 4.3 Life Cycle Analysis of aluminum structures Life Cycle Analyses in the past often resulted in ecological-based decisions restricting the aluminium market. Reasons for that were questionable data quality, non-consideration of hydraulic power and non-consideration of recycling credits. In consequence, the European Aluminum Association (EAA) performed in 1996 an extensive survey, collecting input and output data of environmental relevance from the aluminum industry. Since that time average values have been presented in annual environmental profile reports (EAA 2000), forming a reliable basis for LCA studies of aluminum structures. In addition, it is strongly recommended to base any Life Cycle Assessment on DIN EN ISO 14040: 1997-08 (CEN 1997). According to those guidelines the non-consideration of recycling is only permitted if either nothing or a negligible amount of scrap is recycled or the secondary material has no value. Consequently, recycling is taken into account for closed loops with the ‘Substitution Method’ (Figure 2) and for open loops4 with the ‘Value-Corrected Substitution Method’ (Figure 3). 4 Open loop: change of material properties after recycling either as a consequence of metallic contaminants or to

modify chemical composition.

542

Figure 4. Original state of ‘Alter Hof ’.

Figure 5. Future design.

Figure 6. Model.

5000 4124 4000 2914

3000

2073

2000

1617

1000 0

Aluminum (Gemis)

Aluminum (EAA)

Steel

Zinc-coated steel

Figure 7. KEA [GJ] for the construction of several façade types with consideration of glass panels.

5 EXAMPLE: FAÇADE 5.1 Models, simulations In the course of the research project, the building ‘Alter Hof ’ in Munich, Germany (Figure 4) was made available as practical example for the project teams. The historical building will be completely renovated. Thereby, also a transparent stick façade (Figure 5) is planned, which is the basis for the following simulations. The façade is first modeled with the load carrying system consisting of steel and the terminal strips made from aluminum (Figure 6). A further model is based on the assumption, that zinc-coated steel is used for the structure. In a third model, aluminum is assigned to the load carrying structure as well as to the terminal strips. Since the data for aluminum included in the Gemis data bank differ significantly from the data provided by the European Aluminum Association, simulations based on the EAA values are carried out within an additional model. In respect to steel, the Gemis data rather comply with the values provided by the steel industry. Therefore, a data modification of steel is not necessary. Finally not only construction, but the whole life cycle of the façade is considered. Since glass panels are present in all façade types in the same way, they are not included in the life-cycle orientated simulations. To this purpose the substitution method according to ISO 14040 is applied. Thereby for aluminum as well as for steel a recycling rate of 85% is defined. For the recycling of aluminum a loss of 20% (Figure 3) and for steel a loss of 28% (Figure 4) is assumed. Since steel has to be provided with a corrosion protection, it is obvious to calculate respective losses of material and quality. In contrast to that aluminum is used without any coating and therefore can be recycled directly. No renewal of the steel coating is calculated during use. 5.2 Results The models are evaluated in accordance with the KEA method. The results for construction are presented in Figure 7. The façade consisting of aluminium and glass, calculated with the Gemis 543

1500 1106 1000 704 500

0

Aluminum (EAA, EN ISO 14040)

Steel (EN ISO 14040)

Figure 8. KEA [GJ] for the whole life cycle of two façade types without consideration of glass panels.

data, clearly consumes the most energy. The simulation of the aluminium-glass model based on the EAA data results in lower values but still consumes more energy than the façades using steel for the load carrying structure. Thereby, the results for the model using steel without coating are the lowest. However, it might not be realistic to use steel without any coating since it is quite susceptible to corrosion. As a consequence of that the façade with zinc-coated steel elements presents a more realistic model. The results for the whole life cycle based on ISO 14040 are presented in Figure 8. By considering the whole lifetime together with the allocation of recycling credits, the aluminium-glass façade turns out to be more favourable in respect to energy consumption than the façade using steel for the load carrying structure. Although the assumptions in respect to steel, e.g. recycling rate, loss of quality are based on optimistic scenarios.

6 CONCLUSIONS Aluminium has high potential to perform as sustainable building material. Because of its characteristics many contributions are made to sustainability indicators. Problem areas are compensated by applying respective tools, such as appropriate design, LCA and LCCA. Simulations with the ADT program, combining building properties with ecological data, enable correlations between design and ecological performance. By that it is possible to evaluate the applicability of respective materials and to develop optimized design solutions. The simulations also show that it is absolutely necessary to evaluate structures in respect to their whole life-cycle. In addition recycling credits have to be allocated according to ISO 14040. On this basis the energy intensive production of aluminum can be balanced. Compared to steel, even less energy consumption of aluminum structures over the whole life cycle is observed. However, further investigations are undertaken. Thereby, values and processes implemented in other data banks will be used for simulations. In addition further impact assessment methods will be applied. In addition also other structures such as for example roofs will be examined. These will be further tasks of the involvement in this project of our Section for Light Metal Structures and Fatigue.

ACKNOWLEDGEMENT We wish to acknowledge the financial support of the Bavarian State Ministry of the Environment, Public Health and Consumer Protection in this research project. 544

REFERENCES Bavarian Research Network for Recycling and Waste Management (BayFORREST). 2003. Internet basierte Simulation des Resourcenbedarfs von Bauwerken. Lehrstuhl für Bauinformatik. Research Project F219.Technische Universität München, Lehrstuhl für Bauinformatik. DIN EN ISO 14040: 1997-08. Environmental management – Life cycle assessment – Principles and Framework. European Aluminium Association (EAA). 04/2004. Building, www.eaa.net. Brüssel: EAA. European Aluminium Association (EAA). 04/2004. Key features how to treat aluminium in LCAs, with special regard to recycling issues. www.eaa.net, Brüssel: EAA. European Aluminium Association (EAA). 2000. Environmental Profile Report for the European Aluminium Industry. Brüssel: EAA. Glimm S. 2003. Rotschlamm und Rotschlammdeponien, Publication of the German Association of Aluminium Industry (GDA). Düsseldorf: GDA. Institute for Applied Ecology 2003. Global Emission Model for Integrated Systems (GEMIS), Software tool for LCA, developed 1987–89, current version 4.13. Berlin: Institute for applied Ecology. Krone K. 2000. Aluminiumrecycling – vom Vorstoff bis zur fertigen Legierung. Düsseldorf: Vereinigung Deutscher Schmelzhütten (VDS). Lins S. 2003. Architekten und Bauherrenstudie. Wien: Aluminium-Fenster- Institut. Martens P.N. 2002. Stoffstromanalyse und Nachhaltigkeit – am Beispiel Aluminium, International Journal for Metallurgy Special Edition., 56. Volume, 12/2002. Frankfurt/Main: Industriegewerkschaft Metall prEN 1999-1-1: 2004: Design of Aluminium structures, Part 1-1: General structural rules. Radlbeck C. & Kosteas D. 2004. Calibration Study for the Eurocode 9. Proceedings 8th INALCO Conference 2nd –3rd June. Cleveland: INALCO conference. Suisse Association of Architects and Engineers (SIA) 2000. Kriterien für nachhaltige Bauten. SIA Documentation. Zürich: ETH.

545

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Sustainability by LCCA of aluminium structures E. Dienes, C. Radlbeck & D. Kosteas Technische Universität München, Munich, Germany

ABSTRACT: Life Cycle Cost Analysis (LCCA) is one of the tools which is applied to realize sustainable development. With LCCA in an early phase of the building process, appropriate materials, design and methods can be chosen to assure economic, social and environmental benefits over the whole life-cycle. LCCA of aluminium structures point out the economical potential of aluminium as building material. Although primary aluminium has higher production cost than other construction materials it provides many economic benefits and might be even more efficient than other materials over the whole lifetime of a structure. Furthermore, great economic and environmental benefits arise from using secondary aluminium over several life-cycles. An LCCA was made exemplarily for an aluminum pedestrian bridge to investigate the amount and distribution of expenditures and revenues in the different life-cycle phases such as design/ construction, utilization/maintenance and demolition/disposal. Already the first results show that aluminum as construction material results in cost-efficient structures with high functional quality.

1 INTRODUCTION The institutional foundation for sustainable development was laid in the United Nations Conference on Environment and Development in Rio de Janeiro in 1992. “Agenda 21”, one of the resulting comprehensive action plans obliged the 178 adopting governments and major organizations to develop strategies and take action locally for the equation and integration of economic efficiency, environmental protection and social responsibility. The European Union‘s “6th Environmental Action Programme” came into force in July 2002 and contains legally binding ecological commitments and obligations (Hontelez 2002). Within this strategy “Sustainable Use and Management of Natural Resources and Wastes” have to be carried out until 2012. This incorporates the estimation of material and waste streams in the Community, the further development of waste legislation – including construction and demolition waste – until 2006, as well as the significant overall reduction of waste (volumes generated) and disposal. In addition, a wide range of tools promoting eco-efficient production methods, technologies and the sustainable use of resources are to be developed and applied. Building activities currently account for 40% of the total European man-made waste accumulation (CIB 1999), 40% of the total European energy use (European Commission 2001) and 30% of CO2 -outcasts (Töpfer 1997). In addition, the construction sector is the Community’s largest industrial sector, contributing with approx. 11% to the GNP and having more than 25 million people directly and indirectly engaged (CIB 1999). Owing to the future responsibility placed upon the industry by environmental policy, there is an essential need in the construction sector for a paradigm shift towards the design and marketing of sustainability in terms of resource-efficient, non-toxic, recyclable and functional building structures. As a first prerequisite, appropriate methods and tools have to be applied for the evaluation of parameters such as economic efficiency, environmental friendliness and fulfilment of social requirements. Thereby profitability is a frequently used indicator for business efficiency and can be expressed as ratio of profit divided by the amount invested. 547

High profitability can be reached by maximizing earnings or minimizing costs. Considering all costs is in particular important for capital intensive and long-time investments, such as buildings. During long utilization periods building materials may adversely interact with the environment (e.g. moisture) or the purpose of utilization may change, resulting in increased maintenance costs and a deterioration of the total sustainability performance. Economic evaluations hence must take into account costs arising from all life-cycle phases of a structure. In the following, aluminium structures are investigated in respect to cost-efficiency and economic benefits. For this purpose the method of life cycle cost analysis (LCCA) is introduced and a calculation example of a pedestrian bridge is presented.

2 SUSTAINABLE USE OF ALUMINIUM Aluminium is a very attractive and flexible material that is used in various applications in almost all industries. Main user is the transportation sector, consuming 32% of all manufactured aluminum end-products. This is closely followed by the construction industry accounting for 26%. Especially in the latter sector, aluminum demand has significantly increased over the past few years. The construction sector even presents the main consumer in respect to extruded profiles (51%) and rolled products (11%) (EAA 2001). The characteristics of aluminum, such as durability, flexibility, formability, functionality, lightweight, recyclability, non-toxicity, natural resistance against corrosion, non-combustibility and low maintenance during service life present significant potentials in respect to sustainable applications. Properties can be adjusted to specific application requirements by selecting the appropriate alloy type. In addition, aluminum can be remelted indefinitely, saving at each cycle 95 per cent of the fossil energy compared to the production process. Moreover, aluminum is the most costeffective material to recycle (WasteOnline 2004). Thus, recycling can substantially contribute to the conservation of natural and financial resources. European recycling rates vary in different branches between 41% (beverage cans) and 95% (transportation), with actual 85% in building and construction (EAA 2003). This situation may be improved by developing appropriate logistic systems.

3 LIFE CYCLE COST ANALYSIS (LCCA) Life Cycle Cost Analysis (LCCA) is an appropriate method to investigate all costs arising “from cradle to grave” of a structure and identify major sources of inefficiencies and cost reduction potentials. According to the model of Fabrycky and Blanchard, LCCA may be used as a feedback tool through various stages of the product life-cycle to evaluate compliance of alternative solutions with pre-defined targets, and indicate areas of non-compliance where corrective action is necessary (Blanchard & Fabrycky 1998). Figure 1 shows the typical life-cylce phases of a structure and composition of Life Cycle Costs. In the mobile industry the application of LCCA studies is already common practice. As a consequence of that, especially in vehicle construction, the cost-cutting advantages of using aluminum mainly based on its lightweight are highly estimated. e.g. 0.8 l fuel per 100 km is spared with each 100 kg loss in weight. Obviously, savings in energy consumption and costs during utilization compensate the energy-intensive production and the higher starting price of aluminum. In contrast to that, investment decisions in the construction sector are still based on the amount of acquisition cost and on the warranty conditions. According to a representative survey in Bavaria/Germany, life cycle costs are only considered as authoritative of approx. 25% respondents (Hasselbeck et al. 2004). This clearly indicates the low relevance of life-cycle orientation in construction practice, presumably because of missing knowledge about its benefits and the unavailability of tools for realization. 548

Design, Calculation, Development

“Design Costs“ “Costs of Demolition & Disposal”

Production, Transport, Assembly

“Construction Costs”

Life Cycle Costs

Demolition/ Dismantling, Disposal/ Recycling

“Operation Costs”

Utilization, Maintenance, Repair

Figure 1. Different Life Cycle Phases of a building structure, and composition of Life Cycle Costs.

3.1 Classification of life cycle costs 3.1.1 “Design Costs” “Design Costs” comprise all expenditures for design, calculation, planning, research and development. Thereby costs of contracting architects and engineers, as well as costs for scientific experiments, market analyses and feasibility studies are included. In general, approx. 60 per cent of the life-cycle costs of structures are determined by their design (Herzog 2003, Blanchard & Fabrycky 1981). As a consequence of that, all important factors influencing conditions of usage, demolition etc. (e.g. weather, planned service life, ownership, social preferences) have to be taken into account. For planning investments such as structures, the LCCA-approach of (Woodward 1997) can be applied. Thereby, the following steps are included: 1. 2. 3. 4. 5. 6.

establishment of the operation profile; identification of all cost elements; determination of critical cost parameters; calculation of all costs at current prices; escalating prices at assumed inflation rates; discounting costs to the base period, then summing them up to get the net present value.

3.1.2 “Construction Costs” “Construction Costs” include all expenditures for completing construction works, including the costs of raw materials, surface treatment, joints, transport as well as labor for production and assembly. As a basic principle, costs of the production facility, storage and insurance belong to this category as well. However, if it is not possible to assign those costs accordingly to specific products, they will be added to the general business expenses. In general it can be stated that fully assembled aluminium structures cost 1.2 to 2 times more than steel counterparts. Aluminium structures can compete with cheaper materials mainly because of their extreme light weight, and thus resulting cost savings for raw material (per weight), labor (e.g. for assembly), transport, and insurance. The introduction of EUROCODE 9 “Design of aluminum structures” in the near future, may further increase the utilization of aluminum properties in terms of efficient design and thus result in additional reductions of weight and raw material costs. Furthermore, complex cross sections can be fabricated by the extrusion process. Various joint types can be easily realized due to the workability and weldability of aluminum. Costs of surface treatment are not relevant for aluminium because of its natural resistance against corrosion. However, exceptions are presented by extreme environmental conditions, by direct contact with 549

certain materials, e.g. steel, concrete etc., and by the requirement of special coloring effects. Indirect costs of interfering with public activities (e.g. redirections for traffic, interrupting office hours) during construction can be limited by quick erection of aluminum structures enabled again by its lightweight. In urban areas with high traffic volumes, economic costs of delays and accidents caused by traffic blocks and detours may amount up to 100,000 USD per day (Ahlskog 1996). 3.1.3 “Operation Costs” “Operation Costs” cover all expenditures for inspection, maintenance and repairs as well as running costs e.g. of heating, taxes and supervision or monitoring. Because of its neutral behavior against environmental impacts, the use of aluminium can significantly reduce costs of inspection and maintenance compared to e.g. steel or wood, whose corrosion protection must be renewed periodically. In a recent study approx. 90% of interviewed German companies offering aluminum products for building applications confirmed, that inspection as well as repair are either never or rarely needed to reach the designated life-time of their products (Hasselbeck et al. 2004). Reduction of operation costs may be also achieved by innovative technical solutions and efficient design. An example for that presents the aluminum facade developed by Hydro reducing the energy need for air-conditioning and thus costs for electricity, up to 60 per cent in office buildings (Hydro 2004). 3.1.4 “Costs of demolition and disposal” At the end of service life, structures must be deconstructed. This is followed by disposal or recycling of the building materials. Because of the low density of aluminium transport costs can be reduced in contrast to demolition of heavier materials. Structures designed for assembly (DFA) are easy to dismantle and single modules can be reused or recycled. Aluminium scrap represents a high market value so that disposal of used aluminium is not relevant. By selling aluminium scrap a certain part of the initial capital invested into the structure can be regained. By participating in appropriate recycling systems – e.g. the German A/U/F (“Aluminium and Environment in the Construction of Windows and Facades”) – disposal is avoided, the amount of secondary aluminum is increased and savings in raw material costs are reached in an environmental friendly way. 3.2 Calculation of capital expenditure In line with the methodology of Woodward (see 3.1.1) and the definitions of chapter 3.1 for different cost types, capital expenditure can be calculated using the net present value method by summing up all anticipated expenditures and revenues over the total life-cycle, considering also influence of interest and inflation rates:

where C = capital expenditure for the whole life-cycle at present or at time of the first investment; aAC = acquisition cost (= “design costs” + “construction costs”); aOt = operation cost or revenue in year t; aD = “costs of demolition and disposal” (including also revenues from sales of materials to be recyled/ reused); n = total number of service years; t = respective year of operation; i where i = inflation rate (considered as constant); g = 100 p q = 100 where p = interest rate (considered as constant). For calculating the capital expenditure, the algebraic sign of costs often is chosen as positive while revenues have a negative sign for structures yielding no or hardly ascertainable income (e.g. bridges for public use). If tolls are established, the use of algebraic signs may be changed. During long lifetime of building structures interest and inflation rates usually vary. Consequently, i and p are either estimated as constant, or Equation 1 is changed in order to consider different values of the 550

Costs [€] 220.000 2.71

Wood Aluminium Steel

2.77

120.000 1.52

20.000 0

10

20

30 Years

40

50

60

Figure 2. Life Cycle Costs of a 30 m-pedestrian bridge with three different building materials (Kosteas & Meyer-Sternberg 2000, Bentrup 2000).

respective periods. The estimation of inflation rate is often limited in terms of narrow margins, whereas choosing an appropriate interest rate presents a rather difficult venture. Within Europe, the Euro Interbank Offered Rate1 provides a respective starting basis. Yet unknown costs and revenues can be estimated by increasing current prices with assumed inflation rates. Application limits of assumptions are defined by sensitivity analysis. 4 LCCA OF PEDESTRIAN BRIDGES LCCA results for 2 pedestrian bridges are given in Figures 2–3. The first bridge has a 30 m span, 3 m width, is designed for 4.0 kN/m2 and a service life of 60 years. Three alternatives with different building materials are compared: aluminium, steel and wood. Thereby costs of acquisition and operation are considered. Indirect costs caused by construction works interfering with public activities (traffic blocks, delays etc.) are neglected. Acquisition costs of the aluminium and steel bridges are calculated with tonnage prices including also the costs of assembly and surface coating (Kosteas & Meyer-Sternberg 2000). Cost of the superstructure for open wooden bridges with simple design are derived from (Werner 1986), cost of the substructures and the bearing are aligned to the tonnage of the superstructure. According to DIN 1076 (DIN 1999), operation costs of a bridge incorporate inspection costs and all expenditures necessary to guarantee the designated service life. For aluminium and steel bridges, according to DIN 1076 every 3 years either a main or simple inspection is required, whereby the type of inspection is alternating. For wooden bridges, main inspections have to be carried out every 6 years. In addition, repairs of the anti-corrosion layer of steel are based on specific time intervals. Because of the normal environmental conditions no maintenance costs are considered for aluminium (Kosteas & Meyer-Sternberg 2000). The periodic expenditures for inspection and renewal of the preservation coating for wood are based again on (Werner 1986). The inflation rate is estimated as 2%, interest rate as 5,2% both being constants. To simplify the calculation it is assumed that costs/revenues of a respective year have to be paid/received at year-end. Steps to determine the life-cycle costs are: 1. Escalate current prices for maintenance and inspection (operation) aO with i = 2% inflation rate to the respective year of payment (t):

1 Rate

at which euro interbank term deposits are offered by one prime bank to another. www.euribor.org

551

Costs [€] 40.000 Aluminium bridge 2

1.40

35.000

30.000

25.000 0

6

12

18

24

30

Years

Figure 3. Life Cycle Costs of a 8.64 m-pedestrian bridge made of aluminium.

2. Escalate current price of demolition and disposal aD0 with i = 2% inflation rate to the year of demolition (n):

3. Discount escalated costs aOt and aD0 using i = 2% inflation and p = 5,2% interest rate to the present and sum them up, together with the acquisition cost according to Equation 1. The cost of the aluminium footbridge is given by a quasi monotonously and shallow progressing line due to regular low costs of inspection. High expenses for repair or total renewal of the anticorrosion coating for steel (after 15, 25, 45 and 60 years) and wood (every 7 years) as well as for yearly inspection of the wooden footbridge are demonstrated by the increments in their respective cost curves. The break-even point (time period, beyond that accrued costs for an alternative are lower than for the others) of the aluminum bridge is reached after 25 years which is clearly below the required service life of 60 years. By the end of the service time total costs of the aluminum bridge amount to 1.52 times the value of the acquisition cost. Respective figures for steel and wood are much higher – 2.77 and 2.71. In structures where advantages with aluminum are utilized maximally, as in buildings with special needs, in special environments and erection provisions or with moveable elements, a lower break-even point of approximately 10 years can be achieved. In Figure 3 life cycle costs of another aluminum bridge with 8.64 m span, 2.60 m width and with 5.0 kN/m2 design loads are presented. In this case, the same types of costs are considered as in the example above. However, in addition here also costs of demolition and transport, revenues from selling old scrap to a metal merchant as well as regular costs for maintenance are taken into account. Compared to Figure 2, a similarly low and continuous slope of operation costs can be observed. In total, life cycle costs increase until a factor of 1.4 compared to the initial costs of the bridge. 5 CONCLUDING REMARKS Recent developments and reformations of laws in respect to sustainability, make it necessary also for the construction sector to meet respective requirements for new structures. Thereby, the economic dimension of sustainability involves the application of Life Cycle Cost Analysis. The objectives are to optimize structures over the whole life-cycle and to achieve required economical targets. Moreover, LCCA presents a useful tool to help determine break-even points of different product alternatives and to select the most cost-effective solution. 552

Based on new legal directives, also authorities might review and change their financial system according to sustainability. In the past, running separate budgets for acquisition (design and construction) and operation – as it often happens in the public sector representing major clients for the construction industry – resulted in the effort of construction companies to minimize acquisition costs. A possible life-cycle based orientation of authorities in the future will entail bidding companies to indicate sustainable performance by presenting LCCA. This is already common practice in the USA. There, participants of tenders with big construction volume are already obliged to pursue LCCA by law. Furthermore, the American “National Institute of Standards and Technology” (NIST) has developed a special calculation program for the evaluation of cost-effectiveness for bridges called “Bridge LCC”. Another development, involving LCCA as necessary tool, is presented by the increasing popularity of “Build-Operate-Transfer” contracts. Thereby the construction company (or main contractor) is obliged to operate the structure after finishing construction. By that, operation costs are optimized and economic efficiency over the operation time is intended. This entails sustainable design and high-quality building materials. In this discussion, aluminium has a high potential as a building material. Properties like light weight, low maintenance, corrosion resistance and recycling make aluminum structures cost and energy efficient, especially during service. The application of LCCA showed already, that in spite of high energy and cost consumption during production, aluminium will frequently prove to be more favourable than other materials for the total design life, often even within early phases of the life cycle. REFERENCES Ahlskog, J.J. 1996. Aluminium Bridge Decks – A Viable Economic Perspective. Paper # IBC-96-12. Association of German Producers of Aluminium Windows and Facades (A/U/F). 1998. Ganzheitliche Bilanzierung von Fenstern und Fassaden, Frankfurt: Verband der Fenster- und Fassadenhersteller (VFF) e.V. Bentrup, R. 2000. Marketingkonzept im Industriegütermarkt am Beispiel eines Fußgängerbrückensystems aus Aluminium. Master Thesis. Section for Light Metal Structures and Fatigue & Chair of Business Economics, Management, Logistics and Production, Munich: Technische Universität München. Blanchard, B.S. & Fabrycky, W.J. 1981. System Engineering and Analysis. New Jersey, USA: PrenticeHall. Inc. Blanchard, B.S. & Fabrycky, W.J. 1998. System Engineering and Analysis (3rd Edition). New Jersey, USA: Prentice-Hall. Inc. CIB (International Council for Research and Innovation in Building and Construction) 1999. Agenda 21 on Sustainable Construction. CIB Report, Publication 237. Rotterdam, NL: CIB. DIN 1076 1999. Engineering structures in connection with roads – inspection and test. Berlin: German Institute for Standardization. European Aluminium Association (EAA) 2001, status: August 2004. Market development 2001. Statistics. www.eaa.net. European Aluminium Association (EAA) 2003, status: August 2004. Aluminium Industry Supports Recycling Initiatives. Recycling. www.eaa.net. European Commission December 2001. Integrated Pollution Prevention and Control (IPPC), Reference Document on best available techniques in the glass manufacturing industry. http://www.jrc.es/pub/english. cgi/0/733169: Directorate General, Joint Research Center. Hasselbeck, A., Klein, D. & Riesemann, M. 2004. Methoden für eine nachhaltig profitable Unternehmensführung zur Erhöhung der Wettbewerbsfähigkeit von Aluminiumbauprodukten. Project Study. Section for Light Metal Structures and Fatigue & Chair of Business Economics, Management, Logistics and Production, Technische Universität München. Herzog, K. 2003. Lebenszykluskosten von Baukonstruktionen. In Darmstädter Nachhaltigkeits-Symposium Chapter XVI. Darmstadt. Hydro, status: August 2004. “Hot in the City”. www.hydro.com/de/press_room/features/hot_ in_the_city.html. Hoefle, R. 2003. Lebenszykluskosten von Aluminiumkonstruktionen (LCCA). Master’s Thesis. Section for Light Metal Structures and Fatigue, Technische Universität München.

553

Hontelez, J. (Secretary General) 2002. The 6th Environmental Action Programme of the European Union – Legally Binding Commitments and Obligations. http://www.eeb.org/activities/env_action_programmes/ Special-Report-on-6EAP-December-2002.pdf: European Environmental Bureau. Kosteas, D. & Meyer-Sternberg, M. 2000. Aluminium – Brücken. Eigenschaften und Rahmenparameter für die Anwendung von Aluminium im Brückenbau. In Proc. symp., 24-25 February 2000. 2. Duisburger Aluminium-Tage. Duisburg: SLV Duisburg GmbH, Aluminium-Zentrale e.V. Lingg, B. & Villiger, S. 2002. Energy and Cost Assessment of a High Speed Ferry. KTH Stockholm/ETH Zürich. Töpfer, K. 1997. Nachhaltige Baupolitik zwischen Ökonomie und Ökologie. Bonn-Bad Godesberg: Federal Ministry of the Interior for Land Use Planning, Civil Engineering and Urban Development. WasteOnline, status: August 2004. Aluminium Recycling. http://www.wasteonline.org.uk/resources/ InformationSheets/Aluminum.htm. Werner, G. 1986. Dokumentation und Ermittlung realitätsbezogener und bauart-spezifischer Unterhaltskosten von Holzbrücken. Research project I.4-35658. Stuttgart. Woodward, D.G. 1997. Life Cycle Costing – theory, information acquisition and application. International Journal of Project Management 15(6): 335–344.

554

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Functional assessment of lightweight construction solutions in view of sustainability L. Bragança & P. Mendonça Civil Engineering Department, University of Minho, Guimarães, Portugal

ABSTRACT: The amount of waste produced every year, the exhaustion of resources and the construction solutions currently used on construction may not be sustainable in the future. All these issues lead to the research of new construction techniques, the recycling of waste into useful materials, the re-use of construction materials, etc. Most of the new and innovative solutions arise from the general feeling that something should be done to change the conventional way of construction in order to give an answer to current society concerns: the reduction of energy consumption, the minimization of pollution problems, the maximisation of the use of renewable and/or recyclable materials, etc. The aim of this study is to evaluate the potentialities of using more lightweight construction solutions with respect to functional comfort criteria (thermal, acoustic and visual comfort) and to assess the relative merits of this type of construction in view of maximising sustainability. Beyond the structural behaviour of a building the demand of a better habitat requires also a good performance in terms of serviceability. In this work the performance of lightweight construction solutions (optimized for reducing environmental impact) and conventional construction solutions were compared under the energy costs point of view (construction and heating). The acoustic performance was also studied, but just in order to achieve similar conditions, thus becoming not relevant to the purpose of this study.

1 INTRODUCTION 1.1 Historical evolution of housing construction systems in Portugal In the past centuries, at least until 50 years ago, in spite of an extremely heavy stone or massive brick envelope wall (it arrives to more then 1000 kg/m2 ), some of the construction elements in housing buildings in Portugal were lightweight, mainly timber pavements (approximately 50–100 kg/m2 ), timber/clay dividing walls and timber covering structures (approximately 150–200 kg/m2 ). Recently, with the generalisation of steel reinforced concrete and industrialised hollow bricks, the more usual attitude is to generalise the use of the so called “lightweightened” concrete construction system (with approximately 350–400 kg/m2 for a 0,22 m pavement slab and a similar weight for a double pane hollow brick envelope wall, generally with insulation on the air gap) in conventional housing buildings. We can conclude that, in spite of some relative increment on structural performance, the average weight of a housing building is very similar to 50 years ago, but the environmental impact costs per square metre have increased and the possibilities of recycling their components have decreased (Mendonça 2003). Reducing the specific weight of industrialised construction materials and systems can in fact have a significant role on reducing environmental costs, namely by the use of prefabricated modular systems that require no cranes and other heavyweight equipment to erect and have smaller energy costs associated with transport and even with the construction materials themselves. One main problem is that lightweight buildings are usually characterised by a small thermal inertia that results in an excessive daily thermal temperature swing, and thus they are not usually considered on bioclimatic approaches on temperate climates. 555

Objectives

Reduce the environmental impact of housing constructions

Strategies Re-use Recycle Reduce

Energy consumption on materials (PEC)

Energy operating consumption (maintaining comfort)

Figure 1. General objectives of this study.

1.2 Objectives The general objectives of this work are shown in Figure 1. There are several strategies that can lead to reduce the environmental impact of buildings. Recycle and re-use of the materials and even the buildings itself are possible, but are not the issue to be discussed on this paper. The strategy proposed here will be based on the reduction and how it can be achieved by optimizing the weight on architectural and construction systems. There will be focused two different aspects: one is a research on optimizing the total primary energy consumption (PEC) of construction materials and their transport, the other is based on reducing the energy operating consumptions for maintaining thermal comfort, even using the maximum possible passive solar gains. In order to compare the relative influence of these aspects, measurements were carried out on two solar passive test cells. 2 ENERGY CONSUMPTION ON BUILDINGS 2.1 Reducing energy consumption on construction Reducing the weight of materials used implies smaller environmental damages due to the extraction of prime materials, to their transformation processes and to the work yards, with reduction of the noise, dust, wastes and the consumption of energy during the construction and a proportional reduction on loss factors and specially on transport energy costs. The maximum use of local and less-transformed raw materials, or recycled ones, means reduction. But also the minimum use of those that are not locally available, such as steel for reinforcing concrete, cement, brick and an optimized use of those that, in spite of not being local or low energy, can compensate on savings, over their lifespan, such as glass or insulation. We should have in mind that a road transport by truck implies 2890 kJ/t/km (802,78 kW · h/kg · km), being one of the most pollutant ways for transporting construction materials, as can be seen on Table 1, and this is the most used mode of transport in Portugal. 2.2 Reducing operating energy In what respects the structure and the materials used, bioclimatic housing buildings in South European climates are even more heavyweight than conventional ones. Concrete and brick are 556

Table 1. Primary energy use by different modes of freight transport (Energy Research Group 1999). Emissions (g/T · km)

Water

Rail

Road

Air

CO2 CH4 NOx CO VOCs Energy (kJ/T · km)

30 0.04 0.4 0.12 0.1 423

41 0.06 0.2 0.05 0.08 677

207 0.3 3.6 2.4 1.1 2890

1206 2.0 5.5 1.4 3.0 15839

used in the interior pane of double envelope walls and in pavements, in order to increase thermal storage capacity. But it could be questioned if the overall weight could not be reduced by introducing more accurate systems. When the materials and labour are locally available (as adobe or stone), the environmental cost is reduced, but the increase of the global mass of the building implies other problems, such as the high economical cost of an intensive labour or the difficulty for increasing density by the increment of floors (even to more than two). Thermal mass materials still should be used, but in a rational way, related to local availability and just to fit thermal storage necessities. Some construction elements cannot be always locally available, (such as steel, concrete, ceramics and specially glass), and thus this is an area where optimisation can be even more effective (Mendonça 2003). In housing, the thermal gains could be higher in a direct gain strategy, with the concrete pavement slab, the interior walls and the interior pane of exterior walls taking the role of thermal storage, but the temperature and glare due to excessive solar radiation penetrating the interior occupied areas are a cause of discomfort. Apart from the degradation of the furniture and other equipment, a direct gain strategy is not a good solution, also due to the necessity of daily operating a night mobile insulation system. An indirect gain solution could be more effective in order to keep interior comfort in a more functional way, and guarantees that project values are closer to reality.

3 TEST CELLS STUDY 3.1 Characterization of the test cells study The proposed strategy of reducing the overall environmental impact of buildings was based on a mixedweight housing principle, with a thermal zoning concept and passive solar indirect gain that was expected to lead to an overall weight reduction on construction but without increasing operating energy. A research was undertaken using two test cells simulating areas of the Architectural designs shown on Figure 2. The plan on the left is the proposed mixed weight and mixed use housing unit (working on North area with direct lighting and sleeping on South area with indirect solar gains). The right plan simulates a conventional housing unit (but it has also an optimized solar exposition and mixed direct/indirect solar gains). The test cells studied have a rectangular shape (approximately 6,5 × 3,1 m), both are South oriented and have an horizontally moving window that is able to perform a sunspace or a Trombe wall as shown in Figure 3 on the right side. Test cell 1 is the non conventional cell, where the thermal performance of the mixedweight construction was studied. This test cell is divided in two parts separated by a wood moving partition: an heavyweight South oriented zone (sleeping area) with concrete structure, pavement and ceiling slabs, adobe walls and a North oriented lightweight zone with timber structure and sandwich pavement, ceiling and walls. In the heavyweight area Wall 1 is an adobe thermal gaining wall without insulation and a black painting exterior finishing and Wall 2 is a double pane wall with a 15 cm adobe pane on the interior and a wood cement exterior board with a ventilated 15 cm air 557

Figure 2. Plans of proposed and conventional housing units (Mendonça 2003).

Wall 3

Test cell 1

N Wall 4 Test cell 2

Wall 2 Wall 1

Sunspace

Trombe wall

Figure 3. Test cells’ plan and schematic vertical section of moving window (through wall 1) to create a Sunspace or a Trombe wall (distances in m).

gap with 5 cm expanded cork insulation. The North oriented zone (working area) has sandwich lightweight pavement and ceiling made with wood cement board and expanded cork insulation and triple pane walls with an exterior ventilated 15 cm air gap and an interior super-insulated air gap with 8 cm of expanded cork + 2 cm of coconut fibre. For comparative analysis, a conventional reference cell, named test cell 2 on Figure 3, with the same dimensional characteristics, but made with a conventional construction solution, was also studied. This cell corresponds to a conventional solution on contemporary Portuguese construction and has a construction system based on a steel reinforced concrete structure, with pavement and ceiling on pre-stressed concrete “T” beams and hollow brick and exterior double pane (15 + 11 cm) 558

Figure 4. Test cells’ vertical scheme of the North and South façades.

Figure 5. Test cells’ vertical scheme of the East and West façades (distances in m).

Figure 6. Vertical sections of test cells 1 and 2 – sunspace configuration (distances in m).

hollow brick wall with 4 cm of extruded polystyrene (XPS) placed in the air gap and finished with plaster on both sides. Figures 4, 5 and 6 show the vertical schemes of the façades and a vertical section of each test cell. 3.2 Energy operating consumption Long term energy savings implies more than a correct design of façades. In countries with an annual and daily thermal amplitude oscillating below and above the temperature of ideal interior comfort, such as Portugal, where this study is being made (between a minimum of −2,5◦ C and a maximum of 35◦ C) and a daily thermal amplitude of 10◦ C (Mendes et al. 1989), thermal inertia is even more important than insulation capacity, as its absence can result in a night rapid descent of temperature and a resulting excessive daily thermal swing in the interior. Since the South facing walls can take the main role of thermal gains, the bet can be to optimise their performance, and so to use it mainly for indirect gain. The use of combined solutions of ventilation/heat storage, namely 559

Table 2. Embodied energy and weight of materials used in proposed and conventional test cells. Materials used

Weight (kg)

kWh/kg

PEC (kWh)

200,00

44,48

8896,00

18344,80 2161,35 681,32 884,40

0,33 1,08 2,78 1,11

6053,78 2334,26 1894,07 981,68

75,00 34,00

9,73 19,44

729,75 660,96

106,80 112,50 397,80 16,39 1971,27 306,00 57,80 9,50 107,10 4995,00 83,49 144,00 1,53 3,60

5,11 4,05 1,05 24,19 0,18 1,05 3,90 21,55 1,39 0,03 1,08 0,28 24,19 5,56

545,75 455,63 417,69 396,47 354,83 321,30 225,42 204,73 148,87 134,87 90,17 40,03 37,01 20,02

30693,65 80,00 30573,65

0,18

24943,28 14,40 16061,68

Test cell 1 (Proposed) Aluminium (commercial 30% recycled) Concrete Particle board (cement/wood) Steel (commercial 20% recycled) Insulation (expanded cork particle board) Stainless steel Vulcanized rubber (exterior board fixing sealant) Glass Asphalt/carton shingle Carton/plaster gypsum board Alveolar polycarbonate Timber (local treated pine) Gypsum (projected plaster) Insulation (Coconut fibre) Synthetic varnish Timber floating pavement Adobe Particle board (wood) Lime painting (slaked) Polyethylene shingle (expanded) Plastic painting (water based) Total (with aluminium frame on solarspace) (timber frame) Total (with timber frame on solarspace) Pavement area 17 m2 Total/m2 (with timber frame on solarspace)

1798,45

944,81

Test cell 2 (Conventional) Clay (hollow brick) Aluminium (commercial 30% recycled) Concrete/cement mortar Steel (commercial 20% recycled) Polystyrene extruded (XPS) Stainless steel Glass Asphalt/carton shingle Gypsum (projected plaster) Alveolar polycarbonate Particle board (cement/wood) Timber (local treated pine) Timber floating pavement Plastic painting (water based) Particle board (wood) Synthetic varnish Polyethylene shingle (expanded) Total

9778,13 250,00

1,26 44,48

12320,44 11120,00

32411,60 955,60 54,00 75,00 127,20 112,50 270,00 8,91 153,90 851,13 94,50 11,70 40,32 1,70 1,35

0,33 2,78 27,86 9,73 5,11 4,05 1,05 24,19 1,08 0,18 1,39 5,56 1,08 21,55 24,19

10695,83 2656,57 1504,44 729,75 649,99 455,63 283,50 215,53 166,21 153,20 131,36 65,05 43,55 36,64 32,66

45197,54

41260,33

3013,17

2750,69

Pavement area 15 m2 Total/m2

560

Table 3. Embodied energy, operating energy economical and energetic costs in a 50 years life span.

Test cell 1 2

Sunspace Trombe wall Sunspace Trombe wall

Operating energy cost in life span (a/m2 ) 235 369 323 417

Construction cost (a/m2 )

Embodied energy (kWh/m2 )

Materials transport energy (kWh/m2 )

Operating energy consumption (kWh/m2 )

1111

1470

121,4

1267

2756,6

241,9

2374,5 3728,5 3261,5 4218,5

by the use of trombe walls is an effective method of natural heating during the cold season, when there is enough solar radiation. One problem is that the construction of these interior walls between the window and the occupied zones decrease interior illumination, for they are opaque. The need of a great window surface oriented to South and with its major area closed by thermal gaining opaque walls forces the building to open more to other solar orientations. In the proposed solution the working area for studying, receives natural illumination through a translucent window (in alveolar polycarbonate and timber frame) oriented to North. This North great light capture causes more fluctuation on the interior temperature, but it also permits to have a more uniform lighting for this area, that was expected to have a daytime occupation (working areas). The heavyweight area have a smaller fluctuation and when the partition door is closed, during night hours, the temperature swing in this area is lower than the reference test cell. Summer campaign measurements also revealed that cooling needs were not relevant, so they were not considered (the zone of this study was Guimarães and it is in a Northern temperate area of Portugal – not very far from sea so it still gets some maritime influence). The heating overall energetic needs were measured and calculated using the method proposed by CSTB (CSTB 1988) and these values were compared with the other energy aspects – primary energy of construction materials (PEC) and materials transport. 4 ENERGY COST EVALUATION Table 2 presents the measures of the embodied energy of materials used in the proposed test cell (1) and in the reference test cell (2). As we can conclude from the analysis of this table, in test cell 1 and 2, the aluminium of the exterior window frames, in spite of being lightweight, have a very high PEC. Aluminium was the solution adopted here just for the specific purpose of being a mobile window (it makes a telescopic movement in order to study the influence of the sunspace area in the thermal gains as it was referred previously), and other solutions were not exequible for this purpose. In a real situation, a wood frame on the frontal window of the sunspace or the Trombe wall would have a much smaller embodied energy. Those are the values referred in parenthesis and it can be seen that the total PEC decreases 43% on the proposed solution. On conventional construction, hollow brick and concrete take the greatest portion of the embodied energy. For the comparative cost analysis presented on Table 3, where it can be seen that the proposed solution is a little more economical, the life span considered was 50 years with a 2,5% inflation rate. The operating costs were considered just for the heating season, in a 18◦ C base temperature and heating with electric wall radiators. Note that in certain regions of Portugal, stone would be preferable to Adobe masonry in interior heavyweight walls on proposed solution, but the average final value would be very similar as stone has the same PEC. The reduction in weight was in a great part from industrialised non-locally available components, so the percentage of reduction associated with transport, mainly truck by road was significant. To the transport study was considered that all materials made an average of 100 km. The average distance in the transport of adobe (compacted earth) was considered to be 0 km. 561

5000 4000 3000 2000 1000 0

Embodied + materials transport energy (kWh/m2) Operating energy consumption (kWh/m2) Test Cell 1 sunspace

Test Cell 1 trombe wall

Test Cell 2 - Test Cell 2 trombe wall sunspace

1400 1200 1000 800 600 400 200 0

Energy cost in life span (€/m2) Construction cost (€/m2)

Test Cell 1 sunspace

Test Cell 1 trombe wall

Test Cell 2 - Test Cell 2 sunspace trombe wall

Test Cell 1 - sunspace

Test Cell 1 - trombe wall

30% 40%

60% 70%

Test Cell 2 - trombe wall

Test Cell 2 - sunspace

42% 48% 58%

52%

Embodied + materials transport energy Operating energy consumption

Figures 7, 8 and 9. Comparisons between operating energy and embodied + transport energy of test cells.

A Nordic author says that “The amount of energy that actually goes into the production of building materials is between 6 and 20% of the total energy consumption during 50 years of use, depending on the building method, climate, etc” (Berge 1999). The percentage that most suits the Portuguese reality is maybe closer to 20%, because of the particular amenity of the climate, but we can even state that the amount of energy that goes into the production of building materials can easily reach values between 30 and 48% of the total energy consumption during 50 years of use.

5 CONCLUSION This paper show the potentialities associated with the use of lightweight materials combined with locally available thermal mass materials, in order to achieve a good environmental profile. In 562

the end of the life span of most contemporary housing buildings, the dismantling, treatment and transport of waste materials also represents energy savings. The proposed solution is also easy to dismantle and almost all of its materials are reusable or recyclable, especially if compared with nowadays most common construction system used in Portugal – steel reinforced concrete structure with clay hollow brick walls and pavements. The example presented in this paper shows how the environmental impact measured on the Primary Energy Consumption of materials in the proposed innovative mixedweight test cell can reach almost a 50% of improvement when compared with a conventional one and still having a similar economical cost (even a little lower). In spite of the increasing evolution that lightweight materials and systems achieved in the recent past, namely to their durability and stability there is still a long way to go through, before these solutions can be widely accepted. Mixing them with heavyweight solutions, and proving the fact that this strategy is environmentally suitable to be used in bioclimatic constructions, even to temperate climates as the South European ones, can be a step forward. It could also be concluded that the solar passive optimized solution is more sustainable in a Sunspace configuration then in a Trombe wall configuration.

ACKNOWLEDGEMENT This work is an FCT (Fundação para a Ciência e Tecnologia – Portugal) funded project. The authors also wish to thank Pedro Silva for his help in the treatment of some of the data presented on this paper. REFERENCES Mendonça, P. & Bragança, L. 2003. “Energy Optimization through Thermal Zoning – the outer skin”; Proceedings of the “Healthy Buildings – 7th International Conference”; School of Design and Environment, National University of Singapore; 7–11 December. Energy Research Group. 1999. “A Green Vitruvius – Principles and practice of Sustainable Architectural Design”; James & James Ltd. Mendes, J.; Guerreiro, M.; Pina dos Santos, C. & Vasconcelos de Paiva, J. 1989. “Temperaturas exteriores de projecto e números de graus-dias”; Instituto Nacional de Meteorologia e Geofísica/Laboratório Nacional de Engenharia Cívil, Lisboa. CSTB 1988. “Régles Th-BV, régles de calcul du coefficient de besoins de chauffage des logements - annexes”; cahiers du Centre Sscientifique et Technique du Bâtiment, livraison 292, cahier 2274; Paris, Septembre. Berge, B. 1999. “The Ecology of Building Materials”. Architectural Press, Bath.

563

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Comparative assessment of exterior walls construction solutions’ sustainability L. Bragança & R. Mateus University of Minho, Department of Civil Engineering, Guimarães, Portugal

ABSTRACT: In this document a methodology for the comparative sustainability assessment of construction solutions will be approached. This work intends to be a contribution for the Construction in order to turn this industry more compatible with the sustainable development aims. The presented methodology will be applied to some conventional and non-conventional exterior walls construction solutions in order to find, inside the sample, the most sustainable solution.

1 INTRODUCTION Construction industry is one of the most important European economical sectors, but it still relies too much on traditional construction methods and unskilled handwork, being characterized by an excessive use of natural resources and energy. This implies great environmental, social and economical impacts that could easily be reduced. This industry, in general, and the buildings sector in particular, contributes to the degradation of the environment through the dilapidation of natural resources. Building construction consumes 40% of the raw stone, gravel, and sand globally used each year, and 25% of the virgin wood. Building also account for 40% of the energy and 16% of the water annually used worldwide (Roodman, 1995). This reality is incompatible with the sustainable development aims that seek the balance between the environmental, economic and social dimensions. One of the possible solutions for this problem is the use of building technologies more compatible with the environmental balance. In the last years, even with a small impact, an evolution in this domain has been observed, and now there are new materials and construction solutions more sustainable than the conventional ones. The Project Team, mainly during the Design Phase, has the biggest responsibilities in selecting technologies with high eco-efficient materials content. This kind of materials has low environmental impact during its live cycle, without committing the necessary functional performance for the construction element and the final product economical viability. In the majority of the less developed countries, this subject is still very recent. In these countries, the biggest part of the construction companies and the population in general, are not sufficiently informed about the individual and collective advantages of the “Sustainable Construction” concept. In the developed countries, this subjective is no more an environmentalist’s exclusive flag, being nowadays one of the most important aspects in the construction’s global quality assessment. In the construction market the number products with the sustainable label is increasing, without any kind of control. So, some of those solutions could not present any advantages relatively to the conventional. The sustainable label is adopted in a way to increase sales. Consequently is urgent the development and application of methodologies that could help the design teams in choosing construction solutions that turns the construction’s future more sustainable. 2 SUSTAINABLE ASSESSMENT METHODOLOGIES In the solutions’ sustainability assessment, several parameters could be analyzed, some of them not correlated and/or not expressed in the same units. On the other hand, the way that each parameter 565

influences the sustainability is neither consensual nor unalterable along the time. So, it is difficult to express a solution’s sustainability in absolute terms, through an indicator that integrates all of the analyzed parameters and that allows the quantitative classification of its sustainability. For example, a solution with good environmental performance but without the minimum functional requirements could not be considered sustainable. On the other hand, a solution with good environmental and functional performances, but with much higher life-cycle costs than the conventional one, could not be considered sustainable, because the prohibitive costs are a barrier for its implementation. The sustainability is a relative subject that should be assessed comparatively and relatively to the most widely used solution – conventional/reference solution – in a certain country/local. This way, comparing each of the selected sustainability indicators it is possible to verify, at the level of each one, if the solution in analysis is better or worse than the conventional one. The most sustainable solution depends on the technological limit of each moment. In a construction solution sustainability assessment process, the first step consists in gathering the most relevant functional and technical data about the construction solution. The second step consists in selecting an appropriate method that allows the quantitatively assessment of the sustainability. The methodology to adopt should be simple and flexible, to conveniently help the design teams in choosing a certain technology in detriment of others less sustainable. In certain developed countries, some systems and tools for the sustainability assessment are being implemented or in the development phase. The most important are: Building Research Establishment Environmental Assessment Method (BREEAM), Leadership in Energy & Environmental Design (LEED) and Green Building Challenge (GBTool). The presented methodologies aim the evaluation of the global sustainability of a building. Its application is complex and needs the previous knowledge of some data. Some of the sustainability assessment tools have datasheets that gather some of the needed data, although the data is related with the particular aspects of the country of origin, which turns its application in a different country very difficult. These systems focus mainly the building environmental impact assessment in a global perspective. The sustainability of the construction solutions is one of the analyzed aspects in the buildings’ global sustainability assessment. In this perspective and for the propose of this work a methodology named Methodology for the Relative Assessment of the Construction Solutions Sustainability (MARS-SC), is presented (Mateus, 2004). 2.1 Methodology for the relative assessment of the construction solutions sustainability (MARS-SC) In the MARS-SC the assessment of the sustainability is accomplished relatively to the most applied solution – conventional/reference solution – in a certain place. In this methodology three groups of indicators are approached: environmental, functional and economical. The methodology follows the following steps: 1st Step: Defining the indicators to be evaluated on each group. The number of indicators analyzed on each group can be adjusted depending on the specific characteristics of the construction solution, on its functional demands, on the evaluation objectives and on the available data. Table 1 shows some of the most important parameters that could be analyzed in this methodology. 2nd Step: Calculating the comparison indexes. The comparison between the solution under analysis and the reference solution is accomplished at the level of each parameter through a comparison between indexes. These indexes express the relationship between the value of a certain indicator in the solution under analysis and the same parameter in the conventional solution that allows verifying, relatively to each analyzed parameter, if the solution is better or worse than the conventional construction solution. The indexes are calculated by the equation 1:

566

Table 1. Indicators that can be analyzed in the MARS-SC methodology. Indicators Environmental

Functional

Economical

Global warming potential (GWP) Primary energy consumption (PEC) Recycled content Recycling potential Raw material’s reserves Eutrophication potential

Air born sound insulation Percussion sound insulation Thermal insulation Durability Fire resistance Flexibility of use

Construction cost Utilization cost Rehabilitation cost Demolition cost Residual valor End use treatment cost

Table 2. Indicators score (Ni) through the value of the comparison indexes (Ix ). Ix

Score (Ni )

≤0.6 [0.6;0.8] [0.8;1.0] 1.0 [1.0;1.2] [1.2;1.4] ≥1.4

3 2 1 0 −1 −2 −3

where Ix = index of the indicator x; Vx = value of the indicator x in the solution in analysis; and V  x = Value of the indicator x in the conventional solution. 3rd Step: Giving a score for each indicator. Through the indexes value the score of each indicator (Ni ) is defined, in a scale of values between −3 and 3. If the score is negative, the solution in analysis is worse than the conventional one, at the level of that indicator. Otherwise the solution in analysis is better than the conventional one. The score is given through the Table 2. 4th Step: Graphical representation of each indicator’s score (Sustainable Profile). The indicators’ scores are represented in a radar type graphic with a number of rays equal to the number of indicators in analysis. 5th Step: Determining the solution’s Performance Scores at the level of each group of indicators (NDi ). The solution’s performance is evaluated inside each group of parameters, through equations 2 to 5. With the NDi it is possible to synthesize in one value the solution’s performance inside each group.

where NDA = environmental performance’s score; NDF = functional performance’s score; NDE = economical performance’s score; WAi = environmental indicator (i) weighting factor; WFi = functional indicator (i) weighting factor; WE i = economical indicator (i) weighting factor; m = number of environmental indicators in study; n = number of functional indicators 567

Table 3. Construction solution’s sustainability classification. NS

Sustainability classification

<−1 [−1,−0] 0 [0,1] [1,2] [2,3] 3

Mediocre Unsatisfactory Reference Better Good Very good Excellent

in study; o = number of economical indicators in study; NIAi = environmental indicator (i) score; NIFi = functional indicator (i) score; NIEi = economical indicator (i) score. The weighting factor of each indicator in the determination of the three performance scores is not consensual. At the level of the environmental indicators there are some studies which allow the almost consensual definition of its weights. The most important were the studies performed by the United States Environmental Protection Agency (EPA). EPA’s studies identified, for a list of twelve environmental indicators, the relative importance of each one among the others through their environmental effects (EPA, 1990). In the MARS-SC the weighting factors presented in that study are used directly or by extrapolation. There are no studies about the functional indicators. So, it is considered an equal weight distribution per each indicator. The use of more consensual values could be possible by the application of a Multiattribute Decision Analysis methodology. In the economical indicators domain, considering that the biggest part of the construction solution’s life cycle is related to the use phase, it is suggested that the maintaining and operational costs should have bigger weighting factors than, for example, the construction costs, in the economical performance assessment. 6th Step: Sustainable Score (NS) calculation. Using Equation 6 it is possible to synthesize in on value the solution’s performance at the level of the three vectors considered in the sustainability assessment.

where NS = solution’s sustainable score; W1 = environmental indicators group’s weighting factor; W2 = functional indicators group’s weighting factor; W3 = economical indicators group’s weighting factor. The way that each indicators group influences the sustainability is not consensual. Although, aiming a bigger compatibility between the artificial environment and the natural one without forgetting the functional requirements of the construction solutions, it is current the use of bigger weights for the environmental and functional groups. In this way, in the MARS-SC is used the following distribution of weights: W1 = 0.40; W2 = 0.40; W3 = 0.20. Consulting Table 3 and considering the NS it is possible to classify the construction solution’s relative sustainability. 3 COMPARATIVE ASSESSMENT OF EXTERIOR WALLS CONSTRUCTION SOLUTIONS’ SUSTAINABILITY 3.1 Used methodology The Methodology for the Relative Assessment of the Construction Solutions Sustainability (MARS-SC) is used in the performed evaluation. The sustainability is evaluated, relatively to 568

Table 4. Weighting factors considered in the assessment. Group

Indicator

Indicator’s weighting factor

Group’s weighting factor

Environmental

GWP PEC

0.75 0.25

0.40

Functional

Dn,w U WT

0.33 0.33 0.33

0.40

Economical

CC

1.00

0.20

Air gap (2cm)

XPS (2cm)

Hollow brick (15cm)

Hollow brick (11cm)

Render (1.5cm)

Render (1.5cm)

Figure 1. Conventional/reference solution (Wall 1). Air gap (2cm)

XPS (3cm)

Stone (30cm) Hollow brick (11cm) Render (1.5cm)

Figure 2. Wall 2 construction solution.

the conventional solution, through the comparison of two environmental parameters (global warming potential – GWP and primary energy consumption – PEC), three functional (air born sound insulation – Dn,w , thermal insulation – U and wall’s thickness – WT) and one economical (construction cost – CC). The weighting factors considered in the Performance Scores (NDi ) and in the Sustainable Score (SS) calculation are in Table 4. 3.2 Construction solution’s characterization The conventional/reference solution (wall 1) is one of the most applied technologies in exterior walls in Portugal. The solution is a double (15 + 11 cm) hollow brick wall with a 2 cm thick extruded polystyrene (XPS) layer placed on the air gap. Each surface of the wall is covered by a 1,5 cm thick layer of render. Besides the conventional solution, more three construction solutions were analyzed. Having in mind the biggest relative importance of the thermal behaviour towards the other functional requirements, the construction solutions were defined in a way that their thermal behaviour was, in minimum, equal. The other analyzed construction solutions were: Wall 2: Double pane wall with an exterior stone pane and an interior hollow brick pane (Figure 2). 569

EPS (4cm) Reinforced render (1cm) Hollow brick (22cm) Render (1.5cm)

Figure 3. Wall 3 construction solution. Mineral wool in blankets (14cm)

OSB (1.2cm) EPS (1cm)

Ligh gauge steel profile

Reinforced render (1cm)

2xPlasterboard (2x1.25cm)

Figure 4. Wall 4 construction solution. Table 5. Indicators value.

Wall 3: Wall 4:

Indicator

Wall 1

Wall 2

Wall 3

Wall 4

GWP (g/m2 ) PEC (kW · h/m2 ) Dn,w (dB) U (W/m2 ·◦ C) WT (cm) CC (a/m2 )

46511 197 51 0.70 33.00 46.70

44221 169 59 0.65 47.50 125.90

36538 159 49 0.60 28.50 41.80

49883 171 51 0.23 19.60 133.40

Single pane wall with external thermal insulation with rendering (Figure 3). Light gauge steel frame wall (Figure 4).

3.3 Results Table 5 resumes the obtained results in the sustainable indicators evaluation. Table 6 resumes, for each construction solution, the performance scores of each indicators group and also the respective sustainable score. Inside the analysed sample and in accordance with the considered indicators, the results shows that the most sustainable solution is the single pane wall with external thermal insulation with rendering (Wall 3) and the less sustainable is the double pane wall with an exterior stone pane and an interior hollow brick pane (Wall 2). The light gauge steel frame wall’s (Wall 4) sustainability is similar to the reference solution. At the environmental performance level, Wall 3 is the best solution, while the Wall 4 is the worst. The Wall 4 has the best functional performance and the Wall 2 the worst. Wall 3 has the best economical performance. 4 CONCLUSIONS The project teams have big responsibilities in searching the sustainability in the building and real estate sectors, through the selection and use of construction solutions with improved environmental, 570

Table 6. Sustainability of the construction solutions. Performance Construction Sustainable solution profile Wall 2 CC

GWP 3 2 1 0 -1 -2 -3

WT

Sustainable Relative Env. (NDA ) Fun. (NDF ) Econ. (NDE ) score (NS) sustainability 0.25

0.33

3.00

−0.63

1.75

0.66

1.00

1.16

Superior

−0.50

1.98

3.00

0.00

Reference

Unsatisfactory

PEC

Dn,w

U

Wall 3 CC

GWP 3 2 1 0 -1 -2 -3

WT

PEC

Dn,w

U

Wall 4 CC

GWP 3 2 1 0 -1 -2 -3

WT

PEC

Dn,w

U

functional and economical performances, during their whole life-cycle. The development and use of sustainability’s evaluation methodologies and tools are fundamental aspects for these goals. Analysing the MARS-SC results, it could be observed that they depend on the type and number of each group’s considered indicators and on the relative weight considered for each one. Aiming more consensual results, this methodology should be developed through the pre-definition of a list of indicators for each construction solution. The presented results and methodology intends to be a contribution to the sustainability in the construction industry domain, through the use of construction solutions with improved environmental, functional and economical performances. REFERENCES Mateus, Ricardo, 2004. Novas Tecnologias Construtivas Com Vista à Sustentabilidade da Construção. Dissertação de Mestrado, Departamento de Engenharia Civil da Universidade do Minho, Portugal. United States Environmental Agency (EPA), Science Advisory Board (SAB), 2000. Toward integrated Environmental Decision Making. EPA-SAB-EC-00-011, Washington, D.C.

571

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

The impact of climate parameters on air movement in ventilated roofs air gap E. Monstvilas, V. Stankevicius & R. Bliudzius Institute of Architecture and Construction of Kaunas University of Technology, Kaunas, Lithuania

ABSTRACT: In cool and cold climates, ventilated roofs do not operate as effectively as they do in the Southern countries. For the detailed analysis of the climate factors’ impact to the air movement in the roof ventilated air gap, the calculation method has been compiled with the corresponding software. The air movement velocity in the ventilated air gap, air temperature alteration at the air outlet vent due to the heat exchange in the gap, indoor air temperature, outdoor air temperature and wind speed, building height and aerodynamic building parameters, dimensions of the ventilated air gap and materials, pressure losses in the ventilated air gap are estimated in the suggested calculation method. The main problems of functioning of the ventilated roofs are faced with during the cold periods of the year. At this time, the temperature of roof hydro-cover is close to the outdoor temperature in conditions of the cold and cool climate. Then the pressure difference necessary to cause the air movement in the ventilated air gap depend only on the wind pressure. In this case the temperature impact is quite low. The dimensions of the ventilated air gap necessary to ensure the functioning of the ventilated roof depend on the dominant wind speed in the location and the constructional singularities of the air gap itself. The worst situation is found in the ventilated roofs of timber construction, when the hydro-protective cover is installed on the timber purlins. The high air pressure losses are caused by the big amount of the purlins

1 DESCRIPTION Roofs of all inclines with constantly ventilated air gap are attributed to ventilated roofs. Ventilated roofs are designed to allow air to flow through the roof cavity (the space above the insulation) to remove moist air that migrates into that space. Roofs of this type are widely used in dwelling and industrial buildings. In cool and cold climates, ventilated roofs do not operate as effectively as they do in the Southern countries. The roofs are seldom ventilated satisfactorily to absorb the incoming moisture rapidly enough. The continuing low temperatures cause the water vapour condensation, which turns into frost on cold surfaces before it can exit. This frost continues to build up until spring, sometimes closing the cross section and stopping air movement, thus, damaging the structures. In regions with cold climate, if the roof is vented, the ceiling finishes, ceiling joists, vapour barriers and insulation have a much shorter life span due to this moisture accumulation. The ventilated air gap should be of adequate dimensions to warrant the natural air movement in it. If the dimensions are not appropriately chosen, the air movement in it shall be stopped causing the conditions of moisture condensation; then big amount of moisture could be accumulated in the roof structure leading to the deterioration of structure materials.

2 FIELD OF APPLICATION At present in practice of design and installation of ventilated roofs almost no respect is paid to the climatic data relating to the wind speed, location peculiarities, dimensions of the ventilated 573

air gap and other important parameters, adequacy of which is necessary for the trustworthy roof ventilation. The suggested calculation method enables to forecast the effectiveness of the ventilated roof at a particular geographic locality. 3 TECHNICAL INFORMATION Main symbols and units d θ ρ β U C c

– – – – –

thickness, m; temperature , ◦ C; density, kg/m3 ; surface roughness coefficient; heat transmission coefficient, W/(m2 ·K); – aerodynamic coefficient; – air specific heat capacity, c = 1008 J/(kg·K);

l A h ν Z p ζ

– – – – –

length, m; area, m2 ; height, m; velocity, m/s; friction losses due to local roughness in channel, Pa; – pressure, Pa; – resistance of the local obstacles

Main subscripts to symbols: sum – sum; air – air; e – external; i – internal;

inc out v.a.g. w

– – – –

inlet; outlet; ventilated air gap; wind.

4 STRUCTURAL ASPECTS The principal scheme of roof with the ventilated air gap is presented in Figure 1. The height of the ventilated air gap hv.a.g is not determined by the customary roof design procedure at present. The value is assumed as a constant quantity according to the recommendations arranged by hydro-insulating cover producers despite of the factors predetermining hv.a.g . Hydro-protection cover Underlay of hydroprotection l θair.out pair.out ρair.out

θe

∆h Ui-v.a.g.

νv.a.g. hv.a.g .

Uv.a.g.-e.

θe

θi

pe. ρe

Figure 1. Principal scheme of roof with ventilated air gap.

574

Waterproof diffuse film Thermal insulation Water vopour barrier Bearing structure

5 RESEARCH ACTIVITY If air movement is not sufficient in the air gap of the ventilated roof, the intensive deterioration process of roof structure takes place. After only 4 years of service of the roof under consideration the serious moisture damages have been found (Tywoniak, 2000). The deteriorating action of moisture could be different when the roof structure materials are of organic origin or metal. When roofs details are made of timber, and the ventilation is insufficient, it causes the mould germination on the surfaces of timber details and the rot of them as warm and moist air is promoting the development of microbes (Krus, 2001). When the roof is covered by zinc tin sheets, insufficient ventilation leads to the active corrosion of zinc film on inside surfaces (Zheng, 2003). One of the main problems which occur at the application of the ventilated roofs is the tightness of installed vapour barriers. If there are leakages in the barrier, moist air from the indoor is rising into the roof structure and significant moisture volumes could be condensed there during winter (Hens, 2003). The wind speed, locality type, the height of the building, the aerodynamic coefficients of the building (ENV 1991-2-4:2000, 2000), temperature impact, air pressure losses and some other factors will be included in the calculation of the natural air movement (Juodis, 1998; Technical Note AIVC 29, 1990; Kronvall, 1993). Heat transfer is running between the air and the surfaces of the ventilated roof air gap. Therefore, the temperature of the moving air is growing up or decreasing, in regard to the temperatures on the surfaces, depending accordingly on the indoor and outdoor air environments. The solution of equation, describing the heat transfer balance in the air gap, gives opportunity to determine the value of air temperature at every distance from the inlet vent (Barkauskas, 2000). The roughness of the inside air gap surfaces forms additional resistance to the air movement in the gap. This parameter must also be evaluated at the calculation when the air pressure losses in the air gap are checked (Tixomipob, 1981). Similar investigations are provided by specialists of building ventilation systems. The aims of their calculation models are often close to the problem under consideration, e.g. the estimation of the change of indoor parameters in the building at infiltration of the outdoor air or assurance of the natural ventilation in the selected building area (Andersen, 2003).

6 METHOD OF ANALYSIS For the detailed analysis of the climate factors’ impact to the air movement in the roof ventilated air gap, the calculation method has been compiled with the corresponding software. The air movement velocity in the ventilated air gap, air temperature alteration at the air outlet vent due to the heat exchange in the gap, indoor air temperature, outdoor air temperature and wind speed, building height and aerodynamic building parameters, dimensions of the ventilated air gap and materials, pressure losses in the ventilated air gap are estimated in the suggested calculation method. The incline of the roof is evaluated by the following roof dimensions: height difference between the air inlet and outlet vents and length of the ventilated air gap (the length of the roof pane). The value of air flow velocity in the ventilated air gap is assumed at the initial stage of the calculation. Most of researchers agree that the value of approximately 0.2 m/s is necessary for a normal service of the ventilated air gap in the roof. It is important that the initial presumptions for the air movement in the ventilated air gap were created. That means sufficient air pressure difference between the inlet and outlet vents. The mentioned difference could be created by impact of three main factors such as: different air density at the inlet and outlet vents, the wind effect and deflector or other ventilation operation of equipment. The total value of air pressure difference between the inlet and outlet vents in the ventilated air gap 575

psum , Pa, is determined by:

where: pd – air pressure difference due to different air density values at the inlet and outlet vents of ventilated air gap, Pa; pw – air pressure difference due to wind effect, Pa; peq – air pressure difference due to operation of roof ventilation equipment (vacuum ventilation chimneys, deflectors etc.), Pa. Air pressure difference due to different air density values pd , Pa, is determined:

where: g – gravitational acceleration (g = 9,81 m/s2 ). The temperature of outlet air from the ventilated air gap depends on a number of factors, including the air flow velocity in the gap. The value is determined according to the following expression (Barkauskas, 2000):

where: Ui−v.a.g. – heat transmission coefficient of roof structure, from inside air to ventilated air gap air, W/(m2. K);Uv.a.g.−e – heat transmission coefficient of roof structure, from ventilated air gap air to outdoor air, W/(m2. K); e – base of natural logarithm, e = 2,718; A – constant of integral:

Qv.a.g. – air flow through ventilated air gap of 1m width, kg/s:

The air pressure difference between the inlet and outlet vents, caused by wind effect on the vertical surface pw , Pa, is determined by the following expression:

where: 10 – factor of units transfer from kg/m2 to Pa; Ce – value of aerodynamic coefficient at the inlet vent of ventilated air gap; Ci – value of aerodynamic coefficient at the outlet vent of ventilated air gap; νw.inc , νw.out – wind speed at the inlet and outlet vents of the ventilated air gap correspondingly, m/s; g – gravitational acceleration; (g = 9,81 m/s2 ). The air movement in the ventilated air gap should be ensured on the necessary condition:

where: psum – total air pressure difference between the inlet and outlet vents, Pa; Rx – pressure losses due to the friction in the ventilated air gap sector, at the temperature of 20◦ C. The values found in the tables of ventilation guides according to the equivalent diameter of the ventilated air gap and air flow velocity in the sector under consideration, Pa/m; βx – the coefficient of the 576

surface roughness for the air gap sector under consideration. Depending on the materials used and the air flow velocity, the value could be selected from 1 to 1.48; lx – the length of ventilated air gap sector, m; K1 – correction coefficient of the temperature impact for friction pressure losses. As the alteration of the coefficient is not significant, the constant value could be taken (if the average outdoor temperature −5◦ C, then K1 = 1,07); Zx – air pressure losses due to the resistance of local obstacles in the ventilated air gap sector at temperature of 20◦ C, Pa; α – safety coefficient (α = 1.1–1.15). It is suggested to take α = 1,1. Air pressure losses due to the resistance of local obstacles in the ventilated air gap sector are determined by the following equation:

where: Zx – pressure losses due to the resistance of local obstacles in the ventilated air gap sector at temperature of 20◦ C, Pa; ζ – sum of local obstacles resistances in the ventilated air gap sector. The data on the values of the resistances could be found in the ventilation guides; pdin – dynamic air pressure in the sector, Pa; νx – air flow velocity in the ventilated air gap sector, m/s; ρav – average air density in the ventilated air gap, kg/m3 . If dimensions of the inlet or outlet vents differ from the cross-section of the ventilated air gap, they form additional pressure losses. Therefore they should be estimated as well. 7 RESULTS OF ANALYSIS The main problems of functioning of the ventilated roofs are faced with during the cold periods of the year. At this time, the temperature of roof hydro-cover is close to the outdoor temperature in conditions of the cold and cool climate. Then the pressure difference necessary to cause the air movement in the ventilated air gap depend only on the wind pressure. In this case the temperature impact is quite low. Table 1. Parameters of air movement in the air gap of ventilated roof and their interrelationship. Height of ventilated air gap hv.a.l. , m

Maximum length of ventilated air gap, when air movement is possible at wind speed of 3 m/s

300mm

0,025 0,035 0,05 0,1 0,2 0,3 0,4 0,5

35mm hv.a.l.

2m 2,4 m 3m 7,4 m 16 m 25 m 40 m 54 m

Required wind speed for ventilated air gap of 6 m length

Required wind speed for ventilated air gap of 12 m length

Most frequent roof scheme when the profiled steel sheets with ventilated air gap on timber purlins are used

5,2 m/s 4,7 m/s 4,1 m/s 2,7 m/s 1,9 m/s 1,5 m/s 1,2 m/s 1,1 m/s

577

7,4 m/s 6,5 m/s 5,6 m/s 3,8 m/s 2,7 m/s 2,1 m/s 1,7 m/s 1,5 m/s

The dimensions of the ventilated air gap necessary to ensure the functioning of the ventilated roof depend on the dominant wind speed in the location and the constructional singularities of the air gap itself. The worst situation is found in the ventilated roofs of timber construction, when the hydroprotective cover is installed on the timber purlins. The high air pressure losses are caused by the big amount of the purlins.

8 EXAMPLE OF APPLICATION The calculations are provided according to the method described in the paper with respect to producers’ recommendations of the hydro-protection materials. The results are presented in Table 1. The approximate height of the air gap is forecasted in regard to the data presented in the Table 1 and average monthly wind speed in winter at the locality, where the ventilated roof certain length is intended to be installed. REFERENCES Andersen K.T., 2003. Theory for natural ventilation by thermal buoyancy in one zone with uniform temperature. Building and Environment. Volume 38, issue 11. Barkauskas V., Stankeviˇcius V., 2000. Pastat u
ativar u
šilumin˙e technika. Technologija, Kaunas. ENV 1991-2-4:2000, 2000. Basis of design and actions on structures. Part 2–4: Actions on structures. Wind actions. Hens H. & Zheng R., 2003. Does performance based design impacts traditional solutions? Metal roofs as an example. Research in building Physics. Carmeliet, Hens & Vermeir (ed.). Proceedings of the 2nd international conference on building physics 14–18 September 2003, Antwerpen, Belgium. Juodis E., 1998. V e˙ dinimas. Enciklopedija, Vilnius. Kronvall J. & Sandberg P.I., 1993. Ventilation, heat and moisture conditions in attic spaces. Buildings physics in the Nordic countries. Bjarne Saxhof (ed.). Proceedings of the 3rd symposium 13–15 September 1993, Copenhagen, Denmark. Krus M., Sedlbauer K., Zillig W., Künzel H.M, 2001. A New Model for Mould Prediction and its Application on a Test Roof. Contribution to the II. International Scientific Conference on “The Current Problems of Building Physics in the Rural Building”, Cracow, Poland, Nov. 2001. Tywoniak J., 2000. Hygrothermal performance of building envelopes from environmental point of view. Integrated life-cycle design of materials and structures ILCDES 2000. Proceedings of the RILEM/CIB/ISO international symposium, 22–24 May 2000, Helsinki, Finland. Zheng R., Carmeliet J. & Hens H., 2003. The influence of roofing systems on the corrosion of zinc sheeting in highly insulated zinc roofs – a laboratory study. Research in building Physics. Carmeliet, Hens & Vermeir (ed.). Proceedings of the 2nd international conference on building physics 14–18 September 2003, Antwerpen, Belgium.

Technical Note AIVC 29, 1990. Fundamentals of the multizone air flow model – COMIS. International Energy Agency energy conservation in buildings and community systems programme. Air Infiltration and Ventilation Centre, Great Britain.

578

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

The effect of thermal resistance value to the external deterioration due to climate impact J. Šadauskiene, Dept of Building Materials, Kaunas University of Technology, Kaunas, Lithuania

V. Stankevicius & E. Monstvilas Institute of Architecture and Construction at Kaunas University of Technology, Kaunas, Lithuania

ABSTRACT: Thermal insulation from outside is very popular. The thin finish layers are used for finishes of the effective insulation most frequently. As it could be faced by experience, the thin render finish layer in certain cases is not as durable as expected. The weather durability of the finishes is in direct dependence to the climate parameters. But the additional damaging effect is created by the thermal resistance increase of the structure thus deriving into the whole system. At the analysis of the thermal behaviour of the insulated structures it has been found out, that the moisture volume in the outer layers of wall structure is accumulated more intensively in comparison with un-insulated one. The moisture in the insulation decreases the insulation effectiveness, causes the swell deformations and material’s mellowness. The thin-layered finish thus is worn out quickly and getting damaged. 1 FIELD OF APPLICATION The durability of an insulated wall’, external layers has been estimated according to their resistance to climatic impacts; other additional factors that increase the destroying effect have not been taken into account. The recommended calculation method allows for the prognosis of the impact of the factors of the enclosures’ external layers on the destructive climatic effects in a concrete geographical locality with typical air conditions. 2 TECHNICAL INFORMATION Main symbols and units d – thickness, m; φ – relative humidity of air, %. Sd – water vapour diffusionµ – water vapour resistance equivalent air layer thickness, m; factor,m/s λ – material thermal conductivity, ρ – density, kg/m3 ; W/(m·K); θ – temperature , ◦ C; ◦ The indoor air parameters θ, C/φ, %: 1. 18◦ C/50% 2. 18◦ C/70%

3. 4.

22◦ C/50% 22◦ C/70%

3 STRUCTURAL ASPECTS The calculation moisturous state of the enclosure’s have been carried out according to the EN ISO 13788. 579

_

+

_

+

1 2 4

1 2 3 4

a

b

Figure 1. General constructional wall schemes: a) – non-insulated wall; b) – insulated wall: 1 – interior render; 2 – construction of lightweight concrete blocks; 3 – mineral wool; 4 – thin mineral render. Table 1. The climatic parameters employed in the calculations of the enclosure’s moistures state. Month Climatic parameters

10

11

12

1

2

3

4

5

6

7

8

9

θ,◦ C φ, %:

7.1 86

1.8 88

−2.3 89

−5.2 87

−4.3 86

−0.4 79

5.8 76

12.4 71

15.8 71

16.9 75

16.4 80

11.9 83

Table 2. Physical parameters of the investigated constructions. d

ρ

λ

Type of layer

mm

kg/m3

W/(m·K)

1. Indoor render 2. Supporting part of the wall structure 3. Thermoinsulating layer 4. Thin render 5. Paint

20 240 100 10 0.1

1700 400 140 1700 –

0.9 0.22 0.04 0.9 1

µ

Sd m

6 3.6 1 6 –

0.12 0.86 0.1 0.06 0.3

The principal schemes of the investigated enclosures are presented in Figure 1. The climatic parameters employed in the calculations are presented in Table 1. The physical parameters of the layers of the enclosure are presented in Table 2. The values of the material’s physical parameters have been taken from the LST EN 12524. 4 RESEARCH ACTIVITY The constructions of the contemporary buildings are impermeable to air and water vapour, which determines their higher moisture (Madeleine, 1983). Higher moisture in an enclosure worsens the thermal properties of the wall, which decreases frost resistance and may end in the rise of the biological processes, such as rot fungus, etc. No doubt, it strongly affects the durability of the constructions (Fokin, 1973). The absence of the sufficient humidity balance in an enclosure (between moisture accumulation and drying), offers the possibility for the condensation processes to take place in it in a winter season (Janssens, 1998). When insulating the buildings from outside, the moisture and temperature strain takes place in an insulating material, which may affect the durability of the insulating material (Kunzel, 1999). 580

120 120 100

80

83

80

72

72

98

91 77 71

89 78

77

75

73

80 77

78 August

60

93

80

79

74 71

71 70

69 68

71

67

60

84 78

78 August

64 59

58

40

37

40 Lightweight concrete blocks

20 1

270

265

260

200

80

20

10

0

370

365

360

310

2 260

20

0

140

0

0

10

20

140

Relative air humidity, %

January

100 January

100

Relative air humidity, %

100

100

Thickness of the enclosure, mm

Thickness of the enclosure, mm

Figure 2. The distribution of the relative air humidity in the non-insulated wall when the temperature of the premises is 22◦ C and the relative humidity makes 70%.

Figure 3. The distribution of the relative air humidity in the insulated wall when the temperature of the premises is 22◦ C and the relative humidity makes 70%. 1 – the area of lightweight concrete blocks; 2 – the area of mineral wool.

One of the problems connected with the insulation systems is the incompatibility of the different layers according to the index of their vapour resistance. When water vapour permeability in the enclosure’s thermoinsulating layer is high and that of the enclosure’s external finishing layer is low, it allows for the assumption that the thermoinsulating layer’s drying conditions will be complicated (Ramanauskas, 2000). In designing the building constructions, it should be taken into consideration that the higher water vapour permeability of the wall structure’s thermoinsulating layer, the higher the index of water vapour permeability of the external finishing layer (Shala, 2003). External finishing should possess the following physical properties: low water saturation and high water vapour permeability ( Miniotaite, 2001). Attention has been paid to the finishing material’s vapour transfer coefficient in the case of the mineral wool usage for the insulation of the building enclosures from outside. It should not be less than 40 g/(m2 · d) (prEN 13500). 5 METHOD OF ANALYSIS The wall construction most frequently employed in practice has been chosen for investigation. Its constructional scheme is presented in Figure 1. Mineral wool has been used for the enclosure’s thermoinsulation since its water vapour permeability is high and the maximal moisture amount is accumulated in the intersection of the external layer and the mineral wool. The graphic analysis of the impact of the insulation from outside on the enclosure’s external layer is presented in Figures 2–11. 6 RESULTS OF ANALYSIS With the increase of the thermal resistance the dynamics of the moisture transfer in the whole enclosure grows (Fig. 2 and Fig. 3). During winter the condensation in the insulated enclosure takes place between the thermoinsulated layer and thin render. High water vapour permeability of the mineral wool determines abrupt moisture accumulation by the thin-layer finishing, whose water vapour permeability is lower. Here the external surface serves as a barrier for vapour diffusion from inside out. 581

Premise temperature and relative air humidity

Premise temperature and relative air humidity

182

22C/70%

0

22C/50%

0 18C/70%

0 18C/50%

0

50

100

150

136

22C/70%

0 22C/50%

0 18C/70% 0 18C/50%

0

200

20

40

2

Figure 4. The total moisture amount that might take place during moisture accumulation (i.e. condensation) in the insulated wall according to the calculation results achieved within constant environmental effects.

80

100

120

140

2

Figure 5. The total moisture amount that might take place during moisture accumulation (i.e. condensation) in the non-insulated wall according to the calculation results achieved within constant environmental effects. Mineral wool

0.1

Lightweight concrete blocks

0.09

0.18 0.16

0.08 Temperature gradient, °C/mm

Temperature gradient, °C/mm

60

Amount of moisture Wsum, g/m

Amount of moisture Wsum, g/m

0.07 0.06 0.05 0.04

Render

Render

0.03 0.02 0.01

0.14 0.12 0.1

Lightweight concrete blocks

0.08 0.06 0.04

Render

Render

0.02

0 1

2

3

4

5

6

7

8

0 12

34

56

78

Series1 0.0234 0.0234 0.0955 0.0955 0.0955 0.0955 0.0234 0.0234

Series1

Number of the enclosure's layers. At the bottom - the value of temperature gradient in the layer, °C/mm

0.008 0.008 0.0326 0.0326 0.1792 0.1792 0.008 0.008

Number of the enclosure's layer. At the bottom - the value of temperature gradient in the layer,°C/mm

Figure 6. The temperature gradient in the layer of non-insulated wall.

Figure 7. The temperature gradient in the layer of insulated wall. 120

120

January 100 100

Relative air humidity

84

83

80

100

72

70

78

75

72

69 68

100

100

81

81

Relative air humidity

January 100

80 77

August

60

85

80 74 71

74

68 64

60

67 60

67

100

100

82

81

84 78

August

66

57 49

40

40 20

Lightweight concrete blocks

12

20 0 0

0 0

10

20

80

140

200

260

270

10

20

140

260

310

360

370

370

Thickness of the enclosure, mm

270

Thickness of the enclosure, mm

Figure 9. The distribution of the relative air humidity in the insulated wall after the use of the paint films on its both sides when the temperature of the premises is 22◦ C and relative air humidity makes 70%. 1 – the area of lightweight concrete blocks; 2 – the area of mineral wool.

Figure 8. The distribution of the relative air humidity in the non-insulated wall after the use of the paint films on its both sides when the temperature of the premises is 22◦ C and relative air humidity makes 70%.

582

Premise temperature and relative air humidity

Premise temperature and relative air humidity

1190

22C/70%

369

22C/50%

780

18C/70%

217

18C/50%

0

200

400

600

800

Amount of moisture Wsum, g/m

1000

1200

1761

22C/70%

725

22C/50%

976

18C/70%

282

18C/50%

2

0

200

400

600

800 1000 1200 1400 1600 1800

Amount of moisture Wsum, g/m

Figure 10. The distribution of the relative air humidity in the non-insulated enclosure after the use of the paint films on its both sides, when the paint films have been estimated according to the calculation results achieved at constant environmental impacts.

2

Figure 11. The distribution of the relative air humidity in the insulated enclosure after the use of the paint films on its both sides, when the paint films have been estimated according to the calculation results achieved at constant environmental impacts.

In the period of moisture accumulation, when the wall was insulated from outside, the result achieved was higher accumulation of moisture than in the case of the non-insulated wall (Fig. 4 and Fig. 5). The process of condensation takes place when the air in the premises is warm and humid. After the insulation of the wall structure the temperature distribution has changed. The temperature gradient of the thin-layer finishing when affected by cold winter air decreases, i.e. freezing approaches volumetric freezing (Fig. 6 and Fig. 7). With the increase of the external layer’s vapour resistance the intensity of moisture accumulation grows. Air conditions in the premises determine the amount of accumulated moisture, which is larger in an insulated wall (Fig. 8 and Fig. 9). An both constructions, the condensation takes place between thin render and a paint film (Fig. 10 and Fig. 11).

7 FURTHER DEVELOPMENTS When restoring the insulated buildings, the external thin-layer finishing is repainted. The former paint film is not removed, since the very removal turns out to be problematic (render is thin and mechanically easily destroyed). Therefore each repainting of the thin-layer finishing is related with: the increase of the number of the paint films; the increase of the external surface’s vapour resistance, the growth of the condensation intensity beneath the paint films during winter season, and the decrease of the durability of the render-paint system.The calculations are provided according to the method described in the paper with respect to producer’s recommendations of the hydroprotection materials. The results are presented in Table 1. REFERENCES Rousseau Medeleine Z., 1983. Control of Surface and Concealed Condensation. Building Science Insight. Canada. Fokin K.F., 1973. Constructional Thermotechnique of the Building. Stroitizdat. Moscow. (in Russian) Janssens, A. & Hens H., 1998. Condensation Risk Assessment. Thermal Performance of the Exterior Envelopes of Buildings VII, December. Florida. Kunzel H.M. & Holm A., 1999. Combined Effect of Temperature and Humidity on the Deterioration Process of Insulation Materials in ETICs. Building Physics in the Nordic Countries, August. Gothenburg.

583

Ramanauskas J. & Stankeviˇcius V., 2000. Weather durability of thermal insulation system in external. Technologija. Kaunas. (in Lithuanian) Shala Y. & Makhatka M. 2003. Thermoinsulation of the Building Facades.. Obkos. Prague. (in Lithuanian) Miniotaite R. & Stankeviˇcius V., 2001. The Durability of the Painted Surfaces of the Building Walls. Technologija. Kaunas. (in Lithuanian). PrEN 13500. Thermal Insulation Products for Buildings – External Thermal Insulation Composite Systems (ETICS) Based on Mineral Wool – Specification. 2002-10 Rousseau Medeleine Z., 1983. Control of Surface and Concealed Condensation. Building Science Insight. Canada.

584

Session 10: Robustness 2

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Retrofitting of complex wooden structures by means of mixed reversible technologies: a study case B. Faggiano, A. Marzo & F.M. Mazzolani Dept. of Structural Analysis and Design, University of Naples “Federico II”, Naples, Italy

ABSTRACT: The paper deals with the analysis of the behaviour of complex structures made of ancient wood. With reference to a study case, which is the roofing structures of the Diplomatic Hall of the Royal Palace of Naples, the detailed geometrical and mechanical surveys are shown. On the basis of the results of the structural analysis, by means of a 3D model, the safety checks, according to the EC5 provisions, have identified the capacity of the structure in terms of strength and deformation and then the necessity of a structural retrofitting. Details of the proposed restoration intervention, based on the mixed technologies, are presented.

1 INTRODUCTION In general, the structural and architectural restoration of historical buildings has the aim of studying, interpreting, preserving, rehabilitating safety and service conditions. The intent to restore the bearing capacities of the structure is justified by the need to impede the worsening of degradation leading to collapse and/or to upgrade the architectural structure according to the current safety criteria and/or to change destination of use, what implies the increase of loading conditions as respect to the design ones. In this context, the restoration of constructions made of wood consists in the total or partial substitution of deteriorated elements or in reinforcing interventions. It can be carried out with two different aims: (1) improvement and rehabilitation of the existing structures by preserving the original structural configuration; (2) renovation by changing the original structural configuration due to safety lacks or change of destination or upgrading to current safety codes. The line of demarcation of such types of operation is defined by the acceptability of the existent structure. Actually, the analysis of ancient structures made of wood is cumbersome for some difficulties to be faced for both the material and the structural behaviour characterization. In fact, structures in ancient wood are subjected to imperfections, related on one hand to the material nature (damage for material shrinkage and ring shake due to aging and deterioration, and to degradation due to insect and mushroom attacks), on the other hand to irregularities due to the past technologies. Therefore the study of the existing wooden structures preliminary requires knowledge of materials to the evaluation of the deterioration degree by means of specific procedures and methods. To this purpose, it is compulsory to perform an in-depth survey of the structure and the relevant degradation of construction materials, of the environmental conditions and finally of the service conditions, which the structure has undergone during its life. Thus, it is needed to identify the static function of each wooden element, to characterize the joints between structural members and the restrain conditions, the latter being a topical concern of the wooden structures. Once all the necessary knowledge is achieved, it is possible to evaluate the bearing capacity of the single structural element and its contribution to the behaviour of the whole structure. Finally, the appropriate restoration intervention should be designed. In this context, the paper faces the refurbishment of complex structures made of ancient wood. The attention is focused on a study case, which is the roofing structures of the Diplomatic Hall of the Royal Palace of Naples (Italy). The geometric and mechanical surveys, together with the modelling aspects and the results of the structural analysis were presented in Mazzolani et al., 2004a. The 587

(a)

(b)

(c) C B Particular of fresco A Scaffolding

Diplomatic Hall

Propped up system

Lamp

Figure 1. (a) The Diplomatic Hall of the historical royal apartment of the Royal Palace of Naples; (b) A structural section; (c) Particulars of the internal Diplomatic Hall at present.

analysis of the structural behaviour evidences the weaknesses of the structure in terms of resistance and deformation capacities. The safety checks are performed according to the European provisions, Eurocode 5 (CEN, prEN 1995-1-1 2002). In this framework the paper illustrates the restoration interventions, conceived for upgrading and retrofitting the wooden structure. It is based on the use of the so called mixed technologies, consisting in the combination of techniques, which use different materials (in the specific case new wood, steel, reinforced concrete, rubber) for realizing local strengthening systems of the existent structures. The prerogatives of such retrofitting technologies are the reversibility of materials, the lightness, the ductility, the easiness of supply, transportation and erection. 2 THE STUDY CASE 2.1 General The wooden structure under study is the roof of the Diplomatic Hall of the Royal Palace of Naples (Fig. 1), which is datable to middle XVIII century. The roofing structure of the hall is composed by a complex floor slab-vault structural system, which was typical in important buildings at that time: the false ceiling was usually erected as a wooden vault, at the intrados of which an internal coat of thin canes and a cover of stucco is the base of the paint; the vault leans on the perimeter masonry walls and it is connected to the floor above by means of wooden links, so that the vault is partially suspended to the floor above. In particular, the complex wooden structure is composed by 3 main sub-structures (Fig. 2): A. The vault (Fig. 2A) It is made of wooden grid elements shaped as a reversal boat keel, which constitutes the ceiling of the Diplomatic Hall. The vault structure is composed by a grid of ribs and splines. Ribs simply about to the perimeter masonry walls, whereas they are nailed at the ends at the vault crown. Splines are orthogonal to ribs and fixed to them by a half-lap joint. The horizontal part at the vault crown is realized by a rectangular grid of wooden elements with circular section, which are nailed one each other at the intersection point and enclosed in a perimeter rectangle, whose elements are nailed one each other and to the vault ribs. The vault is supported along all the perimeter by the vertical masonry and it is connected to the upper floor beam by wooden links, which are nailed at both ends, upside to the primary or secondary floor beams and downside to the ribs. At the intrados of the vault a coat of thin canes, lime and plaster, as an overlay of the wooden structure, and the cover of stucco, as the base of the fresco, were realized. B. The beam floor (Fig. 2B) The primary frame of wooden beams is arranged along the minor span. Each beam is composed by two wood stocks placed side by side at the mid-span, connected one each other by nails. Only one large beam is shaped with a rectangular cross-section, being a single member along the whole span. The floor beam is stiffened by means of three different systems: (1) longitudinal secondary beams set under the primary beams, parallel to them and connected to longitudinal 588

(a)

(b) B 16.80

primary beam

B

C A

tie beam of the truss transversal struts

vault crown C A

floor sec. beams

14.20

16.80

A

principal façade

splines

14.20

A

B ribs Y X

tranv. sec. beam

(c)

B

single large beam

(d) king post secondary tie beam queen post metallic plate ties strut struts metallic stirrups primary tie beam Z X

(e)

lamp chain

(f)

Figure 2. (a) Vault; (b) Beam floor; (c) Truss (section C-C); (d) Section A-A; (e) Section B-B.

inclined struts, forming a supporting portal frame, which the wooden links of the vault are connected to; they are placed at all beams, with the exception of three alignments of beams near the perimeter masonries; (2) two transversal secondary beams, orthogonal to the primary beams; (3) a system of three series of four transversal inclined struts arranged in planes perpendicular to the primary beam axes, located at their central part near the transversal secondary beams and the tie beam of the truss. All the wood elements are connected by nails. The floor slab is composed by planks with a semi-circular cross section, over which a layer of lapilli and cement lime mortar is cast. By in situ measurements, the layer thickness varies from about 10cm at the support to masonry to 40cm in the middle span. C. The truss (Fig. 2C) The truss is close to the middle span of the floor slab, orthogonal to the primary floor beams. It is 7.00 m height. Its struts are composed by different wood stocks’ parts, placed side by side and nailed one each other, with a variable cross section along the element axis. The tie beam 589

has a circular cross section, it is connected to some floor beams by means of metallic U shaped stirrups. The structure is composed by two queen posts with a composite cross section, whose parts are connected by nails and metallic bindings, two orders of struts are connected to the queen posts, a central king post with one order of connected struts. Two heavy chandeliers of the Diplomatic Hall are hung to the tie beam. In view of the retrofitting intervention, the vault has been propped up by provisional supports.

2.2 The material mechanical properties According to the constructional practice of that time (Clean ed., 1996), all the wooden structural elements of the beam floor and the truss beam are made of chestnut, whereas the wooden structure of the vault is made by poplar, which is lighter. For the new wood of chestnut and poplar, according to the technical literature (Giordano, 1989) mechanical properties are indicated in Table 1. In the case of ancient wood, a 25% reduction of the mechanical properties as respect to the new wood was applied, for considering defects and degradations, together with working lacks, as it was obtained by experimental investigations (Calderoni et al., 2002; Mazzolani et al., 2004). Concerning the metallic plate ties of the truss, the iron with a tensile strength equal to 320MPa was assumed (Breymann, 1889), according to the strength properties of the material used at the time of the truss erection.

2.3 The structural model A 3D model of the structure was implemented for using in the program of structural calculation SAP2000 vers.7.12 (Wilson, 1998). The modelling assumptions are specified in Mazzolani et al., 2004. Due to the high level of variability and irregularity, which is peculiar of the ancient wooden structures, the geometrical modelling is necessarily affected by some approximation, always on the safe side. Figure 3 shows the extruded model at the truss beam location, in both XZ and YZ planes. Table 1. Mechanical properties of the wood. Wood

γ kN/m3

E0 kN/m2 × 10−3

E90 kN/m2 × 10−3

Nominal mechanical properties of new wood Chestnut 5.80 11380 544 Poplar 3.40 7850 376

ν0

ν90

G0 kN/m2 × 10−3

G90 kN/m2 × 10−3

0.37 0.38

0.46 0.47

4153 2844

185 128

3115 2133

139 96

Predicted mechanical properties of ancient wood Chestnut 8535 408 Poplar 5887 282

XZ plane

YZ plane

Vault layout

Figure 3. Extruded structural models.

590

3D-Vault

Beam floor

2.4 The load analysis Load combinations was applied for the Ultimate and the Serviceability Limit States safety checks, according to Eurocode 5. Concerning the load analysis, for the vault, the dead load is due to the contribution of the lathing, the stucco and the plaster, corresponding to 1 kN/m2 , which is applied as a distribution of forces concentrated at the grid nodes, equal to F1 = 0.45 kN and F2 = 0.35 kN. For the beam floor, dead load is Gk = 7.0 kN/m2 , live load is Qk = 2.0 kN/m2 , considering that the above rooms are for residential use. In order to go along the stress and strain states during the service life of the structural complex from the beginning until today, different load conditions, which the structure is supposed to have undergone, starting from the successive erection stages until the service load condition, were analyzed, according to the identified following phases and corresponding structural models (Mazzolani et al., 2004a): Phase 1. Phase 2. Phase 3. Phase 4. Phase 5. Phase 6.

Erection of the floor structure (Model 1); Erection of the truss (Model 2); Erection of the vault (Model 3); Floor structure completion (Model 4); Connection vault-floor beam and vault completion (Model 5); Service conditions (Model 6).

3 THE RESULTS OF NUMERICAL ANALYSES AND THE SAFETY CHECKS 3.1 Evaluation of the deflection state For estimating the overall deflection, deriving from the complete erection of the wooden structure, the material is modelled as new wood, this assumption being reasonable at the time of the original erection (Phases 1–6, Table 1). Three effects were considered, aiming at evaluating the actual deflection state of the whole wooden structure, according to EC5, for the worst climatic condition: (1) the reduction of the elastic modulus due to degradation typical of ancient wood; (2) the creep, such as the material strain at constant load; (3) the moisture content of material. As reference parameters, vertical displacements (uz ) of the floor beam mid-sections are assumed. The results of the analyses show that beam deflections are below the safety limits when new wood is considered. On the contrary, the degradations and the time dependent effects of creep and moisture contents induce an increase of deflections, which for some beams is beyond the limits (uz,max = 0.09 m versus uz,min = 0.08 m). Moreover the beneficial effect of the truss induces a displacement reduction of about 50% in the connected floor beams. The deformation state of the vault, which determined the crack distribution requiring the restoration intervention of the fresco (Mazzolani et al., 2004a, Fig. 4), was evaluated. In particular, considering the effect of degradation due to ancient wood, creep and moisture, the vault undergoes a flattening, which consists of sagging at the centre, with a maximum displacement equal to 6 cm, and elevation near the masonry supports, with a maximum displacement equal to 1.54 cm. In Figure 4 together with the crack distribution, the model at a typical structural section, the bending moment distribution and the corresponding deformed configuration are drawn. 3.2 Evaluation of the stress state The structural analysis was performed at each erection stage, so that the evolution of stresses from the beginning to the end of the installation can be followed. Aiming at the strength and stability checks, the influence on the material strength of the load application duration and of the wood moisture was accounted for according to EC5, by a reduction factor, which has been assumed as equal to about 0.5 (Mazzolani et al., 2004a). Thus, the design strength listed in Table 2 for the relevant stress conditions and wood species were used. Results emphasized that some ribs of the 591

(a)

(b)

(c)

(d)

Figure 4. (a) The model at a typical structural section, (b) the bending moment distribution, (c) the corresponding deformed configuration, (d) crack distribution at the vault intrados. Table 2. Design strength fd (kN/m2 × 10−3 ). Wood species Stress condition

Chestnut

Poplar

Compression // to the grain (fc,0,d ) Tension // to the grain (ft,0,d ) Bending (fm,d ) Shear (fv,d )

13.0 12.0 20.0 1.9

10.0 6.5 11.0 1.2

vault do not accomplish the strength requirements and all the struts of the floor beam stiffening systems do not satisfy the stability check. 4 THE ADOPTED RETROFITTING SYSTEMS 4.1 The intervention techniques Different retrofitting techniques are conceived according to the needs of each simple substructure. For the beam floor the following retrofitting interventions are planned: 1. Casting of a reinforced lightened concrete slab, which collaborates with the floor beams by means of a system of steel connectors purposely conceived with an “ad hoc” shape. 2. Reinforcement of the unstable longitudinal struts and transverse radial struts, by means of both horizontal and vertical stiffening elements. For the vault, the retrofitting intervention involves the vault-floor links only. In fact, from the examination of the internal actions’ distribution within the vault structural members, it appears that links between vault and floor beams located near the masonry supports are harmful, they working as struts, then exerting an additional bending action within the ribs. The removal of all these links placed at the vault curved sides induces a strong reduction of the internal actions. As a consequence, the retrofitting intervention consists in the removal of all the wooden links, replacing them only at the horizontal vault crown by means of steel ties. The r.c. slab is 20 cm thick, it is lightened by polystyrene blocks, 10 cm high (Fig. 5). The slab is reinforced by a net of steel rebars, 6 cm diameter and 20 cm mesh, disposed at the extrados. The concrete slab is cast on fir planks, 2 cm thick, with the function of fix formworks. The wooden 592

polyst. blocks

TRANSVERSAL SECTION net of steel rebars r.c. slab

LATERAL VIEW

TRANSVERSAL SECTION r.c. slab net of steel rebars

polyst. blocks

fir planks

LATERAL VIEW

fir planks

TRANSVERSAL SECTION r.c. slab net of steel rebars

LATERAL VIEW

fir planks

polyst. blocks

rubber rubber steel sleeve

(a)

rubber filler steel sleeve

rubber steel sleeve

Type A

Type B

Type C

connector

connector

connector

connector

steel pipe 50x50x4

steel plate 30x5

steel plate 30x5

steel pipe 50x50x4

steel plate 40x10

vault crown element

connector

connector

vault crown element

steel pipe 50x50x4

steel pipe 50x50x4

steel plate 40x10

steel pipe 50x50x4 steel plate 40x10

steel pipe 50x50x4

steel pipe 50x50x4

steel plate 40x10

Beam floor-vault connection

long. strut (b)

Types A1-B1-C1

fir planks

89

(c)

r.c. slab polystyrene blocks

long. strut

long. strut

Types A2-B2-C2

long. strut Types A3-C3

net of steel rebars

5

15

steel pipe 50X50X4

5

A

A1

A2

A3

B

B1

B2

C

C1

C2

Number 89

5

13

20

54

31

6

10

6

2

0

12

Type

80

80

M14

90

steel pipe 50X50X4

M14 80

160 steel plate 4X1

10

COLD FORMED STEEL SLEEVE

(d)

Figure 5. (a) Beam-slab connectors, (b) beam floor-vault connection and stiffening systems for longitudinal (c) radial and (d) struts.

planks should be protected from the water of the cast by a plastic film and in any case the water content of concrete should be limited. Wooden beams – r.c. slab connectors are realized by sleeves made of cold formed steel, which are composed by four (types A and C) or five (type B) parts (Fig. 5). They have perpendicular drilled wings at the ends and they are connected each other by means of bolted connections between the wings. The vertical superior wings serve as connectors and they are immersed in the concrete cast, it guaranteeing the transmission of the sliding actions. Therefore, steel sleeves have a twofold function: (1) they exert a transversal ringing action on the beams; (2) they realize the beam-concrete slab connection without the need to drill the beams. In particular, such connectors do not weaken the cross section and produce a beneficial effect of confinement. Due to the variability of the beams’ cross section, in order to avoid to make a number of different sleeve geometries, for the adaptation to the actual beams’ cross section, the system is conceived in only three different types and for each type the adaptation is optimized by the interposition of a layer of rubber, which is vulcanized to the steel sleeve in the workshop. The three different types of sleeve have the following characteristics (Fig. 5): A – Sleeve for beams with a single cross section; B – Sleeve for beams with multiple cross section (in the middle zone); C – Sleeve for the beam with a single rectangular cross section. 593

Figure 6. Connecting and stiffening systems: (a) plan layout of connectors at the floor beams; (b) section A-A; (c) section B-B.

Each type of connector is particularized in type 1, at the horizontal vault crown, and types 2 and 3 at the longitudinal struts. The B type connector is provided with a double layer of rubber, in order to improve the adhesion at the multiple cross section. In particular, the A1 (B1 , C1 ) system is formed by a A type sleeve for the beam-slab connection, a steel plate vertical tie for the vault-floor connection, a steel sleeve astride an element of the horizontal grid of the vault crown. The steel tie is bolted at its ends to the above mentioned sleeves. A2 (B2 , C2 ) and A3 (C3 ) systems are similar to A1 type, but they connect the floor beams to the longitudinal struts, in order to stiffen struts in the vertical plane. Furthermore, tubular horizontal steel profiles connect the vertical braces each other and to the perimeter masonries, in order to stiffen struts in the horizontal plane (Figures 5 and 6). The stiffening system for the transverse radial struts consists in steel square profiles, which are connected to steel sleeves located astride three of the four struts (Fig. 5). The same profile is used for stiffening struts in the horizontal plane. The total number of connecting systems to be used is 236, divided as indicated in Figure 5. The in plan layout of the connection elements is drawn in Figure 6. 4.2 Materials For the steel sleeves the following materials could be used: stainless steel, Fe360 hot galvanized steel, Fe360 hot painted steel, Fe360 hot galvanized and painted steel. Sleeves are tightened to the beams with the interposition of a layer of rubber. The assembly and the tightening of the sleeve parts is realized by means of 8.8 grade Ø14 bolts. For the slab the Rck200 reinforced concrete, 1800 kg/m3 weight is used. Fir planks as the support of the cast and polystyrene block for slab lightening are used. 594

Figure 7. Beam floor static scheme during the slab casting.

4.3 Design criteria for the retrofitting interventions The design of the floor structure is carried out by considering different loading conditions, they representing all the situations that the slab could undergo, from the construction to the service life: 1. Connectors and stiffening systems Sleeve connectors can be considered as ductile elements. The tighten of the sleeve should impede the sliding along the wooden beam. Aiming to this, the rubber layer should guarantee the friction resistance between steel and wood. Finally, all the floor beams are connected to the truss tie by means of metallic stirrups. 2. Concrete slab casting In the erection phase, before casting, wooden floor beams should resist all the dead load due to the beam itself, planks and live loads due to workers and tools. During casting, when the slab has no bearing capacity, it is necessary to analyse the effect of the r.c. cast load and live loads distribution, so that the four static schemes shown in Figure 7 should be analysed. All the ULS safety checks for the wooden beams are performed, according to EC5. Results show that even in the casting phase the stiffening struts buckle, so that it is necessary to install the stiffening supports planned for the intervention, already in this phase. In particular, for the longitudinal sloped struts two vertical and horizontal ties are needed in order to accomplish the stability check, by reducing the buckling length. 3. Completed structure After the connection of the vault to the floor beam, by means of ties at the vault crown, and the completion of the floor, it is possible to remove the provisional props of the vault. The service conditions should then be investigated. In particular, the collaboration of r.c. slab with the wooden beams is considered for resisting the applied loads, as a composite cross section. Live loads are applied in the worst distribution condition (Fig.7). The slab has been reinforced with a 7‰ steel area percentage, what allow to neglect the concrete cracking.

4.4 Procedure for the realization of the retrofitting system The retrofitting intervention consists of the following phases: a. b. c. d. e. f. g. h. i.

Removal of the vault-beam floor links; Demolition of the completion elements of the floor and of partition walls; Installation of sleeves for the beam-slab connection; Installation of connection between beams and the truss tie; Installation of the horizontal and vertical stiffening systems for the struts; Installation of fir planks and polystyrene blocks; Casting of the light r.c. slab; Erection of the completion elements above the floor; Restoration of the vault-floor beam connection by steel ties at the horizontal vault crown.

The necessity of disconnecting the vault from the beam floor before the demolition of the completion elements above the floor is due to the possible recovering of the elastic deformation in the beams, with a consequent elevation of the vault, which could cause damage of the fresco. In order to investigate such situation the numerical models, corresponding to the completed structure 595

Figure 8. Vertical displacement evolution during the retrofitting intervention phases.

in ancient wood (Model 5’) and the completed structure in ancient wood without beam floor completion elements (Model 7) are analysed (the numbering of models follows the one presented in section 2.4). Results show that the vault have an unacceptable lifting, being the maximum computed vertical displacement equal to about 5 cm at the vault crown (Fig. 8). 4.5 Analysis of the retrofitted structure The strength and stability checks, according to EC4 (CEN, prEN 1994-1-1 2001) and EC5 , for the concrete slab and the wooden structure, respectively, are satisfied. Figure 8 shows the deformed configurations of both the vault and the beam floor at each investigated phase of the retrofitting interventions, compared to the ones corresponding to the present structure state. It can be observed that, after the connection of all the beams to the truss tie, the beam floor assumes a continuous deformed shape (Phase d versus Phase b). The effect of the retrofitting intervention on the complete structure is apparent if Phase i and Phase 5’ are compared: after restoration, the maximum vertical displacement for the beam floor as respect to the undeformed configuration is about 2 cm, which is the half value than before the intervention; the vault has a light sagging of about 1cm as respect to the present configuration. When the service loads are applied (Phase j), there is an increment of deflection of the whole structure equal to about 0.5 cm, whereas the effect of the same loads before the restoration (Phase 6’) is quantifiable in about 1cm for the beam floor and 2.5 cm for the vault. In order to examine the deflection at infinite time, the viscous effects for both concrete and wood should be considered. For the concrete it is necessary to amplify deformation due to dead loads only, what is done by reducing the elastic modulus by three times (EC4). For the wood it is necessary to amplify by kdef equal to 2 the instantaneous deformation due to quasi-permanent actions (EC5). In these conditions, the increment of vertical displacement, as respect to the service conditions at the time of the restoration, is equal to about 0.8 cm for the whole structure. In total, the maximum deflection of the vault at infinite time is about 2.3 cm, which could be considered acceptable for the integrity of the fresco. 5 CONCLUSIVE REMARKS The paper deals with the refurbishment of complex structures made of ancient wood, with reference to a study case, which is the roofing structures of the Diplomatic Hall of the Royal Palace 596

of Naples (Italy). The geometric and mechanical surveys, together with the modelling aspects and the results of the structural analysis were presented in Mazzolani et al., 2004a. The analysis of the structural behaviour evidences the weaknesses of the structure in terms of resistance and deformation capacities. Consequently to the safety checks, according to Eurocode 5, a restoration intervention is conceived for upgrading and retrofitting the wooden structure. It is based on the use of the so called mixed technologies, consisting in the combination of techniques, which use different materials (in the specific case new wood, steel, reinforced concrete, rubber) for realizing local strengthening and stiffening systems of the existent structures. The main prerogatives of the proposed retrofitting system are the reversibility of the technology, the lightness, the ductility, the easiness of supply, transportation and erection. By illustrating the studied technical solution, its validity for the accomplishment of the design objectives is clearly evident.

ACKNOWLEDGEMENT Authors would like to acknowledge the Superintendence for Cultural Heritage of Naples (arch. Enrico Guglielmo), which has the authority on the activities developed inside the Royal Palace of Naples, and which entrusted the structural check for the retrofitting of the Diplomatic Hall. REFERENCES Breymann GA. Trattato generale di costruzioni civili con cenni speciali alle costruzioni grandiose, in III vol. Costruzioni metalliche (Costruzioni in ferro) (Metallic structures), Vallardi ed., Milan, Italy, 1889. Calderoni B., De Matteis G. and Mazzolani F.M. 2002. Structural performance of ancient wooden beams: experimental analysis, European timber buildings as an expression of technological and technical cultures, Editions Scientifique et Medicales, Elsevier S.A.S., pg. 217–233. CEN (European Communities for Standardisation), Final draft prEN 1994-1-1 2001. Eurocode 4: Design of composite steel and concrete structures – Part 1-1: General rules – General rules and rules for buildings. CEN (European Communities for Standardisation), Final draft prEN 1995-1-1 2002. Eurocode 5: Design of timber structures – Part 1-1: General rules – General rules and rules for buildings. CLEAN ed. 1996. Manuale del recupero delle antiche tecniche costruttive napoletane dal‘300 all’800 (Manual of the ancient constructive techniques in Naples since ‘300 to ‘800), Napoli. Giordano G. 1989. Tecnica delle Costruzioni in legno (Wooden structure engineering). HOEPLI ed., Milan, Italy. Mazzolani F.M., Faggiano B. and Marzo A. 2004a. Methodology for the analysis of complex historical wooden structures: a study case. Proceedings of the IV International Seminar “Structural Analysis of Historical Constructions (SAHC)”, 10–13 November, Padova, Italy. Mazzolani F.M., Calderoni B., De Matteis G. and Giubileo C. 2004b. Experimental analysis of ancient wooden beams for flexural and shear failure. Proceedings of the 4th International Seminar “Structural analysis of historical constructions”, November 10–13, Padova, Italy.

597

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Moment-resisting timber frames with densified and reinforced beam-to-column connections under seismic loads A. Heiduschke & P. Haller University of Technology, Dresden, Germany

B. Kasal North Carolina State University, USA

ABSTRACT: This paper describes results of the shake-table tests of glue-laminated frames with glass fiber reinforced connections. Typical design (control) was compared with frames that contained densified wood in connection regions reinforced with glass-fiber-epoxy composite. The connection zones used densified wood to increase moment and normal force capacities. Glass-fiber fabric was used in vertically laminated members to mitigate potentially brittle failures resulting from tensile strains across wood fibers. The tests showed superior performance of fiber-reinforced frames that behaved as self-correcting system with plastic deformations in beam-to-column connections and the elastic energy stored in the beams and columns. Analytical models of connections and frames were developed and compared with experiments.

1 INTRODUCTION Timber structures perform extremely well under earthquake loads due to their low mass density, high strength/mass ratio and ductile mechanical connectors. This is especially true for light-frame, lowrise buildings that contain large number of highly ductile nailed connections. Heavy timber frames must rely on a finite number of highly stressed connections that can fail in a brittle manner due to the tensile stresses perpendicular to fibers. Brittle failures are undesirable in earthquake situation since the buildings are permitted to be loaded well beyond pseudo elastic limit and connection failure may result in a catastrophic failure of the entire structure. Reinforcement of connections with composite materials has been researched for some time and most of the experiments dealt with static monotonic loads and simple splices (Haller & Chen 1999)(Haller & Wehsener 2001). Limited work studying cyclic loads was done and the results shown significantly improved capacity and ductility of such connections (Guan & Rodd 2003). Application of reinforced members (either locally or globally) in timber frames is rare and few examples exist. In this work, we have studied the performance of locally reinforced beam-to-column connections, column-to-foundation connections subjected to static cyclic loads and performance of frames subjected to dynamic (shake table tests) loads. We first built and tested scaled models of the connections and frames followed by full-scale tests. The results showed that locally reinforced frames can withstand large inertia forces with no significant damage. The timber frames behave as a self-correcting system capable of large elastic deformations while dissipating significant energy through elasto-plastic connections.

2 MATERIALS AND METHODS Schematic of the research is shown in Figure 1. The 1:4 scaled tests of connections were followed by full-size connections tests. The scaled tests of joints were used to optimize the joint 599

CYCLIC TESTS ON SINGLE CONNECTIONS

NUMERICAL SIMULATION

SHAKE TABLE TESTS ON FULL STRUCTURES

a(t), ϕ(t)

Scaled Frames (UD-R and D-R)

Scaled Corners (UD-UR, UD-R, D-R)

NUMERICAL ANALYSIS

-Quasi-static Cyclic

4 dowels

8 dowels

- Seismic - Sine-sweep - Free Vibration - Impact

F(t), u(t)

12 dowels

Full-size Corners (UD-UR, UD-R, PD-R, D-R)

-M–ϕ - Capacity - Stiffness - Diss. Energy - Damping

- Monotonic - Quasi-static Cyclic 8 dowels

u(t)

Full-size Frames (UD-UR and PD-R)

- Damping - Stiffness

- Seismic - Sine-sweep - Dwell Sine

- Nat. Frequency - Trans. Function - Storey Drift

Figure 1. Schematic of the research plan. Table 1. Testing matrix of the beam-to-column connections (Kasal et al. 2002).

Material (1)

Ultimate capacity Mmax (kNm)

Joint rotation ϕ at Mmax (rad)

Dissipated energy Eϕ at rotation ϕ (kNm*rad) Eϕ = 0.04 Eϕ = 0.2

12 dowels (Ø 3 mm)

UD-UR UD-R D-R

0.516 0.715 1.162

0.143 0.192 0.130

0.040 0.040 0.111

0.625 0.785 2.76

8 dowels (4 × Ø 4 mm and 4 × Ø 5 mm)

UD-UR UD-R D-R

0.518 0.767 1.477

0.173 0.192 0.154

0.044 0.049 0.114

0.58 1.03 3.00

4 dowels (Ø 5 mm)

UD-UR UD-R D-R

0.405 0.549 –

0.100 0.192 –

0.037 0.043 –

0.52 0.85 –

Number of dowels and diameter (Ø)

UD-UR = undensified-unreinforced, reinforced material.

UD-R=undensified-reinforced

and

D-R = densified-

configuration and to compare performance of several beam-to-column connections. To show the differences between reinforced and typical designed connections various specimen were subjected to cyclic loads. Basically 3 different designs were tested: (UD-UR) typical design - undensified and unreinforced joint, (UD-R) undensified and reinforced joint and (D-R) joint with densified material and fiber-reinforcement. Two sets of shake table experiments were performed: 1 : 4 scaled tests, and full-scale tests. The scaled tests of frames were carried out to study the frame behaviour and to develop the test strategy for the large frames. Further the shake table tests in reduced scale were used to verify analytical models that are equally valid at any scale providing that the constitutive and governing equations are equally applicable to both, model and full-size structure. Full-size shake table tests were performed at ENEL-HYDRO facility in Bergamo, Italy and results were used to verify and improve analytical models. 2.1 Beam-to-column connection tests Beam-to column connections are critical in transferring moments and normal forces in the frames. Since the system has little redundancy, failure of a single connection may results in the total loss of the structure. We have tested three different materials in three connection configurations. The testing matrix is shown in the Table 1. European Spruce (Picea Excelsa) with the average density 600

Figure 2. Connection detail and 3D view of a partially densified and reinforced full-size specimen.

of 0.44 g/cm3 was used for model tests. The densification process and its parameters is described elsewhere (Haller 2003). The material had initial average density of 0.44 g/cm3 and the final density after the densification was between 0.7 and 1.0 g/cm3 . Based on the 1:4 scaled test results, configuration 2 (8 dowels) was used as a final connection design. The beam-to-column connection tests followed the cyclic test protocol (DIN EN 12512 1996). Full-scale tests were done using the same test protocol as for the 1:4 connections. The connection architecture of a full-size specimen with a D-R joint design is shown in Figure 2. We have used two different dowel diameters. This was done to maintain the end, edge distances and spacing prescribed by DIN 1052 T2 (1988). We have used DIN rather than EC5 (1995) since the DIN permits smaller distances. The limiting dowel spacing for fiber-reinforced material is currently unknown, but one can reasonably expect that the connectors can be placed closer due to the mitigation of brittle tensile failure modes (Haller 2001)(Zeuggin 2003). 2.2 Shake table tests in reduced and full-scale 2.2.1 Small scale (1:4) shake table tests The scaled tests were performed at the Institute of Theoretical and Applied Mechanics, in Prague, Czech Republic and included the tests of the control, frame with fiber-reinforced joints and a frame using densified and reinforced material. In case of the densified material, entire members were densified. The scaling was done on a geometric basis (joint size) and higher order cross-sectional properties had different scaling factors. The goal of the scaled experiments was to generate the data for analytical models and study the potential failure modes. The frames were subjected to a series of arbitrary loading simulating earthquake load (IEEE 1987) at different magnitudes ranging from 0.1 g to 2.5 g. Sinusoidal sweep tests and impact tests were performed between each arbitrary loading to measure the change in natural frequencies. Arbitrary load tests were followed by a sinusoidal sweep test at the 1st natural frequency to induce large drifts. Accelerometers were placed at each floor and at the vicinity of connection. Rotations between beams and columns were also measured. 2.2.2 Full-scale shake table tests The full-scale tests were performed at the Earthquake Engineering Laboratory of ENEL.Hydro in Bergamo, Italy. Two frames were tested: 1. a typical design of horizontally laminated frame with no reinforcement and densification (control) and, 2. frame with partially densified and reinforced connection zones – see Figure 2. The dimension of the frame are shown in Figure 3a and the test setup and instrumentation in the Figure 3b. Standard equipment was used to measure accelerations and beam-to-column connection rotations. The tests included unidirectional and bi-directional excitations. The bi-directional experiments used the same signal for both directions with a phase angle between them. The value of the phase angle was randomly generated. The detailed description of the materials and test setup can be found in (Kasal et al. 2004). 601

Figure 3. (a) Drawing of the full-size frame and (b) shake-table tests setup and instrumentation (A = acceleromer, RD = displacement transducer). 1,0

1,0

0,8 D-R

Energy (%)

Energy (%)

0,8 0,6 0,4

UD-R

PD-R

0,6 0,4

UD-R

0,2

0,2 UD-UR

0,0 0

0,04

(a)

0,0

0,2

0

(b)

Rotation (rad)

0,04

0,13

Rotation (rad)

Figure 4. Comparison between energy dissipation capacity of scaled (a) and full-size (b) connections. Table 2. Comparison between undensified-reinforced (UD-R) and densified-reinforced (D-R) connections. Frame size

Capacity ratio

Yield rotation ratio

Initial stiffness ratio

Energy index

Scaled Full-size

1.9 2.0

0.78 0.73

3.4 2.05

3.3 1.75

Ratios are defined as the value of the parameter determined for treated connections (D-R) divided by the parameter determined for the control (UD-R).

3 RESULTS AND DISCUSSION 3.1 Beam-to-column connection tests Results from connection tests have been reported elsewhere (Kasal et al. 2002) and here we will summarize the main findings. The energy dissipation capacity of the beam-to-column connections is compared in Figure 4. From the Figure 4 follows that densification and reinforcement significantly increases the energy dissipation capacity of the connections. Because of the brittle failure mode in un-reinforced joints (UD-UR), little energy gets dissipated after splitting of wood. A comparison of scaled and full-size connections shows that the increase in stiffness and energy dissipation is qualitatively different – see Table 2. For the scaled and full-size connections, factors of 3.3 and 1.9 were determined – see Table 2. 602

Table 3. Results of shake table tests of scaled and full frames at arbitrary load of 2.5 g and 1.0 g. Excitation in the plane of the frames. 1

2

3

4

5

6

7

Design

Scale

Inter-storey drift

Max. rotation [deg]

PGA [g]

Max. accel. at 2nd (1st) floor [g]

Accel. at 2nd (1st) floor/PGA

UD-R UD-UR D-R

1:4 Full Full

1/35 1/45 1/70

1.60 1.24 0.78

2.5 1.0 1.0

5.6 (3.7) 3.0 (1.5) 3.0 (1.7)

2.3 (1.5) 3.0 (1.5) 3.0 (1.7)

Figure 5. Comparison between energy dissipation capacity of scaled and full-size connections.

3.2 Frame shake table tests The results of the frame shake table tests are compared in Table 3. The maximum inter-storey drift is defined as maximum relative amplitude at the floor level divided by the storey height. The results show, that the use of densified wood decreased the inter-storey drift up to 45%. In the column 6, the maximum measured accelerations for the 2nd and 1st floors are listed. Values in the column 7 represent the amplification (or response) factors, calculated by dividing column 6 by the peak ground acceleration (PGA). The maximum relative amplitude is calculated based on the maximum rotations in the connections. The relative deformation due to elasticity of the structural members was not measured. Based on the numerical simulation and experimental measurements, about 15% of deformation is due to elastic bending of beams and columns and about 85% due to the rotation of joints. The transfer functions for the scaled and full-size frames for 2.5 g respectively 1.0 g arbitrary loads are shown in Figure 5. From the transfer functions (TF) it follows that the natural frequency of the reinforced frame is higher that the one of the control. It is also evident that the reinforced frame will have higher inertia forces for the same arbitrary signal as compared with the control. This, however, cannot be generalized and depends on the frequency contents of the simulated ground motion. The arbitrary load tests were followed by the sinusoidal dwell test. The goal of the sinusoidal dwell test was to reach extreme amplitudes not achievable at 1.0 g simulated earthquake load. The maximum beam-to-column rotation was 1.24 deg and 0.78 deg for the control and reinforced frames, respectively. The rotations represent 69% and 12% of the connection rotation at maximum capacity. This indicates that even under relatively large inter-storey drifts (4% of the storey height), the joint rotations are small. However, sufficient energy is dissipated – experimentally determined equivalent viscous damping was 16%. Given the fact that the frames behave as self-correcting system (zero residual deformations after the sinusoidal dwell test), most of the energy must be dissipated in connections. The control frame exhibited crack development at the connections zones due to the tensile stress in the direction perpendicular to wood fibers. No connection deterioration was observed in the reinforced and densified connections. 603

0,03 Rotation [rad]

0,02

experiment nonlinear simulation

0,03

experiment linear simulation

0,02 0,01

0,01 0,00

0,00

-0,01

-0,01

-0,02

-0,02 -0,03

-0,03 10

12

(a)

14

16

18

10

20

(b)

Time (s)

12

14

16

18

20

Time (s)

Figure 6. Joint rotations (1st floor) for (a) linear and (b) nonlinear analysis compared to the experiment.

4 ANALYTICAL MODELS For the numerical analysis of the tested frames, a simple 2D model was used to simulate the frame response to the seismic excitation. The elements used for modeling the beams and columns had linear elastic material behavior. Two finite-element models were developed: (1) linear model with linear rotational springs representing beam-to-column connections, and (2) nonlinear model with nonlinear rotational springs. For the analysis with linear rotational springs, we have used 16% viscous damping. The damping ratio was obtained from low-level sweep-sine tests using the half-power bandwidth method. In Figure 6 the rotations of the analysis are compared with the experimental results. To simulate the hysteretic behavior of the beam to column connections a combination of nonlinear rotational springs was used. A bilinear spring with inelastic unloading hysteretic behavior represents the steel fasteners. A bilinear elasto-plastic spring element with gap was used to define the material behavior of the wood. Parallel arrangements of the bilinear elasto-plastic elements was used to simulate the joint behavior and the model simulated the experiment relatively well. 5 CONCLUSIONS Heavy timber laminated frames can perform well in earthquake-prone areas due to their low mass/strength ratio, high energy-dissipation capacity, high fire resistance and self-correcting nature. The ability of the system to return to its original (zero residual deformation) position offers the opportunity to use such systems that are required to retain the original function even after largemagnitude earthquakes. Design on beam-to-column connections is critical and potential brittle failures due to tensile stresses in the direction perpendicular to fibers are mitigated by composite fiber reinforcement. The stiffness and load-capacity of the frames may be limited by the capacity of connections and can be increased by using densified material in the connection areas. The experiments demonstrated high deformability of the frames where beams and columns behaved elastically and connections deformed plastically. While the typical-design frame developed cracks in the connection areas (perpendicular-to-fibers tension), no visible cracks developed in frames with reinforced and densified connections. Linear-elastic analytical model described the structural behavior with sufficient accuracy providing that the equivalent viscous damping is correctly estimated. If such estimate is not available a nonlinear analysis may be necessary. REFERENCES DIN 1052 T2. 1988. Timber Structures – Mechanical joints. Beuth Verlag. Berlin, Germany. EUROCODE 5. 1995. Design of timber structures, Part 1-1:General rules and rules for buildings, German version ENV 1995-1-1. Beuth Verlag, Berlin, Germany.

604

DIN EN 12512. 1996. Timber Structures – Test methods – Cyclic testing of joints made with mechanical fasteners; German version prEN 12512 : 1996, Beuth Verlag, Berlin, Germany. Guan, Z. & Rodd, P. 2003 Modeling of timber joints made with steel dowels and locally reinforced by DWW disks. Structural Engineering and Mechanics, Vol 16, No 4, October 2003: 391–404. Haller, P. & Chen, C. J. 1999. Textile reinforced timber joints and structures. In: Structural Engineering International, No. 4: 259–261. Haller, P. & Wehsener, J. 2001. Verstärkung von Holzverbindungen mit beanspruchungsgerechten textilen Strukturen. TU Dresden, Institut für Baukonstruktionen und Holzbau, Abschlussbericht zum AiFForschungsprojekt Nr. 14 ZBR ½. Haller, P. & Wehsener, J. 2003. Entwicklung innovativerVerbindungen aus Pressholz und Glasfaserarmierungen für den Ingenieurholzbau. Frauenhofer IRB-Verlag, Stuttgart. T3003. Institute of Electrical and Electronic Engineers (IEEE), Standard 344. 1987. Recommended Practice for Seismic Qualification of Class 1E Equipment for Nuclear Power Generating Stations. Kasal, B., Heiduschke, A. & Haller, P. 2002. Fiber-reinforced beam-to-column connections for seismic applications. Proc. of CIB W18 meeting. Kyoto, Japan. Universität Karlsruhe CIB W18/35-7-12. Kasal, I. Pospisil, I. Jirovsky, M. Drdacky, A.Heiduschke & P. Haller. 2004. Seismic performance of laminated timber frames with fiber reinforced joints. Earthquake Engineering and Dynamics. John Villey & Sons Ltd. London. UK. Vol. 33. (5): 633–646. Zeuggin, N. 2003. Temporäre Rautenfachwerkbrücke mit glasfaserarmierten Verbindungen. In: Bauen mit Holz 12/02.

605

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Experimental analysis on mechanical connections for ancient chestnut beams B. Calderoni, G. De Matteis, C. Giubileo & F.M. Mazzolani University of Naples “Federico II”, Department of Structural Analysis and design, Napoli, Italy

ABSTRACT: In this paper the results of experimental tests performed on defect-free ancient wooden specimens connected by means of metal fasteners are presented. The adopted typology of connection is commonly used in ancient floor slabs for long-span beams. In order to investigate on the evaluation of stiffness and bearing capacity of such a connection system, the performed experimental activity have been based on both embedding tests on wooden samples joined by steel bolts and four-point bending tests on wooden beams including the considered connecting system. All the wooden samples have been extracted from chestnut beams aged 1700–1800, withdrawn from an ancient building located in the historical centre of Naples (Italy). In the whole, the obtained results show that the use of such a connecting system is particularly uncertain for ancient wooden beams, since it could affect strongly the global behaviour of the relevant structural element.

1 INTRODUCTION Ancient wooden structures are highly vulnerable when applied in building located in seismic regions, due to both their intrinsic brittle behaviour and the significant degradation phenomena developing during their life. Therefore, it is often necessary to apply appropriate structural reinforcing interventions, based on a preventive evaluation of the bearing capacity of the existing structure. In particular, in this paper the behaviour of ancient wooden beams connected by means of metal fasteners is investigated. The adopted typology of connection was commonly used in ancient floor slabs for long-span beams. In such a case, load bearing beams are joined together by means of mechanical fasteners, for instance iron nails, placed along a very large overlapping zone. The connecting system has to provide adequate strength during the whole natural life of the structural element, despite the deterioration phenomena occurring in the wood. The contact between different material could give rise to high tensile stresses in the base material, especially when these stresses are not oriented in direction parallel to the grain. Then, it should be considered that usually adopted connecting systems for existing wooden structures are not related to specific and scientifically based methods. Therefore, it seems necessary to develop and adopt specific procedures that could be formally accepted, allowing the correct and reliable prediction of their structural behaviour. On the basis of the above considerations, an experimental investigation has been recently undertaken at the University of Naples Federico II. It concerns a series of tests on ancient defect-free chestnut beams, jointed by means of steel connectors. Specimens used in this activity have been extracted from existing beams, aged 1700–1800, withdrawn from a floor structures of a historical building (Calderoni et al., 2002). In the current paper the results obtained from embedding tests on wooden samples and from four-point bending tests on beam connected by means of steel fasteners are shown. The aim of the experimental campaign is to investigate the behaviour of such a connection system, in relation to both stiffness and bearing capacity. Hence, several specimen configurations have been considered, varying the number and the distance of steel connectors in addition to the width of the connected elements. 607

2 EMBEDDING TESTS Embedding tests have been carried out on a series of ancient wooden specimens in which a steel threat bar is inserted. In particular, tests have been performed considering wooden samples oriented both in the direction parallel (L-test) and perpendicular (T-test) to the grain. Four and two tests have been carried out for L and T configuration, respectively. According to the relevant part of the European regulation EN 383, the wood specimens have a 20 × 50 mm2 cross-section and 100 mm height, while the connecting element is a steel thread bar with a nominal diameter dn of 5.00 mm and inner core diameter dc of 3.96 mm. The samples have been loaded in force control by directly applying a compressive load to the two edges of the bar by means of a specific steel device, as shown in Figure 1. The displacement has been measured by a transducer (LVDT) with an accuracy of 1 × 10−3 mm. It was placed at the top of the loading device, whose displacement has been considered as representative of the joint displacement (w). A preliminary test was performed monotonically increasing the load up reaching a joint displacement equal to 5 mm. The corresponding strength of the system has been assumed as the maximum estimated force F,est , to be used for the subsequent tests. For all the tests, the adopted loading history is shown in Figure 2a. The load has been increased up to 0.4 F,est (point 04 in the diagram) and kept constant for about 30 sec (point 14). Then, the load is reduced up to 0.1 F,est (point 11) and then increased again up to the end of the test, conventionally assumed when the joint displacement becomes equal to 5 mm. The embedment strength, σemb,α at an angle α to the grain, has been calculated as:

Figure 1. Specimen shape and testing equipment for test in direction parallel and perpendicular to the grain. 1 0,9 0,8 0,7 0,6 0,5 0,4 0,3 0,2 0,1 0

F/Fest

04

14

time [sec] 11 0

120

240

360

480

600

720

(a)

(b)

(c)

Figure 2. Loading history adopted for embedding test (a). The specimen after a L-test (b) and a T-test (c).

608

where Fmax is maximum force withdrawn by the specimen in the whole range of jont displacement from 0 to 5 mm, d is the diameter of the metal fastener and t is the thickness of the sample. As an example, in Figures 2b and 2c, two specimens, tested one in direction parallel and one in direction perpendicular to the grain, are shown at the end of testing. In Figure 3, the obtained F-w curves are depicted for each test for both L and T configurations. Furthermore, in Figure 4 the σ -w curves are represented grouping the sample having the same configuration. In this case, the stress σ has been calculated considering an uniform stress distribution along the thickness t of the wooden sample and over the bolt diameter dc of the steel bar, as in equation (1). The curves obtained form both L-test and T-test show a similar shape, with a first practically linear branch characterized by a high value of the initial stiffness ki , followed by a second nearly linear branch characterized by a reduction of the stiffness kh , so that the experimental curve can be simulated by a simplified bi-linear relationship. In the case of the L-test, it can be noted that the experimental curves for sample EL 1, EL 2 and EL 4 are quite similar to each other up to a stress value of about 25–30 MPa, all exhibiting similar initial stiffness ki , ranging from 7200 to 8500 N/mm. On the contrary they become more different throughout the development of the second branch, the corresponding stiffness kh ranging from 1000 to 1500 N/mm. The sample EL 3 exhibited an initial stiffness ki significantly higher than the other ones (2700 N/mm), while the stiffness kh resulted again in accordance with the other samples (about 1000 N/mm). 5000

5000 F [N]

F [N] 4000

4000

3000

3000 Fmax (5 mm)

Fmax (5 mm) 2000

2000

EL1

EL2

1000

1000 displacement [mm]

displacement [mm] 0

0 0

2

4

6

8

2

0

10

5000

4

6

8

10

5000 F [N]

F [N]

4000

Fmax (5 mm)

4000

3000

3000 Fmax (5 mm)

2000

2000

EL4

EL3 1000

1000 displacement [mm]

displacement [mm] 0

0 0

2

4

6

8

2

0

10

4

6

8

10

5000

5000 Fmax (5 mm)

F [N]

4000

3000

3000

2000

Fmax (5 mm)

F [N]

4000

ET2

2000

ET1

1000

1000 displacement [mm]

displacement [mm] 0

0 0

2

4

6

8

10

Figure 3. F-w curves of L-test (EL ) and T-test (ET ).

609

0

2

4

6

8

10

The experimental curves for sample ET 1 and ET 2 are practically coincident to each other. They present a first nearly linear branch only up to a value of σ of about 10 MPa and a subsequent second linear branch characterized by a stiffness kh slightly higher than those obtained for L-tests. In Table 1, the experimental embedment strength, calculated for all the samples using both dn (σemb,α (dn )) and dc (σemb,α (dc )) values of the bar diameter, are reported. For L-tests, the mean value is equal to 35.1 MPa, considering dn , and 44.4 MPa, considering dc . This result is in a good agreement with the outcomes of some previously executed compression tests in direction parallel to the grain (Mazzolani et al., 2004b), where a peak stress value of about 45 MPa was obtained. The mean values of σemb,a (dn ) and σemb,α (dc ) for T-test specimens are equal to 24.4 MPa and 30.8 MPa, respectively. These values are about 30% lower than the ones of L-tests, while they are much higher than the ones obtained from compression tests executed throughout transversal direction, which resulted in mean of about 5 MPa (Mazzolani et al., 2004b). According to Eurocode 5, the value of the characteristic embedment strength for new hardwood and bolted connections up to 30 mm bolt-diameter, fh,α,k , at an angle α to the grain, can be computed as:

where: fh,0,k = 0,082 (1-0,01d)ρk is the characteristic embedment strength parallel to the grain k90 = 0,90 + 0,015d ρk is the characteristic timber density (kg/m3 ) In our cases, ρk was equal to 620 kg/m3 , obtained by an experimental measure of the density on the tested samples. Therefore, according to EC5, the characteristic embedment strength parallel to the grain fh,α,k (to be compared with the one obtained from L-tests – α = 0◦ ) corresponds to 48.8 MPa and 48.3 MPa, considering dn and dc respectively. This is in good accordance with the mean value of σemb,0 (dc ) obtained from tests (see Table 1). 60

60 σ[MPa] 50

40

40

30

w= 5 mm

σ[M Pa]

50

30 EL 1 w= 5 mm

20 10

ET 2

EL 3

10 w [mm]

EL 4

w [mm]

0 0

2

6

4

ET 1

20

EL 2

8

10

0 0

2

6

4 (b)

(a)

Figure 4. Comparison among σ -w curves: L-test (a) and T-test (b). Table 1. Main values of embedment strength and local stiffness. Specimen

Fmax (5mm) [N]

σemb,α (dn ) [MPa]

σemb,α (dc ) [MPa]

EL 1 EL 2 EL 3 EL 4 ET 1 ET 2

3005 4002 4500 3042 2436 2670

29.1 38.7 43.5 35.1 23.2 25.7

36.8 48.8 54.9 36.9 29.3 32.4

610

8

10

On the other hand the characteristic embedment strength perpendicular to the grain fh,α,k (to be compared with the one obtained from T-tests – α = 90◦ ) should be equal to 49.5 and 50.9 MPa, considering dn and dc respectively. This is significantly over-estimate if compared with the results obtained by tests and on the unsafe side too. 3 BENDING TESTS ON BEAM-TO-BEAM CONNECTION Bending tests on beam-to-beam connection have been carried out on a series of 5 defect-free ancient chestnut specimens working in the grain direction, connected to each other by steel thread bars, the same used for the embedding tests. According to the present European provisions for bending tests, a simply supported arrangement of the specimen, loaded by two symmetric concentrated forces (four-points bending test), has been adopted for testing, in order to submit the connecting zone to a constant bending moment (Figure 5). Table 2 reports the geometrical characteristics of the specimens, which differ for dimensions of cross section and number of connecting steel thread bars (Nf ). In all cases the connectors are distributed along a 12 cm connecting zone. In particular, sample BC 1, BC 4 and BC 5 have different width and equal Nf = 3, while sample BC 2 and BC 3 have the same dimension of BC 1 (cross-section 2 × 5 cm2 ) but Nf varying from 3 to 5. The bolt spacing is then variable and equal to 60, 40 and 30 mm for Nf = 3, 4 and 5 respectively. Tests have been performed under force control, by applying a quasi-static loading history. The imposed force has been progressively increased up to the collapse of the specimen. The load has been transferred to the beam by interposing a rigid steel profile between the loading jack and the specimen (Figure 5). The external supports were made of two steel half-cylinders. Deflections have been measured by means of displacement transducers (LVDT) placed at the edges of all the connecting elements and at the top of the loading device. In Figure 6, for all the samples, the experimental curves Acting Force (F) vs. B1-Displacement () are depicted, where the point B1 corresponds to the location of the first connecting bar, i.e. the one closest to the external support.

Figure 5. The loading scheme for bending tests. Table 2. Main dimensions for specimens tested in bending. Cross section

Span

Overlap zone

Number of

Bolt

Specimen

L [mm]

Lo [mm]

Width b [mm]

Depth h [mm]

connecting bars(Nf )

spacing [mm]

BC 1 BC 2 BC 3 BC 4 BC 5

580 580 580 580 580

200 200 200 200 200

20.5 20.7 21.0 3.04 3.95

50.6 50.4 50.7 50.7 50.7

3 4 5 3 3

60 40 30 60 60

611

6000

6000

F [N]

F [N] 5000

E13000 MPa

5000

E13000 MPa

4000

4000

3000

3000

BC2

2000

2000

BC1

1000

1000 displacement [mm]

displacement [mm]

0 0

2

4

6

8

0

10

6000

0

2

4

6

8

10

10000 F [N]

F [N]

E13000 MPa

5000

8000 E13000 MPa

4000 6000 3000 4000 2000

BC4

BC3 2000

1000

displacement [mm]

displacement [mm] 0

0 0

2

4

6

10

8

0

5

10

15

20

10000 F [N] E13000 MPa

8000 6000 4000

BC5

2000 displacement [mm] 0 0

5

10

15

20

Figure 6. F- curves for all tested specimens.

Only for comparison, in the same diagrams, the initial linear branch of the F- curve, corresponding to an ideal defect-free ancient chestnut beam having cross-section and length equal to those of the tested sample, are drawn. Such curves have been theoretically defined, adopting a longitudinal elastic modulus Elong equal to 13000 MPa, which has been obtained from a previous experimental campaign performed on defect-free ancient chestnut beams without internal connection (Mazzolani et al., 2004b). Note that from the same tests has been derived also a conventional elastic maximum stress of the basic material (σf ) equal to 43 MPa, so that the ideal curves have been stopped to the value of the force (Fult,id ) corresponding to the flexural resistance of the beam evaluated on the base of this stress. In Figures 7a and 7b, the experimental F- curves are grouped for specimens BC 1, BC 2 and BC 3 (having the same cross-section dimension), and for specimens BC 1, BC 4 and BC 5 (having the same number of fasteners), respectively. 612

5000

3000

F [N]

F [N] 2500

4000

2000 3000 BC1

1500

BC1

BC2 1000

BC4

2000

BC3

BC5 1000

500

displacement [mm]

displacement [mm] 0

0 0

2

4 (a)

6

0

8

5

10

15

20

25

(b)

Figure 7. F- curves for specimens BC 1, BC 2 and BC 3, having same cross-dimension and different N (a), and for specimens BC 1, BC 4 and BC 5, having variable cross-section dimension and N = 3 (b). Table 3. Mean values of ultimate force and displacement for tested beams. Specimen

Fult [N ]

ult [mm]

Fult,id [N ]

S(B1) [N]

σc (dc ) [MPa]

BC 1 BC 2 BC 3 BC 4 BC 5

2254 1936 2118 3320 3588

5.3 3.9 4.1 9.0 16.8

4560 4568 4689 6788 8820

1550 1198 1165 2282 2467

19.1 14.6 14.0 19.0 15.8

It can be noted that all samples exhibited an initial flexural stiffness significantly lower (about 40%) than that of the corresponding ideal beam without internal connection. This means that the connection leads to a reduction of stiffness, so that the connected beam is equivalent to an unique ∗ of about beam made of a conventional base material having a longitudinal elastic modulus Elong 8000 MPa. Specimens BC 1, BC 2 and BC 3, having the same cross-sections, showed experimental curves practically coincident, nearly linear up to the collapse, which occurred for a value of the acting force (Fult ) of about 2000 N (see Table 3), quite independent of the number of applied fasteners (Nf ). On the contrary, the F- curves for specimens BC 1, BC 4 and BC 5 show similar shape, but very different values of collapse load. Such a results is obviously due to the different cross-section of the samples of this group. Note that all the curves present a linear elastic behaviour up to about 1500 N. As reported in Table 3, the experimental collapse load forces (Fult ) for connected beams are significantly lower (up to 55%) than the ultimate load (Fult,id ) of the corresponding beams without internal connection, evaluated with reference to the conventional maximum elastic stress of the basic material (σf ). The typical exhibited collapse mode of the tested beams is shown in Figure 8. For all the tests, the failure occurred in correspondence of the connection. A crack started from the first external bolt B1 and propagated quite horizontally along the line of holes. Evidently the failure was due to an excess of tensile stress (σt,90 ) in direction perpendicular to the grain, before the attainment of the maximum embedment resistance, as determined from the corresponding tests. In fact, if a linear reaction force distribution among the steel bars is assumed, the generic reaction force in a bar S(Bi ) can be easily determined by means of the equilibrium equations with reference to the acting bending moment in the connecting zone. The maximum contact stress σc in the base material occurred in correspondence of bolt B1. It was calculated through the application of eq. (1), where Fmax is assumed equal to S(B1). As shown in Table 3, for all the samples σc ranges between 14 and 19 MPa, resulting largely lower than the main value of the embedment stress σemb,90 (dc ) in 613

Figure 8. Typical collapse mode.

transversal direction. On the other hand, such a contact stress value seems to be realistic in relation to maximum stress value for tensile failure throughout the grain transversal direction. Anyway, such outcomes should be verified on the basis of appropriate tests on beams without internal connections where the failure occurs due to excess of tensile stress throughout such a direction. 4 CONCLUSIONS The experimental campaign on defect-free ancient chestnut samples connected by means of steel bolts have provided interesting information on the local and global behavior of the tested specimens. In particular, embedment strength measured in direction parallel (σemb,0 ) and perpendicular (σemb,90 ) to the grain have been obtained. A mean value of 44.4 MPa has been calculated for σemb,0 considering dc bolt-diameter, which is in good agreement with both the ultimate compressive stress in the same direction and the characteristic embedment strength computed by EC5 for new hardwood. On the contrary, for σemb,90 a mean value of 30.8 MPa has been obtained, it resulting much lower than the one suggested by EC5. Bending tests on 5 connecting systems have been also performed, varying the number (Nf ) and the distance of steel connectors in addition to the thickness of connected elements. The tests show a remarkable loss of stiffness and bearing capacity in comparison with a whole beam having the same geometry. In fact, as an average value, a percentage reduction of longitudinal elastic modulus and ultimate strength of about 40% and 55%, respectively have been obtained. In all examined cases, the failure mechanism occurred in the wooden sample, throughout the longitudinal section in correspondence of the steel bars, due to an excess of tensile stress in the transversal direction before the attainment of a remarkable beam embedment. In order to provide a better understanding of the failure mechanism due to tensile stress in the direction perpendicular to grain, additional tests are necessary. Therefore, further research developments will be addressed to this aspect. REFERENCES Calderoni, B., De Matteis, G. & Mazzolani, F.M. 2002. Structural performance of ancient wooden beams: experimental analysis. In European Timber Buildings as an Expression of Technical Cultures. Editions scientifiques et medicales Elsevier SAS. EN 383/94 “Timber structures – Test methods – Determination of embedding strength and foundation values for dowel type fasteners”. Mazzolani, F.M., Calderoni, B., De Matteis, G. & Giubileo, C. 2004a. Experimental analysis of ancient wooden beams for flexural and shear failure. In Proceedings of the 4th International Seminar “Structural Analysis of Historical Constructions”. Padova, November 10–13. Mazzolani, F.M., Calderoni, B., De Matteis, G. & Giubileo, C. 2004b. Experimental analysis of ancient chestnut beams by small specimens. In Proceedings of the 4th international seminar “Structural Analysis of Historical Constructions”. Padova, November 10–13. Tampone, G. 2001. Acquaintance of the ancient timber structures. In Lourenco, P. B. & Roca, P. (ed.) Historical Constructions. Possibilities of Numerical and Experimental Techniques.

614

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Injection renovations of cracked joints A. Kudzys Vilnius Gediminas Technical University, Vilnius, Lithuania

R. Simkus KTU Institute of Architecture and Construction, Kaunas, Lithuania

ABSTRACT: Structural behaviour and response peculiarities of the moment-resisting RC frame systems subjected to horizontal extreme action effects caused by wind velocity pressures or earthquake motions are discussed. Results of experimental trials on the fragments of cast-in-situ RC multi-storey buildings with load-carrying walls and girderless floors as frame systems are presented. The cracking process, hysteresis loops, energy dissipation and resistance of primary and repaired joints are analysed. An effect of injections of epoxy resin grouts on the joint resistance and deformations is under consideration.

1 INTRODUCTION Brittle behaviour and significant damages in RC frame systems corroborate a necessity to perfect a performance of frame structures subjected to actions effects caused by wind velocity pressures and earthquake motions. The assessment of carrying capacity of renovated structures of existing buildings exposed to extreme horizontal forces is actual and urgent problem. It must be underlined that a renovation of damaged joints is rather difficult or impossible in many cases. Of special interest is the behaviour of floor-wall joints of frame system buildings due to lackage in physical experiment data. Besides, researchers and engineers do not attach great importance to investigations on the efficiency of renovation measures in rebuilding of cracked RC members and their joints. Under extreme lateral actions the joints of cast-in-situ RC moment-resisting frames from beams and columns or frame-type systems from plain (girderless) floors and walls receive tremondous stresses from adjacent members causing inelastic strains, bending and shear cracks. RC frames and frame systems subjected to these actions belong to the hysteretic structures with must be able to dissipate energy by means of ductile hysteretic behaviour of joints and horizontal members (beams or slabs). Vertical structural members (columns or walls) must remain elastic at all cases. Non-linear structural behaviour and resistance of cracked beam-column joints of RC frames are rather well investigated (ACI-ASCE Committee 352 1985, AIJ 1988, NZS 4203 1992, CEB 1996), but do not exist data and recommendations on the floor-wall joints of frame-type systems. Besides, researchers and civil engineers do not attach great importance to investigations on the efficiency of renovation and strengthening measures in rebuilding of cracked RC members and their joints of multi-storey buildings. This paper is devoted to present data and recommendations for building engineers on new type framless RC buildings and their floor-wall joint behaviour. It is dedicated to researchers and builders drawing their attention on structural renovations by resin injections and coating compounds. 615

2 TRIAL ON FLOOR-WALL JOINT FRAGMENTS 2.1 Arrangement and test of specimens The experimental research were carried out in order to investigate and ascertain the structural behaviour, failure mechanism and performance of primary and injection renovated interior and exterior joints of multi-storey RC buildings subjected to lateral out-plane seismic and other extreme actions. The test specimens (Fig. 1) presented full-scale fragments of multi-storey RC frame-type systems consisted from cast-in-situ walls and girderless floors. Thus, the specimens imitated model fragments of multi-storey uniplanar RC frames. Cast-in-situ walls and continuous multihollow floors with stay-in-place precast thin mould slabs corresponded to requirements of advantage construction technology in sustainable RC house- and office- buildings. The special test equipment which an actuator which simulated reiterated reversal lateral forces V was constructed, erected and used measuring at all loading steps strains of horizontal and vertical components of the fragments, shear deformations and horizontal displacements of joint concrete cores, cracking development and strains of reinforcing bars. A structural behaviour of floor-wall fragments were also evaluated rather exactly by the finite element models. Computer simulations of joint models were done in regard to the cracking and failure performance data obtained from the experimental tests. 2.2 Injection renovations of cracked specimens Resin grout injections and compound covers belong to rational repair practice for the renovation of structures of buildings, construction and civil engineering works [Kamaitis Z. & Kudzys A. 1982]. Therefore, they may be successfully applied for restoration of repairable members and joints of RC frames failed due to abnormal construction work or extreme live loads, storm wind velocity pressures, earthquake motions and other actions.

100

V

N

500

+

G

–

550

G

1 00

100

500

500

200

200

G

V

500

N

100

(b)

(a)

200 (c)

100

1000 (rad)

+0.0025

+2 +1

0 -0.0025 -0.0050

1000

1000

200

(rad)

+3

+0.0050

100

-1 -2 -3

+0.04 +0.03 +0.02 +0.01 0 -0.01 -0.02 -0.03 -0.04

100

11 9 7

10

8

3 4 56

Figure 1. Test fragments of exterior (a) and interior (b) joints, their loading program (c).

616

12

13

Good penetration, assured adhesive ability and low viscosity of resins or other binding agents allow to diminish the pressure of grout injections and at the same to simplify the renovation and strengthening technology. But the injection renovations can not be used for strengthening of structures of RC frames or frame-type systems which beams or floor slabs lost their initial positions and changed in shape due to inelastic strains of reinforcing steel bars. Means, materials and technology peculiarities for the beam-column or floor-wall joint renovations depend on concrete cracking and cover damage intensities. That’s why, the injection renovation methods and specific practical application features must be adapted to structural reconstruction conditions. But at all cases, an injection repairability of the cracked members and joints, their structural performance (strength and ductility) after the renovation may be checked and assessed objectively only according to experimental trial data. Therefore, after the first stage testing of the primary (uninjured) fragments the specimens were purposely returned to the initial position, repaired by injections of epoxy resin grouts into concrete cracks and replacing damaged concrete covers with epoxy compounds. The restored fragments were tested under the same loading and measurement program.

3 BEHAVIOUR OF PRIMARY AN REPAIRED JOINTS 3.1 Cracking process The floor-wall joint fragments of multi-storey RC buildings subjected to gravity and reiterated transient lateral (horizontal) forces were sufficiently ductile. Their behaviour, practically, did not differ from cracking and deformation peculiarities of rigid beam-column joints of uniplanar frame systems. In the interior primary (uninjured) specimens the initial bending cracks were observed at slab zones adjacent to the joint core. Large flacking area in the joint core concrete developed at the

(a)

(b) 2 1

1

wall

2

2

wall

2

1

1

Figure 2. Carked joints of primary (a) and repaired (b) specimens when cracks are caused by forward (1) and backward (2) leteral (shear) forces.

617

(a)

(b) V

V





(c)

(d) V

V





Figure 3. Hysteresis loops of angle δ versus shear force V of primary (a, c) and repaired (b, d) internal (a, b) and external (c, d) joints.

seventh loading cycle. After that injury, the shear deformations increased considerably and cracking process led to the joint core failure in shear at the thirteenth cycle. In the floor-wall joints of the repaired fragments the cracking process was identical to this of the primary structural members. But in the repaired slabs small bending cracks appeared from the first loading cycle. As the rule, new shear cracks in small quantities were observed from the third loading cycle. The shear cracks appeared in the centre of joint cores after the sixth loading cycle when joint displacement was quite significant. The maps of shear and bending cracking of joints before attaining ultimate limit state of the fragments are presented in Figure 2. It is not difficult to be sure that cracking processes and failure natures for the primary and repaired joints were identical. Therefore, the behaviour of primary and repaired joints may be assessed using the same analysis models. 3.2 Hysteresis loops and energie dissipation The maximum lateral forces Vmax of interior and exterior joint fragments corresponded the plastic inter-storey drift angle, respectively, δ = 2–3% and 1–2%. Dymiotis et al. (1999) ascertained that for multi-storey frames a failure of columns occur shortly after the drift angle δ = 3% is retained. Thus, with the reference on our test data the presumption that the value δ = 3% may be accepted as limit drift angle is in force both for the beam-column and floor-wall rigid joints. The behaviour of hysteresis loops of shear forces V and drift angles δ was analysed in consideration to the energy dissipation or loop area alteration ratio λed at each loading cycle. This ratio was calculated by the equation:

where Af and Ag are the areas of forward and backward cycle loops; A is the total area of rectangular envelope of cycle loops. Considerable differences were not observed in the shape of the hysteresis loops of inter-storey drift angles for the primary and repaired joint specimens with different failure natures (Fig. 3). The comparison of the hysteresis loop alteration ratio λed by (1) in the primary and repaired specimens led to the conclusion that epoxy resin injections may considerably change the energy 618

ed % 25 20 15 10 5 0

1

2

3

5

7

9

11

13

cycle

Figure 4. Energy dissipation ratio λed for primary (1) and repaired (2) specimen of internal frame joint.

dissipation characteristics (Fig. 4). At early loading cycles the ratio λed was small for the primary joints and rather great for the repaired ones due to possitive effect of epoxy resin grouts on steel bar and concrete bond resistance. The structural behaviour of tested fragment till 13th loading cycle may be explained by the ductility performance of the primary and repaired joints. It is to be rejoiced that non-linear concrete and steel reinforcement properties, inelastic bond slip versus shear stress relationship, joint core concrete cracking and horizontal force-displacement envelope curves of the joints for computer simulation sufficiently concurred with experimental trial results. 4 RESISTANCE OF PRIMARY AND REPAIRED JOINTS 4.1 Design mechanical model The inelastic deformations of frame system joints may be repeated by low-cycle earthquake motions or storm wind actions. In this case, the bearing capacity and stability of frame systems can be guaranteed only in an elastic behaviour of columns or walls when the resistance of joints is sufficient ignoring the presence of transverse reinforcement. At all cases, it is expedient to base complicated behaviour of the frame joints on the concepts of the ultimate strength and the action effects distribution on concrete core faces immediately adjacent to the critical sections of horizontal and vertical frame members. The design mechanical model for the resistance assessment of frame joints is presented in Figure 5. This model representing the diagonal compression conventional strut mechanism was suggested by Paulay et al. (1978) and adopted by AIJ (1988) and CEB (1996). It corresponds a loading cycle situation before residual strains in tension reinforcing bars of beams or slabs and full-depth diagonal cracking is reached. The resistance in compression of the concrete strut is closely connected with the concrete compressive strength and distances between stabilized inclined cracks. There are very few experimental investigations with should be sufficient for assertainment of the effect of joint mechanical and geometrical properties on cracking processes. According the recommendations of AIJ (1988) and Vasilev et al. (1970), the predicted resistance of frame joints may be calculated by the formula:

where fc is the concrete cylinder compressive strength; bj is the effective joint with; χ is the cracking factor which values are 0.35, 0.4 and 0.45 when the ratio of cross-section depths of horizontal and vertical members hh /hv is 1.5, 1.25 and 1.0, respectively. 619

Fu =R

hh zh

Fu =R zv hv

Figure 5. Design mechanical model and inner forces for interior joints of RC frames and frame-type systems. Rt/Rpr 1.1 1 0.9 0.8

S1

S2

S3

S4

S5

Figure 6. Ratio of test (Rt ) and predictive (Rpr ) strengths of primary (1) and repaired (2) specimens of internal frame joints.

4.2 Load-carrying capacity The ultimate load-carrying capacity of floor-wall interior and exterior joints were reached by their core concrete failure in shear. They were capable to withstand the predicted action effects caused by lateral forces. In the ultimate limit state of joints the tensile stress σs in mild reinforcing bars of floor slabs exceeded their yield strength fy from 20 to 75% and from 18 to 68% for the primary and repaired specimens, respectively. For the exterior joint specimens the ratio σs /fy ≈ 1.25 corroborated the ACI-ASCE Committee 352 (1985) directions. Stiffness characteristics of the floor-wall joints in all computer simulation models were in good agreement with the experimental data. The disposition of principle stress distributions in the joint core of RC frames concluded and corroborated the experimental trial data that the diagonal compression strut is predominant in the cracked joints. The test resistance Rt of core concrete struts of the joints was calculated using measured values of tensile, compressive and shear forces. The predictive resistance Rpr was calculated by the equation (2). Due to strengthening of the bond between concrete and reinforcing bars the load-carrying capacity of repaired joints was no less as in uninjured joints (Fig. 6). Therefore, the resistance of the primary and repaired joints of RC frame systems may be assessed using the same design method. The main concrete strut mechanism model well founded by experimental and computer simulation trials data is acceptable as the reliable approach in the assessment of resistance throughout the response of the primary and repaired joints of RC frame systems of multi-storey buildings. 620

5 CONCLUSIONS The resisting mechanism and design mechanical model for the primary (uninjured) and renovated cracked joints of RC moment-resisting frame-type systems of multi-storey buildings exposed to extreme climate and seismic actions should be presented and assessed by the main concrete strut model which is acceptable both for primary and renovated structures. The resin injections not only help to renovate fractured, disturbed and cracked zones of RC members and joints but they may also allow to increase effectively a bond between concrete and reinforcing steel bars, a rigidity, ductility, stiffness and strength of structures exposed to low-cycle bending moments and shear forces. The expedience to renovate the cracked members and joints of existing RC frames or frametype systems by the resin grout injections is corroborated by experimental trial data on structural behaviour of the primary and renovated frame fragments. The injection of bending and shear cracks by resin grouts and the covering of fractured and destroyed concrete surfaces by resin compounds are simple, rational and effective means for the renovation of RC structures being able to withstand unfavourable and dangerous actions. REFERENCES ACI-ASCE Committee 352. 1985. Recommendations for design of beam-column joints in monolithic reinforced concrete structures. Journal of the ACJ, V. 85, N◦ 3, 266–283. AIJ. 1988. Design guidelines for earthquake resistant reinforced concrete buildings based on ultimate strength concept Architectural Institute of Japan (in Japanese). CEB. 1996. RC frames under earthquake loading. State of art report. London: Thomas Teilford. Dymiotis, Ch., Kappos, A. J. & Chryssanthopoulos, M. K. 1999. Seismic reliability of RC frame with uncertain drift and member capacity. Journal of Structural Engineering, V. 125, N◦ 9, 1038–1047. Kamaitis Z. & Kudzys A. 1982. Rèparation du béton précontraint fissure par injections des polymers. The Ninth International Congress of the FIP, Stockholm. NZS 3101. 1992. Code of practice for design of concrete structures. Standard Association of New Zealand, Wellington. Paulay, T., Park, R. & Pristley, M. J. N. 1978. Reinforced concrete beam-column joints under seismic actions. Journal of the ACJ, V. 75, N◦ 11, 583–893. Vasilev, A. P. et al. 1970. Joint of precast reinforced concrete structures. Proc. of Articles. Moscow: Stroiizdat (in Russian).

621

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Seismic vulnerability assessment of RC structural walls M. Fischinger & P. Kante University of Ljubljana, Ljubljana, Slovenia

ABSTRACT: Mean return periods and mean 50-years probabilities of exceeding the estimated capacity of the buildings with RC structural walls have been calculated using simplified probabilistic approach based on total probability theorem (Cornell’s method). Maximum overall drift and maximum story drift, calculated indirectly from maximal tension and compression deformations at the bottom of the boundary columns of the wall, have been selected as damage measures. Effective 3D multiple-vertical-line-element model was used in inelastic response analysis. Model was calibrated by large-scale shaking table experiments. In general seismic vulnerability has been acceptable even for walls reinforced by minimum reinforcement. In addition to the height of the building and intensity of the seismic zone most important parameters influencing seismic vulnerability of buildings with RC walls are wall-to-floor ratio and confinement of the edges in compression.

1 INTRODUCTION Building systems with reinforced structural walls prevail in some central European countries housing millions of people. Such systems might not be designed for the appropriate level of seismic demand in the past and they were frequently reinforced with nominal minimum reinforcement. It is not clear what will be the behaviour of such walls in the case of exceptional loading or in the cases when overstrength is exhausted. Possible problems include compression failure or buckling of thin edges, large tension deformations due to the rocking of walls and related possibility of (brittle) fracture of the tension reinforcement. A number of research projects related to seismic assessment of RC structural walls have been initiated in Europe to support the development of Eurocodes. Two of them, in which University of Ljubljana actively participated, will be briefly described in Section 2. In parallel, an effective, yet relatively simple model for inelastic seismic analysis of RC structural walls has been developed and verified by experimental results (Section 3). The model was used in the inelastic response history analysis to construct “incremental dynamic analysis – IDA” curves as a part of the case study of Cornell’s methodology presented in Section 4. The case study is conceived to complement the general description of the methodology presented in the parallel data-sheet (Fischinger, 2004).

2 EXPERIMENTAL WORK Large-scale lightly reinforced RC cantilever structural walls (Figure 1) were studied in the frame of the CAMUS (Conception et Analyse des Murs sous Séisme) program on the shaking table at the CEA in France. Experiments and research were supported by international benchmark studies – blind analytical predictions (Fischinger et al, 2002a). Within the ECOLEADER shaking-table experimental program »Seismic Performance of Lightly Reinforced Structural Walls in Low to Moderate Seismicity Areas«, performed by the University of Ljubljana, L3S Grenoble and INSA Lyon using shaking table facilities at the LNEC, Lisbon, the research has been extended: (a) to investigate the influence of simultaneous loading in both 623

Figure 1. CAMUS test.

Figure 2. ECOLEADER test.

force

displacement

Figure 3. Bi-directional (3D) version of the multiple-vertical-line-element model (MVLEM) and vertical spring hysteretic properties.

horizontal directions, (b) to address walls with T (H) cross-section, (c) to investigate the free edge of a T (H) shaped walls in compression, (d) to investigate the behaviour of coupled walls, (e) to evaluate the behaviour of coupling beams in thin wall, and (f) to calibrate and further develop numerical models. Two 1:3 specimens (Figure 2) were loaded by a sequence of accelerograms with increasing intensity in both horizontal directions simultaneously. 3 ANALYTICAL MODELLING 3D multiple-vertical-line-element (MVLEM) has been developed to model inelastic seismic behaviour of RC walls (Figure 3). In the model, several vertical springs are connected by rigid slabs at the top and bottom level. They simulate axial and 2D flexural behaviour of the wall segment using simple hysteretic rules. Horizontal springs are modelling shear behaviour. The model successfully predicted inelastic response of the tested walls at various seismic intensities and was further refined after the experiments. In comparison to standard beam-column elements, MVLEM is able to model the shift of the neutral axis (related to the rocking of the wall), shear behaviour, local deformations of the reinforcement and confinement of the edges in compression. 624

L=15m

d=5m

Awall

h=0.2m 1%

ρ1=

Afloor

0.2% 1%

0.75m (0.15d)

Figure 4. Typical 5-storey structural wall with minimal reinforcement.

Awall

Afloor

Figure 5. Idealized floor plan of the RC wall building.

4 CORNELL’S APPROACH FOR PROBABILISTIC SEISMIC PERFORMANCE ASSESSMENT – A CASE STUDY FOR CANTILEVER STRUCTURAL WALLS 4.1 Basic concepts At Stanford University an approach for probabilistic seismic performance assessment based on total probability theorem was developed (Cornell & Krawinkler, 2000). This methodology was applied in the case study for cantilever walls representing idealised RC buildings. The mean return periods and mean 50-years probabilities of exceeding the estimated capacity of the structures were obtained as the final result. In the paper the shortened term for “the mean 50-years probability of exceeding the estimated capacity of the structure” is used: “50-years probability”. In general, the problem is characterized by three random elements: the ground motion intensity (IM – intensity measure), the demand on the structure and capacity of the structure (DM – damage measure). More detailed overview of the methodology has been explained within related COST publications (Fischinger et al, 2002b and Fischinger, 2004). 4.2 Reinforced concrete structural walls and seismic loading The case study refers to idealized RC buildings with structural walls. The area of the wall was kept constant (Awall = 1.0 m2 ; length: 5.0 m, thickness: 0.2 m). The varied parameters included: number of stories (n = 5, 10) and wall-to-floor ratio (ρ1 = 1%, 1.5%, 2%, 3%) – Figures 4 and 5. Axial force was calculated based on one half of the tributary area (considering the other half is carried by walls in the perpendicular direction). Longitudinal reinforcement of the RC walls was defined according to minimal demands of Eurocode 8 and to the old Slovenian seismic standards. The minimum percentage of the longitudinal wall reinforcement was 1% in boundary areas (Abc = 0.75 m × 0.2 m = 0.15 m2 ) and 0.2% in the rest of the wall cross section (Arest = Awall − 2Abc ). Accordingly, the total percentage of the longitudinal reinforcement in the wall was 0.44%. The confinement of the boundary columns of the wall has been also studied. The structural walls were modelled with in-plane multiple-vertical-line element model (MVLEM), which has been implemented into the up-to-date OpenSees framework (McKenna and Fenves, 2000). 625

1.0E-02 DGA 0.250g 0.225g 0.175g 0.100g

H(PGA)

1.0E-03

k0 1.703·10-5 1.210·10-5 1.212·10-5 5.226.10-6

k 3.8645 3.8159 3.3219 2.7735

1.0E-04

1.0E-05 0.0

Region with DGA=0.250g Region with DGA=0.225g Region with DGA=0.175g Region with DGA=0.100g 0.1

0.2

PGA [g]

0.3

0.4

0.5

0.6

0.7

0.8

Figure 6. Hazard curves for Slovenian seismic hazard regions (different design ground accelerations – DGAs) with idealization parameters.

Fourteen accelerograms were selected from the European Strong-Motion Database for nonlinear dynamic analyses (see Fischinger et al, 2002a, page 122). 4.3 Hazard curves In the presented study, peak ground acceleration (PGA) was adopted as intensity measure because seismic hazard curves were available in terms of this quantity. Hazard curves are presented in Figure 6. They represent the annual probability that the random acceleration of ag at the site will equal or exceed some specified level of ground acceleration. A standard deviation of the acceleration hazard has not been taken into account in the presented study. 4.4 Demand on the structure Maximum overall drift (GDR – global drift ratio) and maximum story drift (IDR – interstory drift ratio) have been selected as damage measures. Maximum overall and story drifts were calculated with nonlinear dynamic analyses for different accelerograms and different values of peak ground acceleration. After IDA curves, which link maximum overall (GDR) or story (IDR) drifts versus peak ground acceleration (PGA) had been determined, statistics of the drift demand given the intensity measure was obtained. Median drift demand and standard deviation of natural logs σDR were determined with assumption of lognormal distribution of maximum drifts. Finally, median drift demand was approximated (fitted median drift demand) with expression , where a and b are constants. Consequently the term median IDA (incremental dynamic analysis) curve represents median drift demand curve for the wall model of each structure (Figure 7). 4.5 Capacity of the structure Median drift capacity values for selected damage measures (GDR, IDR) have been calculated indirectly from maximal tension and compression deformations at the bottom of the boundary columns of the wall (Table 1). Consequently there were two capacity criteria used: the tension criterion (the maximal tension deformation εsu = 5.0%) and the compression criterion (three different maximal compression deformations: εcu = −0.4% for unconfined boundary column, εcu = −3.7% for confined boundary column in 5-story wall and εcu = −5.1% for confined boundary column in 10-story wall). Confinement of the wall boundary columns was calculated according EC8 (DC High). Capacity standard deviations of natural logs for randomness of the seismic action (σCR ) are included into Table 1. 626

0.6

PGA (g)

0.5 0.4 0.3 Median Drift Capacity with σCR Median Drift Demand Fitted Median Drift Demand Median Drift Demand ± σDR

0.2 0.1 ^

0.0 0.0%

C=0.695% (σCR =0.0019) 0.5%

1.0%

1.5%

Maximum Overall Drift (GDR)

Figure 7. Demand IDA curves of the structural wall (n = 5, ρ1 = 1.5%, min. reinf., conf. bound. columns).

Table 1. Median drift capacity values with standard deviations for both damage measures (GDR, IDR). Structural wall

GDR

σCR

Criteria

IDR

σCR

Criteria

n = 5, ρ1 = 1.0%, conf., As,min n = 5, ρ1 = 1.0%, unconf., As,min n = 5, ρ1 = 1.5%, conf., As,min n = 5, ρ1 = 1.5%, unconf., As,min n = 5, ρ1 = 2.0%, conf., As,min n = 5, ρ1 = 2.0%, unconf., As,min n = 5, ρ1 = 3.0%, conf., As,min n = 5, ρ1 = 3.0%, unconf., As,min

0.794% 0.343% 0.695% 0.409% 0.648% 0.463% 0.610% 0.584%

0.0040 0.0229 0.0019 0.0146 0.0005 0.0065 0.0002 0.0058

tens. comp. tens. comp. tens. comp. tens. comp.

0.752% 0.329% 0.653% 0.388% 0.581% 0.428% 0.538% 0.523%

0.0287 0.0417 0.0098 0.0720 0.0454 0.0328 0.0434 0.0422

tens. comp. tens. comp. tens. comp. tens. comp.

n = 10, ρ1 = 1.5%, conf., As,min n = 10, ρ1 = 3.0%, conf., As,min

0.807% 0.795%

0.0419 0.0318

tens. tens.

0.916% 0.804%

0.0303 0.0413

tens. tens.

Notes: conf./unconf. … confined/unconfined wall boundary columns; As,min … minimal longitudinal wall reinforcement.

4.6 Seismic performance assessment Hazard curve, demand curve and capacity value are coupled by the total probability theorem (Cornell’s approach). The result is the mean annual frequency of limit drift violation (mean annual probability of exceeding the estimated capacity of the structure) Pf , which can be determined according to Cornell & Krawinkler (2000) as

Mean return period (RP) and mean 50-years probability (Pf50 ) for each structure can be calculated by the expressions

627

Table 2. Mean return periods and mean 50-years probabilities for maximal overall drifts (GDR). Seismic region – DGA [g] p50 f

RP [years] Structural wall

0.250

0.225

0.175

0.100

0.250

0.225

0.175

0.100

n = 5, ρ1 = 1.0%, conf., As,min n = 5, ρ1 = 1.0%, unconf., As,min n = 5, ρ1 = 1.5%, conf., As,min n = 5, ρ1 = 1.5%, unconf., As,min n = 5, ρ1 = 2.0%, conf., As,min n = 5, ρ1 = 2.0%, unconf., As,min n = 5, ρ1 = 3.0%, conf., As,min n = 5, ρ1 = 3.0%, unconf., As,min

295 73 959 326 1699 947 3892 3457

448 112 1423 491 2501 1406 5671 5045

1024 295 2621 1054 4266 2599 8693 7858

4927 1677 10145 4810 15180 10095 27490 25288

15.62% 49.74% 5.08% 14.25% 2.90% 5.14% 1.28% 1.44%

10.58% 36.02% 3.45% 9.69% 1.98% 3.50% 0.88% 0.99%

4.77% 15.61% 1.89% 4.64% 1.17% 1.91% 0.57% 0.63%

1.01% 2.94% 0.49% 1.03% 0.33% 0.49% 0.18% 0.20%

n = 10, ρ1 = 1.5%, conf., As,min n = 10, ρ1 = 3.0%, conf., As,min

173 796

267 1193

724 2391

4051 9948

25.15% 6.09%

17.08% 4.11%

6.68% 2.07%

1.23% 0.50%

Mean return periods and mean 50-years probabilities for GDR for different structural walls in different Slovenian seismic hazard regions are presented in Table 2. 4.7 Conclusions In most cases predicted seismic vulnerability of the RC structural walls built in Central Europe has been acceptable. This is particularly true for all buildings in low seismic zone (DGA = 0.1g) and most 5-storey buildings. Seismic vulnerability may considerably increase for higher buildings (n = 10) in higher seismic zones (DGA >0.175 g). Important parameters influencing seismic vulnerability of buildings with RC walls are wall-to-floor ratio and confinement of the (thin) edges in compression. REFERENCES Cornell, C.A. & Krawinkler, H. 2000. Progress and Challenges in Seismic Performance Assessment. PEER Center News, 4(1): 1–3. Fischinger, M., Isakovic, T. & Kante, P. 2002a. Response of a Structural Wall – Blind Prediction and Calibration of the Model. Proceedings of the 7th U.S. National Conference on Earthquake Engineering (CD-ROM), EERI, Boston, USA. Paper no. 0314. Fischinger, M., Fajfar, P., Dolšek, M., Zamfirescu, D. & StratanA. 2002b. General methodologies for evaluating the structural performance under exceptional loadings. Structural Integrity under Exceptional Actions. Proceedings of the international seminar: Improvement of buildings’structural quality by new technologies, Lisbon, 19 and 20 April 2002. Luxembourg: Office for Official Publications of the European Communities, pp. 127–141. Fischinger, M. 2004. Performance based seismic engineering (PBSE). COST C12 – WG2, Data-sheet 1–1. McKenna, F & Fenves, G.L. 2000. An Object-oriented Software Design for Parallel Structural Analysis. AdvancedTechnology in Structural Engineering, Proceedings of the Structures Congress, ASCE,Washington D.C., USA.

628

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Ductile design of CBF steel structures A. De Luca, E. Mele & E. Grande Dep. of Structural Analysis and Design, University of Naples Federico II

ABSTRACT: The use of concentric braced frames (CBF) in seismic areas has been dissuaded by the past codes because of non-ductile behavior of this structural typology during large cyclic deformations in the post-buckling range. Only in the last twenty years this typology has been included in the class of ‘ductile structures’ by major codes, and the relevant code design provisions have been subjected to significant changes. In this paper the design procedure and the governing design parameters for CBF are examined with reference to European and Italian seismic codes; some case studies (3 and 6 storey CBF buildings) are designed, both in high and low seismic zone, and the effect of some design choices (i.e. number of CBF bays) is discussed and evaluated in terms of structural weight and design overstrength.

1 INTRODUCTION Concentric braced frames (CBFs) are very efficient structural systems for resisting lateral forces due to wind or earthquakes because they provide complete truss action. However, CBFs do not appear to be too common in seismic areas before the 1960s, even though their use has been dated back to the 1920s in non-seismic areas. The reason for this is related to the poor ductility and dissipation capacity of this framing system, mainly resulting from early cracking and fracture of bracing members or connections during large cyclic deformations in the post-buckling range. In fact, in past earthquakes, including the recent 1994 Northridge and 1995 Kobe events, a significant number of CBF structures suffered extensive damage, requiring extensive repair and upgrading work. Thanks to a great deal of experimental and theoretical research carried out starting from the late 70’s both in Japan and in USA, new design rules for ductile design of CBF have been introduced (and largely modified during the years) in seismic codes. This is also the case of Europe, and of Italy, where new seismic codes, recently released, also include provisions for ductile CBFs. However, these provisions seem to drive to a cumbersome design process and sometimes to practically impossible design solutions. In this paper the design procedure and the governing design parameters for CBF are examined with reference to European and Italian seismic codes; some case studies (3 and 6 storey CBF buildings) are designed, both in high and low seismic zone, and the effect of some design choices (i.e. number of CBF bays) is discussed and evaluated in terms of structural weight and design overstrength.

2 DESCRIPTION OF BUILDINGS AND DESIGN LOADS With reference to 3 and 6 storey steel buildings, characterized by the same plan dimensions (Figure 1a), ductile concentric braced frames (CBFs) have been designed according to Eurocode 8 provisions (EC8, May 1994 and January 2003 versions) and new Italian seismic code (OPCM 3274, March 2003). The buildings’ plan is rectangular, 54.8 × 36.6 m, with bay spans (at column 629

3,96 3,96 3,96

23,77

3,96 3,96

3,96

11,89 3,96 3,96 3,96

9,11

9,11

(a.)

(b.)

Figure 1. (a) Building plan; (b) Elevation of CBF bays. Table 1. Seismic weight and base shear of 3 and 6 storey buildings. Base shear: Vb (kN)

Vb /W [%]

LOW s.z.

LOW s.z.

HIGH s.z.

6514 6637 6914

9.4 9.6 10.0

21.9 22.3 23.2

12995 12272 12782

9.4 8.9 9.2

21.9 20.7 21.5

Seismic code

Seismic weight (kN) W = (Gk + ei Qk,I )

3 st.

EC8’94 EC8’03 OPCM’03

29775

2784 2856 2976

6 st.

EC8’94 EC8’03 OPCM’03

59391

5568 5260 5477

HIGH s.z.

centerline) of 9.144 m. The interstorey height is equal to 3.962 m, while the global height is 11.88 m and 23.77 m, for the 3 and 6 storey buildings, respectively. For the design of the buildings’ structure, different options concerning the number of single braced bays in each principal direction, have been considered: in particular, the cases of 8, 6 and 4 not-adjacent braced bays in each direction have been selected for both the 3 and the 6 storey buildings. In this way, 1/8, 1/6 and 1/4 of the building base shear has been used for the seismic design of the single braced frame; in order to obtain results which only reflect the effect of these seismic design choices and are not affected by the tributary gravity loads acting on beams of the braced bay, the floor has been oriented in the direction parallel to the braced bay. European wide flange sections (HE series) have been adopted for beams, columns and diagonal members. A single steel grade, Fe510 (fy = 355 MPa), has been used for all structural members of the 3 storey frames, while in the 6 storey frames the beams and columns are made of steel Fe510, and steel Fe360 (fy = 235 Mpa) has been used for the diagonal elements. The dead and live gravity loads for both buildings are: dead loads Gk = 4.166 kN/mq, for intermediate stories and Gk = 4.05 kN/mq for roof storey; live loads Qk = 2.00 kN/mq for all storeys. The design of the CBFs has been carried out both for low and high seismic zones, which respectively correspond to PGA values of 0.15 g and 0.35 g (both in EC8 and OPCM codes). Soil profile type B and behavior factor q equal to 4 (ductile CBFs), have been adopted in the definition of the design seismic action. In table 1 the seismic weight W, the base shear Vb and the ratio Vb /W are provided for the 3 and 6 storey buildings, both for low and high seismic zones, and with reference to the EC8’94, EC8’03 and OPCM’03. The static base shear has been evaluated in the design spectrum at the building fundamental period, estimated through the approximate formulae provided by the codes. 630

In the following, the different case studies, are identified by labels, accounting for: the number of storeys (3s, 6s), the number of braced bays (4b, 6b, 8b), and the seismic zone (L, H). Thus, with this notation, all case studies are: 3s4b-L, 3s6b-L, 3s8b-L (3 storeys; 4, 6 and 8 bays, respectively; Low seismicity) 3s4b-H, 3s6b-H, 3s8b-H (3 storeys; 4, 6 and 8 bays, respectively; High seismicity) 6s4b-L, 6s6b-L, 6s8b-L (6 storeys; 4, 6 and 8 bays, respectively; Low seismicity) 6s4b-H, 6s6b-H, 6s8b-H (6 storeys; 4, 6 and 8 bays, respectively; High seismicity) In the next paragraphs, the EC8 and OPCM provisions for ductile design of CBF are briefly reported and the resulting design procedure is outlined; then, the above CBF structures are designed, and the influence of code provisions on the design of ductile CBFs is evaluated and discussed in terms of structural weight and over-strength.

3 CODE PROVISIONS FOR DUCTILE CBF Design rules for the seismic design of CBF steel structures are specified in many standard codes with the purpose of guaranteeing a stable and ductile behavior. The fundamental design criterion adopted by EC8 and OPCM-03 code provisions, is that concentric braced frames shall be designed so that yielding of the diagonals in tension will take place before failure of the connections and before yielding or buckling of the beams or columns. For this purpose, many design indications are provided, in particular concerning limitations of the slenderness ratio of diagonal elements and capacity design criteria for beams and columns belonging to the braced system. Concerning the slenderness ratio, while in EC8-94 and OPCM-03 only an upper limit is specified (i.e.: λ ≤ 1.5), upper and lower limits are given in EC8-03 (i.e.: 1.3 < λ ≤ 2.0). In the case of three and six storey CBFs considered in this paper, the minimum and maximum sections (for the HE series) that can be adopted for diagonal members are: • 3 storey frames (Fe510; L0 =9.96 m):

• 6 storey frames (Fe360; L0 =9.96 m):

The sections adopted for diagonal members indirectly affect the choice of column and beam sections. In fact, in the light of the capacity design approach, the axial forces to be used for the design of beams and columns of the braced frames (here appointed as N∗) are computed by amplifying the values obtained from structural analysis, through the (minimum or maximum) over-strength factor of diagonal members:

where: Nsd,g is the axial force in beams and columns due to gravity loads, Nsd,o is the axial force in beams and columns due to seismic action; αmin , min and αmax are the over-strength factors of 631

diagonal members, respectively defined, in EC8’94, EC8’03 and OPCM’03, as:

Particularly, while in the case of EC8-94 and EC8-03 the minimum over-strength (αmin , min ) has to be considered as amplification factor, in the case of OPCM-03 the maximum one (αmax ) is considered. Moreover, according to Eurocode 8-03, in order to satisfy a homogeneous dissipative behaviour of the diagonals, it should be checked that the maximum overstrength i . . . does not differ from the minimum value  by more than 25%. With reference to a scheme with tension-only diagonal members, the first step in the design of CBF system is the evaluation of diagonal axial forces. Moreover, considering the codes slenderness ratio limits of diagonal members, minimum and maximum adoptable diagonal section can be evaluated. In this way, the choice of diagonal sections depends both upon axial stress level and slenderness ratio limitations. The choice of diagonal sections also affects the design of the columns and the beams. In fact, the design of these elements is based on the amplified axial forces evaluated according to equation (3). 4 DESIGN OF CBFs FOR THE 3 STOREY BUILDING In table 2 are reported the cross sections adopted for the column and the diagonal elements of the 3 storey CBFs, designed according to the above specified seismic codes. It can be observed that in the case of EC8-03 and OPCM-03 (table 2a), it was not possible to differentiate the member sections depending on both seismic zone and number of braced bays, while in the case of EC8-94 (table 2b) different sections could be adopted for low and high seismic zone and for the cases of 4, 6, 8 CBF solutions. The different outcome of the structural proportioning is mainly related to the differences in the formulae (3) which provide the amplified forces N∗ to be used for column design: both EC8’03 Table 2a. Storey building: sections of CBF structural elements (EC8’03, OPCM’03). EC8-03

OPCM-03

St.

Diagonal

Column

Diagonal

Column

1 2 3

HE220A HE200A HE160A

HE340B HE340B HE340B

HE220A HE220A HE220A

HE500M HE500M HE500M

Table 2b. 3 storey building: sections of CBF structural elements (EC8’94). EC8’94 3s4b-L

3s4b-H

3s6b-L

St. Diagonal Column Diagonal Column Diagonal Column 1 2 3

3s6b-H

3s8b-H

Diagonal Column Diagonal Column

HE220A HE550B HE260A HE450B HE220A HE550M HE220A HE400B HE220A HE500B HE220A HE550B HE260A HE450B HE220A HE550M HE220A HE400B HE220A HE500B HE220A HE550B HE220A HE450B HE220A HE550M HE220A HE400B HE220A HE500B

632

and OPCM’03, in fact, introduce the diagonal overstrength factor for amplifying only the seismic axial force (Nsd,o ); therefore, by changing the base shear acting on the single braced bay (due to differences or in number of braced bays either in seismic zone), the values of overstrength in diagonals and of the seismic axial forces in columns change in proportional manner: their product, and hence the value of N∗ , remains constant. On the contrary, in the EC8’94 formula (3), both gravity and seismic axial forces (Nsd,g and Nsd,o ) are amplified by the overstrength factor, thus, since Nsd,g does not change by changing the base shear, the value of N* does not remain constant. Concerning the effect of diagonal slenderness limitations, it can be remarked that, both in the case of EC8-94 and OPCM-03, for all the CBFs (except 3s8b-H), it has been adopted the minimum diagonal section (HE220A, which is derived from λ ≤ 1.5), even though a smaller section could be enough for the design axial forces. It is worth observing also that, in the case of EC8’94 (table 2b), the 3s8b-L structure has not been proportioned since it was not possible to select practical cross section for columns (of the HE series) that satisfied the capacity design provisions, i.e. with axial resistance larger than N*, as provided by the relevant equation (3). In fact, due to slenderness limitation, the minimum allowable diagonal section is greatly redundant (for the building with 8 braced bays and in low seismic zone), thus the low design stress level, and the consequent high over-strength factor, gives rise to very high value for N∗ , and the column sections should be greater than HE1000M. In figure 2 it is shown the comparison between the buckling resisting axial force Nbd of columns (shown in the charts with horizontal lines) and the amplified axial forces N∗ (shown in the charts with vertical bars), according to examined codes and with reference to designed frames. It is possible to note that the values of the column resisting forces, Nbd , are close to the amplified design forces (N∗ ) of the first storey columns. For the second and third storey, instead, the difference between resisting force and design force is greater, since the same column section has been adopted for all storeys. It is interesting to note that, according to EC8-94, the amplified axial column forces N* evaluated for the 3s6b-L (Low seismic zone) are larger than in the case 3s6b-H (High seismic zone). The reason for this apparent contradiction is a consequence of the major role that the overstrength of diagonal members (which is strictly related to the slenderness limitations) plays in the value of the amplified design force N∗ . In fact, by adopting the mimmum section for the diagonals (HE220A) according to the slenderness limitation, the diagonal stress level in low seismic zone, is significantly smaller than in high seismic zone, thus the overstrength factor, which amplifies the axial forces in columns, is significantly larger than in high seismic zone. This drives to column cross sections required for low seismic zone CBF (HE550M) larger than the ones for high seismic zone (HE400B).

5 DESIGN OF CBFs FOR THE 6 STOREY BUILDING As already mentioned, in the cases of the six storey CBFs, steel Fe360 (fy = 235 MPa) has been used for the diagonal members, while steel Fe510 (fy = 355 MPa) has been used for the column and beam elements. The choice of a different steel type, characterized by yield stress smaller than in the case of the 3 storey CBFs, has been necessary in order to reduce the over-strength factors of diagonals members. Despite of this choice, the design of the 6 storey CBFs has been practically possible only with reference to EC8-94 and EC8-03: according to OPCM-03 code, the amplified axial column forces to be considered for the design of the column cross sections were larger than the axial resistance of the largest section available in the HE series (i.e. HE1000M section). The case of 4 braced bays in each principal direction could not be designed in high seismic zone, neither with reference to EC8’94 and EC8’03. It is worth mentioning that, the building plan dimensions are quite large (nearly 2000 square meter), thus the global shear in high seismic zone for the 6 storey building (see table 1) cannot be resisted by a limited number of CBFs, which also satisfy the conditions (2) and (3). 633

10000

10000

3s8b-L

3s8b-H 8000

Nbd(HE500M) - OPCM/03

6000

N* (kN)

N* (kN)

8000

Nbd(HE500B) - EC8/94

4000

Nbd(HE500M) - OPCM/03

6000 4000

Nbd(HE340B) - EC8/03

Nbd(HE340B) - EC8/03

2000

2000 0

2

1

(a)

EC8-94

0

3 STOREY

EC8-03

1

2

(b)

OPCM-03

10000

EC8-03

10000

3s6b-L

3s6b-H

Nbd(HE550M) - EC8/94

8000

Nbd(HE500M) - OPCM/03

6000

N* (kN)

N* (kN)

8000

Nbd(HE400B) - EC8/94

4000

Nbd(HE340B) - EC8/03

2000

Nbd(HE500M) - OPCM/03

6000 Nbd(HE340B) - EC8/03

4000 2000

0

1

(c)

2

EC8-94

0

3 STOREY

EC8-03

1

(d)

OPCM-03

10000

10000

8000

Nbd(HE500M) - OPCM/03

N* (kN)

Nbd(HE450B) - EC8/94

4000

EC8-03

Nbd(HE340B) - EC8/03

2000

OPCM-03

3s4b-L

8000

6000

3 STOREY

2

EC8-94

3s4b-H

N* (kN)

3 STOREY OPCM-03

Nbd(HE500M) - OPCM/03

6000

Nbd(HE550B) - EC8/94

4000

Nbd(HE340B) - EC8/03

2000

0

1

(e)

2

EC8-94

0

3 STOREY

EC8-03

1

(f)

OPCM-03

2

EC8-94

3 STOREY

EC8-03

OPCM-03

Figure 2. 3 storey: comparison between axial resisting forces and amplified design forces for columns. Nbd (HE360M)

10000

Nbd (HE800B)

8000

Nbd (HE340M)

10000

Nbd (HE320M)

6s6b-L

EC8/94 Nbd (HE700B)

8000

Nbd (HE650B)

4000

6s6b-H

N* (KN)

N* (KN)

EC8/03 6000

Nbd (HE650B) Nbd (HE500B)

6000

Nbd(HE600B) Nbd (HE450B)

Nbd (HE550B) EC8/03 Nbd (HE400B) EC8/94

4000 2000

2000 0

1

(a)

2

3 EC8-94

4 EC8-03

5

0

6

1

(b)

storey

2

3 EC8-94

4 EC8-03

5

6 storeys

Figure 3. 6 storey: comparison between axial resisting forces and amplified design forces for columns.

In figure 3, as already shown for the 3 storey CBFs, it is reported the comparison between the buckling axial force of the columns, Nbd (horizontal lines), and the amplified design axial forces (N∗ ), evaluated according to the different code provisions (different vertical bars); due to space limitations, the chart refer only to the 6s6b-H and 6s6b-L solutions, even though the discussion of results is extended to all case studies. 634

Table 3a. 6 storey building: sections of CBF structural elements (EC8’94). EC8’94 6s4b-L

6s6b-L

6s6b-H

6s8b-L

6s8b-H

St. Diag.

Col.

Diag.

Col.

Diag.

Col.

Diag.

Col.

Diag.

Col.

1 2 3 4 5 6

HE600B HE600B HE550B HE550B HE500B HE500B

HE200A HE200A HE200A HE180A HE180A HE180A

HE500B HE500B HE450B HE450B HE400B HE400B

HE340A HE340A HE340A HE340A HE300A HE300A

HE1000M HE1000M HE900M HE900M HE800M HE800M

HE180A HE180A HE180A HE180A HE180A HE180A

HE500B HE500B HE450B HE450B HE400B HE400B

HE280A HE280A HE280A HE260A HE220A HE180A

HE360M HE360M HE340M HE340M HE320M HE320M

HE240A HE260A HE260A HE240A HE200A HE180A

Table 3b. 6 storey building: sections of CBF structural elements (EC8’03). EC8’03 6s4b-L

6s6b-L

6s6b-H

6s8b-L

6s8b-H

St. Diag.

Col.

Diag.

Col.

Diag.

Col.

Diag.

Col.

Diag.

Col.

1 2 3 4 5 6

HE650B HE650B HE600B HE600B HE550B HE550B

HE160M HE160M HE160M HE180B HE180A HE140A

HE650B HE650B HE600B HE600B HE550B HE550B

HE180M HE180M HE160M HE160M HE180B HE140A

HE800B HE800B HE700B HE700B HE650B HE650B

HE160M HE160M HE160M HE180B HE180A HE140A

HE650B HE650B HE600B HE600B HE550B HE550B

HE160M HE160M HE160M HE180B HE180A HE140A

HE650B HE650B HE600B HE600B HE550B HE550B

HE160M HE160M HE160M HE180B HE180A HE140A

It can be observed that, while in the case of high seismic zone, the amplified axial column forces evaluated according to EC8’94 are larger than the EC8’03 ones, for the low seismic zone the opposite happens; this is mainly due to the limitations of EC8’03 on slenderness ratio, as well as on difference between maximum and minimum over-strength factors; both these limitations greatly restrict the range of cross-sections which can be used for diagonals, and differently affect the structural solution in high and low seismic zone.

6 OVERSTRENGTH AND STRUCTURAL WEIGHT In figures 4 and 5 the comparison between the over-strength factors of CBFs (defined as the ratio between the base shear driving the brace to yielding and the design base shear, Vb,brace,y /Vb,brace,d ) are shown. It can be noticed that for the 3 storey CBFs, the maximum value of over-strength factors is larger than the adopted code behavior factor (q = 4); this means that column and beam elements of braces are designed with reference to axial force values larger than those evaluated from the elastic spectrum. For the 6 storey CBFs, only in the case of 6s8b-L (8 braced bays, low seismic zone) designed according to EC8’03, the over-strength is greater than 2, while in the other case it is close to 1. Therefore it can be stated that, as a consequence of the code design process, the global overstrength of CBF designed for low seismic zone are very large, particularly for low rise structures. 635

MINIMUM OVER-STRENGTH FACTORS

MINIMUM OVER-STRENGTH FACTORS EC8-94

EC8-03

OPCM-03

EC8-94

4 NO. OF BRACES

NO. OF BRACES

4

6

8

EC8-03

OPCM-03

6

8

0

1

2

3

4

5

6

0

1

2

Vb,brace,y / Vb,brace,d

3

4

5

6

Vb,brace,y / Vb,brace,d

Figure 4. 3 storey CBFs: over-strength: high seism. (left); low seism. (right). MINIMUM OVER-STRENGTH FACTORS EC8-94

MINIMUM OVER-STRENGTH FACTORS

EC8-03

EC8-94

4 NO. OF BRACES

NO. OF BRACES

4

6

8

EC8-03

6

8

0

1

2

3

4

5

6

0

1

2

Vb,brace,y / Vb,brace,d

3

4

5

6

Vb,brace,y / Vb,brace,d

2,60 EC8/94-LOW seism.

2,40

EC8/94-HIGH seism EC8/03 LOW & HIGH seism.

2,20

OPCM/03 LOW & HIGH seism

2,00

MRF-SOLUTION HIGH seism.

1,80 1,60 1,40 1,20 N.° OF BRACES

1,00 0

2

4

6

8

10

GLOBAL WEIGHT/GRAVITY SYSTEM WEIGHT

GLOBAL WEIGHT/GRAVITY SYSTEM WEIGHT

Figure 5. 6 storey CBFs: over-strength: high seism. (left); low seism. (right). 2,60

EC8/94-LOW seism. 2,40

EC8/94-HIGH seism. 2,20

EC8/03-LOW seism. 2,00

EC8/03-HIGH seism 1,80 1,60 1,40 1,20 N° OF BRACES

1,00 0

2

4

6

8

10

Figure 6. Ratio of building structural weight to gravity system weight.

In figure 6 the global structural weight of the building, normalised to the weight of the gravity structural system is provided as a function of the number of the braced bays (4, 6, 8), designed according to EC8’94, EC8’03 and OPCM’03, for the 3 storey building (figure 6 left) and for the 6 storey building (figure 6 right). In the figure the line representing a perimeter moment resisting frame solution (MRF) in high seismic zone is also reported for comparison. The figure 6 left, which refer to the 3 storey building, allows to state that the design according to EC8’03 drives to the minimum structural weight of CBFs. This is strongly related to the limit on diagonal slenderness (2.0), significantly higher than the ones imposed by the other two codes (1.5), which allows to adopt smaller sections for diagonal elements, and, consequently, also for column elements. The comparison with the MRF solution shows that only in the 4-braced-bays cases the incidence of the structural weight of CBFs is less than in the perimeter MRF solution, for both EC8’94 and 636

EC8’03, while the design according to OPCM drives always to structures heavier than the MRF solution (also for the 4 braced bays solutions). For the 6 storey building (Figure 5 right), the CBFs designed according to the two versions of EC8 have quite similar weight for the 4-braced-bays solutions, while for 6- and 8-braced-bays solutions, larger differences can be observed.

7 CONCLUSIONS New design rules for ductile design of CBF have been recently introduced, and often largely modified during the years, in seismic codes. In Europe, and also in Italy, the new seismic codes (EC8 and OPCM), recently released, include specific design provisions for ductile CBFs. However, these provisions seem to drive to a cumbersome design process and sometimes to practically impossible design solutions. In this paper the design procedure and the governing design parameters for CBF are examined with reference to EC8 and OPCM seismic codes; some case studies (3 and 6 storey CBF buildings) are designed, both in high and low seismic zone, and the effect of some design choices (i.e. number of CBF bays) is discussed and evaluated in terms of overstrength and structural weight. On the basis of the results obtained for these case studies, it can be observed that the fulfillment of the design requirements concerning section compactness, bracing slenderness and capacity design of non dissipative elements (beams and columns), often drives to very heavy braced frames, in some cases even heavier than an equivalent MRF solution, despite of the better inherent efficiency of the CBF structural configuration. In addition, the design overstrength of the CBFs often approaches, and sometimes overcomes, the value of the force reduction factor q used in the design. Finally, the application of the design provisions does not automatically ensure the attainment of a diffuse yielding pattern in the dissipative elements (diagonals) of CBFs. However these results need for a confirmation coming from a larger number of case studies and from a deeper analysis of the cyclic inelastic behavior of CBF structures. REFERENCES Eurocode 8, May 1994. Design provisions for earthquake resistance of structures ENV1998 1-2 Eurocode 8, Dec.2003. Design of structures for earthquake resistance prEN 1998-1 Ordinanza del Presidente del Consiglio dei Ministri n.3274, March 2003. Primi elementi in materia di criteri generali per la classificazione sismica del territorio nazionale e di normative tecniche per le costruzioni in zona sismica

637

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Influence of the hole clearance on the bolted joints rotational characteristics A. Kozłowski & Z. Pisarek Rzeszow University of Technology, Rzeszow, Poland

ABSTRACT: Design joint characteristic, i.e. moment – rotation curves are predicted according to code regulation, like EC-3 – part 1–8. This method and many available analytical models are based on perfect joint performance. In the most of the real joint arrangements imperfections can be observed, which influence joint behaviour. The following stages can be observed in bearing type joints behaviour: under erection loading connection behaves like slip – resistance, showing rather high stiffness. After increasing of loading, slip occurs and joint started to work for shear and bearing. This initial slip and rotation decrease the initial stiffness of the joint. The main aim of this paper is to present the influence of the hole clearance on the joint characteristic. The analytical formulae for determination of M-φ characteristics of bearing type joint with consideration of the hole clearance are presented and compared with experimental tests results. KEYWORDS:

semi-rigid steel joint, imperfections, hole clearance, experimental tests.

1 INTRODUCTION Distribution of the internal forces in a framed structures depends on the characteristics of the joints. In the global analysis joints characteristics based on analytical models of the joints are used. Many available analytical models are based on perfect joint performance and consider only two parameters: resistance and initial stiffness. In the most of the real joints arrangements imperfections can be observed, like lack of fit, random errors in bold positioning etc., which influence joint behaviour. This is especially important in bearing type of bolted connections (A category). For fitting facilities, standard clearance, oversized or slotted holes for bolts are applied. On construction stage, joints are loaded by small forces coming from the self-weight of the construction. In this phase connectors usually do not transfer the loads by bearing and joints work as slip – resistant connections. In exploitation phase, increasing loading cause the displacement in the joint and bolts started to carry the loads by bearing. Not taking into consideration holes clearances can lead to increase estimated beams deflections and joints rotations. This make that joints which in static analysis are assumed as semi – rigid behave as the nominally pined joints. Magnitude of the initial displacements of the joints is random variable and depends on the types of the joints, precisions of component elements fabrications of joints and construction fitting way.

2 INFLUENCE OF HOLE CLEARANCE ON JOINT CHARACTERISTICS Influence of hole clearance on joint characteristics for simple shear connection was presented by (Wierzbicki, 2000). For perfect joint, initial displacements δ is equal to hole tolerance  (Fig. 1). In real joints, initial displacement depends on assembly bolt positioning in the hole (Fig. 2). Value of the initial displacement differs from 0 to 2. 639

P [kN] 100 80 60 40 20 0 0.0

0.5

1.0

1.5

2.0 δ [mm]

Figure 1. Example characteristics of the simple shear joint.

P

P

P

P

P

P

δ=0

δ=∆

δ = 2∆

Figure 2. Possibly initial bolt positioning in the hole. n2 ,d

bb rb

tfb

aw

a1

n2 ,d

hw twb

tw

bw n1 ,d

w

V as ls

bs tz az

twc

a5 a4 c a6 a8 a7

hc

a a3 n1 ,d

bz a10 a10 a9 az

tfc

a2 a hb

tfb ts hz

bc

Figure 3. Geometrical dimensions of symmetrical beams to column minor axis bolted joint with use of an assembly table and a gusset web plate.

Assessments of the initial rotations of more complicated joints subjected to bending moment are difficult. Initial rotation of such joints depends not only on movement like in simple shear joints, but also on displacement of other components of the joint (Raus & Ziemia´nski 1998, Astaneh et al 2002). 3 EXPERIMENTAL TESTS The main aim of experimental tests was to analyse the behaviour of beam to column connections. The experimental tests were performed on symmetrical joints where beams were connected by bolts to minor axis column. The specimens of the connection consisted of a seating Tee section bracket bolted to beam lower flange and fin plate welded to column web and bolted to beam web. Tested joint and its geometrical dimensions are presented in the Figure 3. 640

Figure 4. View of the specimen on the test rig.

Test program consisted of 11 full – scale specimens of the joint. Each type of specimen was triple repeated, for possibility of performance statistical treatment. Any test specimen was composed of part of the column, and two beam elements. As variable factors were adopted: height of beam hb , number of bolt rows connecting beam flange to an assembly table n1 , and number of bolts connecting beam web with gusset web plate n2 . The experimental tests results have shown also significant influence of hole clearance on M-φ characteristic of the joint. Test specimens were made in the plain steel factory with typical treatments for quality. Also assembling of the beam elements of the specimens was made in typical way, placing beams horizontally on the test rig. Bolts in specimens were preloaded with controlled tightening to obtain the preloading force of 0,5 Fp,C , where Fp,C is the design preload force. View of the specimen on the test rig is shown in Figure 4. Detailed tests results are presented in (Pisarek, 2003). 4 COMPARISON OF TEST RESULS WITH ANALITICAL MODEL WITHOUT CONSIDERATION OF JOINT IMPERFECTION Based on the component method, the analytical model for computing moment resistance, stiffness and rotational capacity was elaborated. Analytical model is presented in details in (Pisarek, 2003). Comparison of M-φ curves obtained from test and perfect analytical model for selected joint is shown in the Figure 5. It can be seen, that during experimental tests increase of joint rotation under small bending moment occurred. Observed increase of joint rotation was caused by slip in the connection between beam web and gusset web plate. The magnitude of this slip is variable and depends on joint imperfection. Analytical and mechanical models of the bolted steel joints presented in literature, as well as EC3 1-8 method do not consider joint imperfections. 641

M [kNm]

30 25 20 15 10

test results analytical model

5 0 0

10

20

30

40

50 60 φ [x10-3 rad]

Figure 5. Comparison of M-φ curves obtained from test and perfect analytical model for specimen “h”.

5 ANALYTICAL MODEL WITH CONSIDERATION OF JOINT IMPERFECTIONS Assuming, that the ultimate linear displacement in the joint is 2 (before contact stresses between bolt and hole occurred), joint rotation can be calculated as:

where h = the distance from bolts in shear to center of joint rotation; and φ0 = rotation of the joint which is the result of canceling the hole clearance. Considering beam as a simple supported, midspan deflection under uniform load can be predicted from equation:

where q = value of the uniform loading; Lb = length of the beam; E = the modulus of elasticity; and J = the second moment of beam cross – section. Rotation of nominally pined joint for such beam under uniform loading is equal:

Midspan deflection of the beam related to rotation of joint can be taken from:

where φ = the rotation of the joint. 642

Assuming ultimate value of midspan deflection as fult = Lb /300, value of the rotation of the joint can be predicted from equation:

Assuming that for small value of angle tgφ = φ it can be received  ≤ 0, 0055 h. For example, for IPE 180 beam, for the mean distance from bolts h = 150 mm, and for  = 1 mm: 0,0055 h = 0, 75 mm >  = 1 mm. That means that average hole clearance applied in such joint make that for this beam ultimate deflection can be exceeded. In joints, where bolts are located near the center of joint rotation can develop danger exceeding of beam deflection. Therefore it is necessary to study joint behaviour in the assembly stage. From the analysis of tests results it was observed, that for the most specimens a mean value of clearance  could be accepted and used in the calculation. On the other hand analyzed bolted connection have a certain initial resistance coming from friction force caused of initial bolts preloading. For further analyses it was adopted a model with allowance of joint imperfection based on modified Richard – Chen function (Hong et al 2002):

where MRd = the ultimate resistance of the joint, M0 = initial resistance of the joint, which is the result of frictions force caused by initial bolts preloading, Sj.ini = the initial stiffness of the joint, without consideration of the initial slips in connections, Sp = the stiffness of the joint, with consideration of the initial slips in connections, n = the shape parameter calculated from approximation function, it can be taken as 1,7. Values of parameters for this function can be taken as for ideal joints. Only stiffness of the joint with consideration of the initial slips is depended on the hole clearance. Values of this stiffness can be predicted from equation:

The resistance of the connection M0 depend on the value of friction force due to level of bolts preloading and can be predicted from equation:

where Fi.s,R = the slip resistance of the individual bolt; and zi = the distance of the individual bolt to the center of joint rotation. Comparison of M-φ curves obtained from test and analytical models with consideration of the initial slip in shear connections and without imperfections for few tested joints is shown in Figures 6–10. From the comparison of the tests results and analytical models it can be observed, that approximation function with consideration of the hole clearance predicts better joint behaviour. 643

M [kNm]

20 18 16 14 12 10 8 6

test results model with imperfections model for perfect joint

4 2 0 0

10

20

30

40

50

φ [x10-3 rad]

M [kNm]

Figure 6. Comparison of M-φ curves obtained from test and analytical models for specimen “b”. 14 12 10 8 6 4

test results model with imperfections model for perfect joint

2 0 0

20

40

60

80

100 120 φ [x10-3 rad]

M [kNm]

Figure 7. Comparison of M-φ curves obtained from test and analytical models for specimen “c”. 60

50

40

30

20 test results model with imperfections model for perfect joint

10

0 0

10

20

30

40

50

φ [x10-3 rad]

Figure 8. Comparison of M-φ curves obtained from test and analytical models for specimen “d”.

644

M [kNm]

16 14 12 10 8 6 test results model with imperfections model for perfect joint

4 2 0 0

10

20

30

40

50 φ [x10-3 rad]

M [kNm]

Figure 9. Comparison of M-φ curves obtained from test and analytical models for specimen “e”. 30 25 20 15 10 test results model with imperfections model for perfect joint

5 0 0

10

20

30

40

50 60 φ [x10-3 rad]

Figure 10. Comparison of M-φ curves obtained from test and analytical models for specimen “h”.

6 CONCLUSIONS Comparison of tests results with two analytical models: one with consideration of hole clearance and second without imperfections can lead to the following conclusions: – initial imperfection play important role in the behaviour of bolted shear joints and should not be omitted, – in the preliminary stage of loading, semi-rigid joint with shear bolts behave like hinge, what can increase beam deflection. This is especially essential for low beam section, – presented model shows quite good agreement with test results, – to avoid an initial joint rotation in such joint it is recommended to use slip-resistance or fitted connection. REFERENCES Astaneh-Asl A., Liu J., McMullin K.M. 2002. Behavior and design of single plate shear Connections. Journal of Constructional Steel Research 58: 1121–1141.

645

Hong K., Yang J.G., Lee S.K. 2002. Moment-rotation behaviour of double angle connections subjected to shear load. Engineering structures 24. Pisarek Z. 2003. Shaping of the composite joints in steel structures. Ph. D. Thesis. Rzeszow University of Technology. Rzeszów. (in Polish). PrEN 1993-1-8. Eurocode 3. 2001. Design of Steel Structures. Part 1.8: Design of joints. CEN. Raus R., Ziemia´nski L. 1998. Modelling of the three bolts lap joints with particular attention to beginning phase of joint works. Proc. of 2nd Workshop Semi-rigid joints in metal structures, Rzeszow, 27–28 November 1998, Rzeszów, (in Polish). Wierzbicki., S. 2000. Influence of contact effect on load bearing capacity and deformability of nodes made of steel – using the FEM in structural modelling. Proc. of 3rd Workshop Semi-rigid joints in metal and composite structures, Warsaw, 24–25 November 2000, Warszawa, (in Polish).

646

Session 11: Structural integrity 2

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Identification, repairing, strengthening and revitalisation of existing buildings structures in seismic prone areas Z.Lj. Bozinovski Institute of Earthquake Engineering and Engineering Seismology, University “St. Cyril and Methodius”, Skopje, Republic of Macedonia

K. Gramatikov Civil Engineering Faculty, University “St. Cyril and Methodius”, Skopje, Republic of Macedonia

ABSTRACT: To satisfy the requirements for aseismic design of high-rises in seismically active regions, it is necessary to apply a modern concept of design that, apart from strength and deformability, shall consider plastic excursions and seismic energy dissipation ability. Namely, in the process of design or during repair and strengthening of existing, usually masonry structures, care should be taken to incorporate elements that shall eliminate the unfavourable behaviour of masonry and shall enable its favourable behaviour under strong seismic effects. Most frequently, a solution is sought in combinations of masonry with other materials that are characterized by a higher ductility and tensile and shear strength. Based on the performed synthesis of results from analytical and experimental investigations of elements of masonry systems in the world and in our country, as well as the investigations carried out in IZIIS – Skopje, proposed a procedure for design and analysis of new structures as well as repair and strengthening of damaged masonry systems exposed to static and dynamic effects.

1 INTRODUCTION Masonry structures, and masonry in general have been among principal construction materials for centuries. However, the experience from occurred earthquakes points to unfavorable behaviour of masonry structures. Masonry structures have suffered mass failure and heavy structural damage during past earthquakes. To satisfy the requirements for aseismic design of high-rises in seismically active regions, it is necessary to apply a modern concept of design that, apart from strength and deformability, shall consider plastic excursions and seismic energy dissipation ability. Namely, in the process of design or during repair and strengthening of masonry structures, care should be taken to incorporate elements that shall eliminate the unfavourable behaviour of masonry and shall enable its favourable behaviour under strong seismic effects. Most frequently, a solution is sought in combinations of masonry with other materials that are characterized by a higher ductility and tensile and shear strength. Design of new masonry structures or repair and/or strengthening of existing masonry structures is done by satisfying the requirements of the valid technical regulations, based on the most recent knowledge on seismic design and behaviour of this type of structures, controlling the strength, stiffness, deformability and capability of seismic energy dissipation of the bearing elements and the system as a whole. Based on the performed synthesis of results from analytical and experimental investigations of elements of masonry systems in the world and in our country, proposed is a procedure for design and analysis of new structures as well as repair and strengthening of damaged masonry systems exposed to static and dynamic effects (Fig. 1). 649

Figure 1. Flow – chart presenting the analytical procedure.

Masonry structures are massive and have a high bearing capacity of walls under compression, insufficient bearing capacity under tension, low ductility capacity and inadequate connection of the structural elements into a whole, particularly improper interconnection of the bearing walls in both orthogonal directions and inadequate connection of the bearing walls with the floor structures. The existing structures are most frequently constructed as stone and solid brick masonry structures. In the course of their existence, many of these structures have been exposed to natural 650

and man-made disasters as are the earthquakes and have suffered considerable structural damage. Part of them have not suffered considerable damage but are, nevertheless, non-functional and with improper non-structural elements that do not satisfy the present requirements. Therefore, a necessity arises for their reconstruction, adaptation and enlargement. To satisfy the requirements laid down in the existing technical regulations and modern knowledge on behaviour of this type of structures exposed to gravity and seismic effects, it is necessary to perform their repair, strengthening, reconstruction, adaptation, enlargement and revitalization. There are two essentially different cases. The first requires repair and strengthening because of existing damage and the second requires strengthening only. In both cases, the structures should be provided with the required seismic resistance against the seismic effects that are expected at the considered site. At the same time, revitalization of non-structural elements and installations is done. The experience from past earthquakes has pointed out that structures having different structural systems that are appropriately aseismically designed for gravity and seismic loads are capable to sustain a strong earthquake without collapse. This refers also for repaired and strengthened structures. The analysis of existing and repaired and strengthened structures requires including of several parameters having an essential influence upon strength, deformability, mechanism of behaviour of the bearing structural system exposed to vertical – gravity and horizontal – seismic effects. The problem becomes even more complex by the possibility of using several types of materials in a single structure as one of the possible ways of repair and strengthening of masonry structures. Repair and strengthening of structures damaged by an earthquake means serious modification of their strength and deformability characteristics, with modification of the strength and deformability characteristics of the elements and the system as a whole, whereby the behaviour of the structural system under future earthquakes is improved. Strengthening generally means increase of the strength and deformability characteristics of existing elements and/or adding of new elements to increase the resistance of the structure to horizontal loads. The problem becomes even more complex by the presence of several types of materials in a single structure, with characteristic points in the working diagram for quite different deformation amplitudes. With the proposed procedure, the bearing elements are included in nonlinear range of behaviour, but with a controlled and dictated mechanism of behaviour up to ultimate states of strength and deformability. Used in the procedure are experimental and analytical methods for definition of the necessary parameters. For the concrete type of masonry, needed are minimal experimental investigations of the characteristics of the material and the behaviour of elements under cyclic loads. The procedure is further analytical, in correlation with the experimental results. The procedure consists of analysis of external effects, vertical and horizontal, static and dynamic, experimental tests of the characteristics of materials and behaviour of bearing elements under cyclic loads, proportioning of the constituent elements with controlled and dictated behaviour in all phases of behaviour up to failure and response of the system under real seismic – dynamic effects. Developed are computer programmes for analysis and proportioning of elements of masonry systems with controlled behaviour and programme for nonlinear behaviour of the systems as a whole, under actual seismic effects, in which all the elements affecting structural response are included with their strength and deformability characteristics. Presented in the paper shall be the concept of repair, strengthening and revitalization of damaged and non-damaged structures as well as the procedure for analysis and design of stable and economically strengthened masonry structures.

2 DESIGN CONDITIONS The design conditions contain several parameters associated with the bearing structural system, the purpose, the required seismic protection, the location of the structure and other parameters in accordance with the valid technical regulations. The purpose of the structure is important from the 651

Figure 2. P-δ diagram, St. Sofia Church, obtained during “in situ” test.

Figure 3. P-δ diagram, Palace Sponza, obtained during “in situ” test.

aspect of definition of the required level of resistance to seismic effects, in accordance with the mentioned regulations. The level of required seismic protection is given in the seismological map. It is defined by special seismological-geological investigations of the considered site. The strength, stiffness and deformability characteristics of the materials as well as the ability of elements to dissipate seismic energy through linear and nonlinear behaviour have a different influence on the behaviour of the building structure exposed to vertical – static and horizontal – dynamic loads. To provide conditions for analysis of safety of existing structures as well as proper and most adequate design of repair and strengthening of the principal structural system, it is necessary to define the strength and deformability characteristics of the bearing walls under vertical and horizontal loads as well as the behaviour of the structure under an earthquake. Definition of the characteristics of materials, particularly strength and deformability capacities is done analytically or experimentally. The most appropriate method is the experimental that can be carried out in laboratory conditions or “in situ” on a certain bearing wall (Fig. 2 & 3). The minimum necessary parameters that should be defined are: ultimate compressive stress of masonry, ultimate tensile stress of masonry, elasticity modulus of masonry and sliding modulus of masonry. 652

3 ANALYSIS OF THE STRUCTURE During the serviceability period, the structures are exposed to the effect of vertical and horizontal, static and dynamic, permanent and temporary, normal and abnormal loads. The vertical loads, dead or life, are mainly static loads depending on the purpose of the structure, the finishing of the floor and facade structures, the insulation of the internal and external walls and the roof structure. The vertical loads should be more precisely defined, since the intensity of seismic forces and the response of the structure are directly related to the amount of vertical loads, i.e., masses of the structure. The mathematical model of the structure is defined as a system with masses concentrated at the level of floor structures, connected by springs and dampers. For the defined vertical and horizontal loads, linear static and dynamic analysis is performed for the purpose of obtaining the periods, the mode shapes, the storey stiffness and the relative displacements. The static quantities, bending moments, shear and axial forces are checked for characteristic frames in both orthogonal directions. Based on the storey stiffnesses, defined on the basis of geometrical characteristics of bearing elements, the defined relative storey displacements, the defined seismic effects on the considered location with intensity and frequency content for different return periods and the corresponding design criteria, first of all the ductility capacities of the bearing elements, through the dynamic response of the building structure, optimization of equivalent seismic forces along the height of the building is performed and hence optimization of strength, stiffness and deformability of the structure. Static analysis of the building structure under vertical loads and optimal equivalent seismic forces is performed in both the orthogonal directions. Defined are the static quantities, bending moments, shear and axial forces, for each element and each direction. The above holds for design of new structures. For analysis of repaired and strengthened masonry structures, static analysis is performed for vertical loads and equivalent seismic forces according to regulations. The main purpose of the static analysis, when analysis of elements up to ultimate states of strength and deformability is performed, is to distribute the seismic forces per bearing elements and find out the axial forces in the vertical elements due to the gravity loads of the floor and roof structures, along with the dead weight of the massive walls. For the structural elements with geometric characteristics, characteristics of materials and position in the structure, analysis of the elements is done and hence analysis of the structure is performed up to ultimate states of strength and deformability. Involved in the analysis are several types of elements characteristic for masonry structures. For several types of walls, as are stone walls, brick walls, stone or brick walls with reinforced-concrete jackets, framed brick masonry with reinforced concrete vertical and horizontal belt courses and stone walls with a concrete coating, there is a simple, but sufficiently exact way of determining the strength and stiffness capacities in the linear range of behaviour. The deformations in the same range are defined by the linear strength – stiffness relationship. This way of defining the ultimate states is not sufficiently exact for the reinforced-concrete elements as are columns and walls due to the impossibility of controlling the failure mechanism, particularly from the aspect of definining the deformation at which it occurs. Generally, for all possible elements occurring in the masonry structures, analysis of strength capacity can be done in the simple way of defining strength and stiffness capacity, whereas care should be taken as to deformability, knowing its importance for the behaviour of the structure under an earthquake. Considering the complicated behaviour of reinforced-concrete elements, a more precise way of determining their ultimate strength and deformability value is given, with control of the mechanism of behaviour from the beginning of loading up to failure. This is important because we are often forced to define simultaneously the ultimate states of masonry and reinforced-concrete elements so that, by their superposition, one arrives at storey strength, stiffness and deformability capacities in both the orthogonal directions of the structure(Fig. 4). The most important for the behaviour of the structures is their response under static and dynamic loads. Masonry structural systems, as well as other systems, exposed to dynamic - seismic effects of moderate and high intensity, suffer nonlinear deformations of the constitutent elements, by which 653

K=4160 [kN] LP=0.11 δy =1.13 [cm] δu = 4.20 [cm]

Representative bilinear storey diagram

Cumulative storey diagram – RC elements

Qy = 4700 [kN] Qu = 6100 [kN] D = 3.72

Cumulative storey diagram – Masonry wall elements

Individual RC elements Individual masonry wall elements

Figure 4. Force – displacement diagram for masonry and R/C elements.

the initial strength and stiffness is decreased considerably, depending on the size of deformations and the number of times they are iterated. 4 EVALUATION OF THE STABILITY OF THE STRUCTURE AND THE NEED FOR REPAIR AND STRENGTHENING Based on the analysis of the existing state of the building structure and damages to the structure, the elements of occurred damages and the reason of occurred damages are defined. Such considerations are important for selection of possible and necessary measures for repair and strengthening of the structural system. The strength and deformability capacities of the bearing elements and the system as a whole are compared to those required (according to the regulations) and those required for the analyzed structural response under expected earthquakes on the considered site, with intensity and frequency content. If the strength and the deformability capacity is less than the required, it is concluded that the building structure does not have sufficient strength and deformability and therefore it needs repair and strengthening. Each concrete structure is a case for itself and there are several ways of repairing and strengthening it. The solution for repair and/or strengthening depends on: seismicity of the site, local soil conditions, type and age of the structure, level and type of damages, time available for repair and/or strengthening, equipment and man power, restoration and architectural conditions and requirements, economic criteria and necessary seismic safety. Selected are several variant solutions for repair and strengthening. Analysis of each solution is done and an insight into the advantages and disadvantages is obtained from several aspects. Out of these, selected is the most adequate solution from the economic aspect and the stability aspect according to the required seismic protection. Generally, some possible ways of repair and strengthening of different types of masonry elements and buildings is given in the subsequent text. Out of several analyzed possible solutions for repair and strengthening of the main structural system, selected is the most favourable from the aspect of: stability, i.e., fulfillment of the design criteria according to regulations, possibility for realization of the solution, available materials, economic justification, fulfillment of social requirements and satisfying of aesthetic requirements (Fig. 5 to 8). 654

Figure 5. Storey P- diagram, Army Club in Bitola, direction X-X, repaired state.

Figure 6. Storey P- diagram, Army Club in Bitola, direction Y-Y, repaired state.

Figure 7. Storey P- diagram, primary school Radobor, direction X-X, existing state.

655

Figure 8. Storey P- diagram, primary school Radobor, direction X-X, repaired state.

5 ANALYSIS OF THE STABILITY OF THE REPAIRED AND THE STRENGTHENED SYSTEM For the selected solution of repair and strengthening, analysis of the stability of the structure is performed in order to confirm the necessity and the justification of the undertaken activities. For the repaired and strengthened system, the same analysis is done as that for the existing state of the structure, defining the strength, stiffness and deformability capacities of the structure and comparing them with the required quantities for expected earthquakes with intensity and frequency content as well as the limitations of storey displacements from other reasons. The maximum displacements, defined as required, are compared to the capacities of displacement, whereas the plastic excursions show how many times the structure enters into the nonlinear range of behaviour.

6 CONCLUSION ON THE STABILITY OF THE STRUCTURE AND THE VULNERABILITY LEVEL Based on the strength and deformability capacities of the bearing elements and the structure as a whole and on the basis of required strength and deformability for expected seismic effects with intensity and frequency content, conclusions are drawn regarding the stability of the structure and its vulnerability level. It is of exceptional importance to bring the strength, stiffness and deformability of the structure within the frames of the requirements according to the valid technical regulations and latest knowledge on the behaviour of masonry structures exposed to gravity and seismic effects. REFERENCES Bozinovski, Lj.Z. 1997. Procedure of analysis of masonry building structures to be constructed in seismic prone areas’, Fourth National Conference on Earthquake Engineering , Ankara, Turkey, 17–19 September, 1997. Bozinovski, Lj.Z. 2000. Concept of repair, strengthening and revitalization of existing damaged and nondamaged masonry structures, INDIS 2000 and CIB W-63, Novi Sad, Yugoslavia, November 22–24, 2000.

656

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Micro-modeling of RC frames with masonry infill L. Krstevska & D. Ristic Institute of Earthquake Engineering and Engineering Seismology, Skopje, Republic of Macedonia

ABSTRACT: Presented is the developed analytical micro-model for refined simulation of the non-linear behavior of RC infilled frames under strong earthquakes. The concept is based on discretization of the structural system by a finite number of nonlinear discrete components connected at the nodal points of the model and knowledge of the stress-strain relationship for each material in the structure (micro-modeling concept). The discrete components are different for the individual structural elements of the system, their main characteristic being nonlinearity. The application of the proposed concept applying the nonlinear discrete components and using the originally developed complex computer program EURO NORA is performed by modeling and seismic analysis of a plane RC frame model with masonry infill. The verification of the proposed analytical models is presented using the obtained results from shaking table tests on a two storey-two bay model of the plane frame. The results obtained from the extensive theoretical studies confirmed the applicability of the proposed micro-models.

1 INTRODUCTION Masonry as an infill is used in a great number of different variants in respect to the type of bricks, type of mortar, presence of reinforcement in the masonry, way of loading and alike. Following the scientific-research trends in Europe and in the world in the field of earthquake engineering and stability of systems, noteworthy is the considerable number of cooperative projects that involve research teams from a number of countries and treat structural systems with masonry infill from different aspects. Within the frames of such an international cooperative project INCO COPERNICUS PROJECT No. IC15-CT97-0203, “TOWARDS EUROPEAN INTEGRATION IN SEISMIC DESIGN AND UPGRADING OF BUILDING STRUCTURES”, realized in the period October 1997 – January 2001, financed by the European Commission, complex investigation of reinforced-concrete frame structures with masonry infill has been performed from several aspects, by participation from seven renowned European institutions: University of Ljubljana, Slovenia, University in Bucharest, Romania, Institute for Structures and Architecture, Slovakia, ISMES, Bergamo, Italy, University in Bristol, Great Bretain, JRC, Ispra, Italy and Institute of Earthquake Engineering and Engineering Seismology (IZIIS), Skopje, Republic of Macedonia . The research activities within the project related to refined nonlinear numerical simulation of RC frames with infill have successfully been realized in IZIIS, Skopje, Republic of Macedonia. A new concept for nonlinear refined micro analysis of frame structures with infill of non-reinforced and reinforced masonry, with or without openings for doors or windows has been proposed. The verification of the applied analytical procedure through determination of the nonlinear response of these systems to earthquake effect showed that the nonlinear response of frame systems with masonry infill can successfully be simulated by use of the proposed concept for refined nonlinear micro-modeling and by application of the developed EURO NORA programme. 657

2 PROPOSED CONCEPT OF NONLINEAR MICRO-ANALYSIS 2.1 General approach The proposed concept is based on discretization of the structural system by a finite number of nonlinear discrete components connected at the nodal points of the model. The concept itself is based on knowledge of the σ -ε relationship for each material incorporated in the structure, i.e., the material from which the structural elements are made (micro-modeling concept). The discrete components are different for the individual structural elements of the system, their main characteristic being nonlinearity. For micro-modeling of an integral RC frame system with masonry infill, corresponding models of nonlinear discrete components (NDC-models) have been developed to represent the specific nonlinear characteristics of the different structural elements and materials of the system: 1. NDC-FRAME model – nonlinear model of discrete components for the beam elements, for modeling of the behaviour of the structural elements of the frame – columns and beams; 2. NDC-INFILL model – nonlinear model of discrete components for modeling of infill elements; 3. NDC-MORTAR model – nonlinear model of discrete components for modeling of behaviour of mortar in all the contact zones; 4. NDC-REINFO model – nonlinear element for modeling of the behaviour of reinforcement, in the case of reinforced masonry. 2.2 Phenomenological models of discrete components – NDC models 2.2.1 NDC-FRAME model To model the nonlinear behaviour of structural reinforced concrete or steel elements of the structure, the developed micro-nonlinear finite element presented in Fig.1 is used. This model has originally been developed and proposed by Prof. Dr. Danilo Ristic and has multiply been applied and verified based on the results from numerous experimental tests. The 2D element is defined at two nodal points i and j that have three global degrees of freedom each: axial deformation and two rotations at the ends. The inner forces and the forces at the ends are related to the adopted referent axis of the element, defined by nodal points i and j, while the external forces act upon the nodal points of the element in the direction of the global degrees of freedom of the node.

Figure 1. Finite element, NDC-FRAME model.

658

The distribution of nonlinearity along the finite element is controlled by a finite number of sub-elements defined between previously specified cross-sections – interface elements. The main purpose of these interface elements is computation and including of nonlinearity and its propagation along the element depending on the previous loading history and the current level of axial force. Defined for each interface element are: (a) axial stiffness–stress–strain (σ –ε) relationship with included characteristic parameters, and (b) bending stiffness–moment–curvature (M –ϕ) relationship providing thus realistic simulation of the nonlinear behaviour of the element during quasi-static or dynamic effect. The interface element has two local degrees of freedom: axial deformation εas at the plastic center of gravity of the cross-section and curvature ϕs of the cross-section. Each interface element is divided into layers-axial surfaces and for each layer are defined the hysteretic stress-strain (σ –ε) relationships. 2.2.2 NDC-INFILL model Refined nonlinear modeling by discrete components of structural infill elements is achieved through the infill unit (modulus) composed of six axial nonlinear springs or nonlinear discrete components, Fig. 2. By successful analytical presentation of the realistic nonlinear hysteretic relations for each separate discrete component, the characteristics of nonlinear behavior and the failure modes of the infill under cyclic load effect can realistically be simulated. 2.2.3 NDC -MORTAR model Analogously, refined nonlinear modeling by discrete components of mortar elements is achieved through a similarly assembled nonlinear mortar unit whose standard shape consists of six equivalent nonlinear discrete components-axial springs, presented in Fig.3. Adopting a successful analytical presentation of actual nonlinear hysteretic relations for each discrete component also in this case, the characteristics of inelastic behavior and failure modes of mortar under the effect of generalized cyclic loads can also be realistically simulated. 2.2.4 NDC-REINFO model With this discrete nonlinear element, modeling of the nonlinear behaviour of the infill reinforcement is done, should there is a case of reinforced masonry. In accordance with the introduced analytical approach to nonlinear micro-modeling, the proposed nonlinear model for the reinforcement elements placed in masonry consists of a nonlinear discrete component - axial spring with adopted characteristic nonlinear axial force-deformation (P−δ) relationship, Fig. 4. By appropriately adopted analytical presentation of the current nonlinear hysteretic relationship, a realistic simulation of inelastic behaviour of reinforcement layers under earthquake or other cyclic effect is provided for each discrete component.

Figure 2. NDC-INFILL model – “brick modulus” formed by six discrete components, axial nonlinear springs.

659

Figure 3. NDC-MORTAR model, discrete components of nonlinear axial springs.

Figure 4. NDC-REINFO model, for modeling of reinforcement in case of reinforced masonry.

3 COMPUTER PROGRAMME EURO-NORA FOR NONLINEAR ANALYSIS OF RC FRAME STRUCTURES WITH MASONRY INFILL, BASED ON THE MICRO-MODELING CONCEPT To implement the concept of micro modeling in the analysis of nonlinear response of structures, an optimized analytical procedure was developed which can be summarized into the following steps: • idealization of the system with nonlinear infill components and frame elements; • identification of unknown displacements at the nodes by which the displacement response of the idealized structure is completely defined; • derivation of equations of equilibrium of forces corresponding to the nodal displacements; • solving of the equilibrium equations over unknown displacements and • computation of stresses in elements or distribution of forces using the known nodal displacements. For structural analysis with application of the proposed micro-modeling concept, a computer programme EURO-NORA was developed by Prof. Dr. Danilo Ristic (2001). This programme 660

Figure 5. Plane and spatial infilled frame models tested dynamically.

offers the possibility for performance of the following three types of analysis of two-dimensional structures: • static linear and nonlinear step-by-step analysis (under time-dependent loads); • analysis of initial dynamic characteristics; • dynamic linear or nonlinear step-by-step analysis (time-history response).

4 NONLINEAR MICRO-MODEL OF RC FRAME WITH INFILL To verify the practical application of the formulated nonlinear micro-model for real numerical simulation of nonlinear behaviour of RC frame structures with masonry infill under earthquake excitation, modeling and seismic analysis of two different plane frame models with masonry infill tested experimentaly in laboratory conditions have been performed. The example presented in this paper is related to a two-storey two-bay RC infilled frame (tested experimentaly in the Laboratory of ZRMK in Ljubljana, Slovenia) which is a part of the spatial frame model tested in the Laboratory of the University in Bristol, UK, designed and constructed to a scale of 1:4, Fig. 5, according to the true replica modeling technique. The physical model is an RC frame infilled with plain masonry without openings. The spatial frame model has been tested on the seismic shaking table in Bristol applying uniaxial excitation of modified sine type in the frames of a common envelope. 4.1 Formulated analytical micro-model Applying the proposed micro-modeling concept, the analytical model for the two storey infilled frame is formulated as presented in Fig. 6. The analyses are performed for nonlinear analytical model for the prototype of the scaled model, having the following global geometry: Lp = 5.84 m and Hp = 5.12 m. According to the proposed micro-modeling concept, the analytical model has the following characteristics: • • • • • • • •

total number of nodal points NP = 729 total number of boundary nodes NBN = 27 total number of degrees of freedom NDOF = 2106 total number of basic nonlinear elements NEL-B = 2058 total number of dummy elements NEL-D = 2058 total number of all discrete elements NEL-TOT = 4116 total number of frame elements NFEL = 130 total number of brick elements NBEL = 864 661

Figure 6. Analytical model of nonlinear discrete components for the model of the frame tested on seismic shaking table in Bristol, UK. M2-P: H, IEI=1.5g, NP=378, Dir-x Displacement (m)

Displacement (m)

M2-P: H, IEI=1.5g, NP=729, Dir-x 0.008 0.006 0.004 0.002 0 -0.002 -0.004 -0.006 -0.008 0

2

4

6

8

10

0.004 0.002 0 -0.002 -0.004 0

Time (s)

2

4

6

8

10

Time (s)

Figure 7. Displacement response in x-direction for the model, nodal points NP = 729 and NP = 378.

• total number of mortar elements NMEL = 1064 • final number of used hysteretic relations NHYS = 5936 The seismic analysis for the model is performed under the same dynamic excitation like the one applied during the experimental testing of the model on the seismic shaking table at the University in Bristol. The analysis was performed applying a solution time step of t = 0.001 s and a total time solution time T = 10 s. The maximum input intensity was Amax = 1.5 g, as an intensity which caused serious damage to the physical model during the experimental testing and the period of the input sine type function was equal to the first period of model prototype, Tsin = 0.08 sec in order to produce vibration of the model in resonance conditions in the initial state. Selected results obtained by the analysis of the model M2-P are presented in Fig. 7 to 9. 662

M2-P: H, IEI=1.5g, EL=66, NP=378

M2-P: H, IEI=1.5g, EL=53, NP=27

moment (kN*m/10)

moment (kN*m/10)

60 40 20 0 -20 -40 -60 0

2

4

6

8

8 6 4 2 0 -2 -4 -6 -8

10

0

2

4

6

8

10

Time (s)

Time (s)

Figure 8. Moment time history, EL = 53, NP = 27 and EL = 66, NP = 378. M2-P, EL=78, NP=729

M2-P, EL=65, NP=378

20

35

15

28 21

force F (kN/10)

force F (kN/10)

10 5 0 -5

14 7 0

-7 -14

-10 -21 -15

-28

-20 -0.006 -0.004 -0.002

0

0.002

0.004

-35 -0.003

0.006

displacement D (m)

-0.002

-0.001

0

0.001

0.002

0.003

displacement D (m)

Figure 9. Hysteretic relationship force-displacement in x-direction, EL = 78, NP = 729 and EL = 65, NP = 378 for the model.

Figure 10. Axial and transverzal forces (in kN/10) for the model at the moment of max. displacement.

Presented diagrams of shear and axial forces (Fig. 10) calculated in the elements of the NDCFRAME models (frame elements, columns and beams) have the expected distribution for a frame system under lateral force action. For the maximal deformation of the system during the dynamic excitation, dmax = 3.73 mm in NP = 729, the higher values of shear and axial forces at the first storey are obvious. This storey has been much more damaged then the second storey. An interesting parameter showing the possibilities of the applied computer program for performing a complex nonlinear analysis of a system modeled applying the proposed micro-modeling 663

M2-P, Dx-NP=53

M2-P, Dy-NP=53 Displacement (m)

Displacement (m)

0.00006 0.00004 0.00002 0 -0.00002 -0.00004

0.00003 0.00002 0.00001 0 -0.00001 -0.00002 -0.00003

-0.00006 0

2

4

6

8

0

10

2

4

6

8

10

Time (s)

Time (s)

Figure 11. Time history of displacement for nodal point NP = 53 in x and y direction.

concept is displacement of the brick-mortar contact nodal point in the right bottom corner, NP=53, Fig. 11. In this area high compression forces and crashing of mortar is expected. After the dynamic action, the nonlinear displacement for this point in both directions is obvious.

5 THE MOST IMPORTANT CONCLUSIONS Based on the performed investigations, the following important conclusions were drawn: • the proposed concept of micro-modeling has a lot of advantages since it enables detailed simulation of the characteristics of the nonlinear response of the integral system of masonry infill frames; • the concept has unique advantages since it enables realistic simulation of a complex asymmetric hysteretic response of the frame under the effect of variable dynamic axial forces; • the characteristics of the infill components include simulation of opening of cracks, closing of cracks, failure under tension, failure under compression, modeling of contact brick-mortar, mortar – column contacts, mortar - beams contacts, combined contacts, etc., and • the concept has a lot of advantages since it enables modeling of ordinary infill, infill with openings, non-strengthened infill, strengthened infill, different types of infill, etc. • Finally, based on the proved successfulness of the originally developed computer programme EURO-NORA and the successfulness of the presented theoretical results, it was proved that the developed concept of a micro-model possesses theoretical generality and high preciseness. Considering the offered advantages, the proposed concept of a nonlinear micro-model (NDC Model) is expected to be widely applied in modern earthquake engineering. REFERENCES Krstevska, L. & Ristic, D. 2004. Seismic response of RC infilled frames – micro-model for non-linear numerical simulation. 13th WCEE, Vancouver, B.C., Canada, August 1–6 2004. Krstevska, L. 2002. Development and Application of Non-linear Micro Models for Evaluation of Seismic Behaviour of RC Frames Infilled with Plain and Reinforced Masonry, (Ph.D. Disertation), University of Skopje, IZIIS, Skopje, Republic of Macedonia. Ristic, D., 1988. Nonlinear Behaviour and Stress-Strain Based Modeling of Reinforced Concrete Structures Under Earthquake Induced Bending and Varying Axial Loads, (Ph.D. Disertation), School of Civil Engineering, Kyoto University, Kyoto, Japan. Ristic, D. & Krstevska, L. 2001. Final Report (1.10.1997–31.01.2001) for the INCO-COPERNICUS Contract IC15-CT-97-0203 – Towards European Integration in Seismic Design and Upgrading of Building Structures (EUROQUAKE), University of Bristol, G. B.

664

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Design of low-yield metal shear panels for energy dissipation G. De Matteis, A. Formisano, F.M. Mazzolani & S. Panico University of Naples “Federico II”, Department of Structural Analysis and Design, Napoli, Italy

ABSTRACT: Stiffened shear panels are a useful tool for passive seismic protection of steel framed buildings. Different metal materials can be used, namely low yield strength steel, mild steel and aluminium. The choice of the material depends on the demand, measured in terms of strength, strain hardening and ductility on the applied panels, which is a function of seismic requirements of the whole structure that can be obtained by a global dynamic analyses. The energy dissipation capacity of shear panels derive from using appropriate reinforcing ribs to prevent shear buckling in both the elastic and plastic field. The spacing and the flexural inertia of ribs depend on the material features as well as on the required shear panel deformation capacity. A wide numerical study has been carried out aiming at providing a useful design tool. Therefore, a number of shear panels have been numerically tested, considering different materials, plate slenderness and ribs arrangement. In the current paper the results of this parametrical study are provided.

1 INTRODUCTION In the conventional seismic design methods, the input energy under extreme earthquakes is absorbed by means of large plastic deformation of the beams, with the consequent damage of structural members. Actually, a new seismic design method, also known as Damage Tolerant approach (Wada et al., 1992), consists in protecting the primary framed structure from seismic damage through the use of special devices acting as hysteretic and/or viscoelastic dampers. Shear panels represent one of more convenient systems for passive seismic protection of framed buildings against low and high intensity earthquakes thanks to their remarkable lateral stiffness and strength. Their dissipative mechanism is based on the metallic yielding technology and if they are provided with appropriately selected stiffeners, they can also give rise to a large energy dissipation capacity related to the large portion of panel where shear plastic deformations take place before that local and global buckling phenomena occur. Unstiffened shear panels dissipate energy through the metal yielding along tension diagonals only and are characterized by cyclic behaviour with pronounced degradation of stiffness and strength owing to out-to-plane displacements produced by shear buckling of the plate. Instead, stiffened shear panels present a dissipative mechanism due to pure shear deformation thanks to inhibited buckling due to the lateral confining action carried out by the ribs. In order to have a pure shear dissipative mechanism it is necessary to use stiffened plates so to delay shear buckling in the plastic field. This purpose can be achieved by adopting a material characterized by a high E/fy ratio, which allows a larger width-to-thickness ratio for shear buckling of steel plate. So far, Low Yield Strength (LYS) steel has been proposed (Tamai et al., 1991; Tanaka et al., 1998). The low yield strength steel is a type of steel that, due to the small amount of carbon and alloying elements, has a nominal yield stress of about 90–120 MPa, the same Young’s modulus as conventional steel and a nominal elongation over 50%. It should be considered that the low yield point ensures the energy dissipation yet for smaller deformation levels, as in the case of wind and moderate earthquakes, working as dampers also at the serviceability limit state. For serving as sacrificial elements and to protect also non-structural elements for smaller seismic motions, it could be profitable to use a metal with even lower yield stress. To this aim the interest moved toward the possibility to use pure aluminium as metal to build dissipative panels (De Matteis 665

et al., 2003). Thanks to its low percentage of alloying elements, the pure aluminium is able to give a yield stress lower than LYS steel and simultaneously, due to the lower specific weight, to reduce the overloading on structural elements. Besides, contrary to LYS steel, almost pure aluminium, with a high percentage of purity, is easily available on the world market. In this paper the results of a numerical study carried out on shear panels made of four different metals and analyzed by sophisticated FEM models are provided. This study is framed within a large experimental programme recently undertaken at the University of Naples ‘Federico II’. For the sake of comparison, other two, more traditional materials have been considered as well, namely mild steel (Fe 360) and the wrought aluminium alloy EN-AW 5154A. The obtained results outline the influence of the material features on the response of shear panels, emphasising also the different effect of the buckling paths, namely global shear buckling of the panel, flexural buckling of the stiffener and local buckling of the single panel portion. Also, appropriate design charts are provided, allowing the determination of optimal panel configurations (stiffener inertia and plate aspect ratio) in relation to the performance required to shear panels measured in terms of strength and deformation capacity.

2 DESIGN CRITERIA OF DISSIPATIVE SHEAR PANELS A dissipative shear panel based on metallic yielding technology is able to provide a stable cyclic behaviour if it is designed in such a way to avoid any buckling phenomenon before the plastic deformations occur. For this reason it is necessary to check that the yielding of the panel takes place for loading levels lower than the ones corresponding to the buckling of the panel itself. This aim is reached if the following condition is applied:

where α is the material hardening ratio, τu is the ultimate shear stress and τy is the yielding shear stress. In a stiffened shear panel the shear buckling phenomenon can be both of local type and of global type (Figure 1). In the first case the instability waves are confined within the portion of plate enclosed by longitudinal and transversal stiffeners (Figure 1a). While in the case of global buckling, owing to their limited inertia, the stiffeners are involved in the buckling shape according to a flexural buckling form (Figure 1b). The buckling problem of shear plates in the plastic field has been approached by extending at the inelastic range the formulas valid in the elastic range, by replacing the Young’s modulus with an appropriate reduced modulus Er = µE, where µ < 1 is the plasticity factor (Bleich, 1952). In this way the critical shear stress can be obtained by using the same formula of elastic buckling

(a)

(b)

Figure 1. Shear buckling phenomena for stiffened shear panel: (a) local, (b) global.

666

(Timoshenko & Gere, 1961):

where ν is the Poisson’s modulus and kτ is the shear buckling coefficient, which depends upon the boundary restraints and the aspect ratio a/bw , where a is the spacing of transverse stiffeners, bw and tw are depth and thickness of the plate, respectively. For instance, the plasticity factor µ was determined by applying the unified theory of plastic buckling for an infinitely long plate of 24 ST aluminium alloy in uniform shear (Stowell, 1948). Stowell gave the values of µ in the form of curves plotted versus the intensity of the shear stress τ . It was found that µ is nearly independent of the degree of restraint at the long edges and √its values may be well approximated by a function of the tangent-modulus Et according to the Et /E-curve , which provides conservative values for the critical shear stress. The limit value of the normalized slenderness parameter λ¯ w , which allows the fulfillment of eq. (1), may be obtained by eq. (2) by assuming the Poisson’s modulus ν equal to 0.33 :

Equation (3) shows that the limit value of slenderness parameter λ¯ w decreases to the increasing of hardening ratio α and to the decreasing of the plasticity factor µ. , which is the tangent-modulus in the plastic field in related to. The slenderness parameter λ¯ w can be also evaluated according to the EC9 provisions:

b∗w = bw1 for local buckling and b∗w = bw for overall buckling (Figure 2). The shear buckling coefficient kτ can be calculated by the relationships given by the EC9 for local and overall buckling.

a

z

z

z

15tw 15tw

b

z

bw2

Longitudinal stiffener

bw

b

hw

bw1

Transverse stiffener

a

(a)

(b)

Figure 2. Geometrical parameters for stiffened shear panels.

667

BEAM elements F

SHELL elements

SPRING elements

Loading frame

Shear panel Panel configuration Type F (a)

(b)

(c)

Figure 3. FEM modeling: (a) considered panel geometry, (b) FEM model, (c) adopted mesh.

3 THE FINITE ELEMENT MODELLING 3.1 General The ongoing experimental activity foresees the use of aluminium panel specimens measuring 1500 by 1000 mm, with thickness of 5 mm and stiffened with longitudinal and transversal open rectangular-shaped stiffeners. The ribs have been drawn from the same aluminium sheets utilized for the shear plates. Different rib arrangements will be considered in order to assess their influence on the ductility and hysteretic behaviour of aluminium shear panels. For each rib arrangement, the spacing and depth have to be selected accordingly, in order to fulfil the requirements given in Section 2 and in particular to ensure a shear buckling of the plate delayed after the shear yielding with a preliminary hardening ratios α. Shear panels will be loaded in shear by a very rigid and pinned steel frame equipped with lateral braces. Aiming at developing appropriate parametrical study, appropriate FEM models have been set up so to investigate the effect of the main influential parameters on the performance of the structural system under consideration (Panico et al., 2003; De Matteis et al., 2004). In particular, in this phase of the study, the shear panel under consideration is characterized by stiffeners alternatively placed on each side of the plate in order to have 4 × 6 portions having dimensions of 250 × 250 mm (‘type F’ in Figure 3a). The monotonic and cyclic behaviour of such panel with different configurations in terms of aspect ratio and stiffener depth have been investigated by using the ABAQUS non linear numerical analysis program, taking into account the influence of geometrical imperfection, the actual inelastic properties of the material, the influence of ribs and the interaction with the surrounding loading frame (De Matteis et al., 2004b). Beam and column members of the loading frame are obtained by using double channel shapes (depth equal 200 mm) and connected to each other with pin joints at the four corners. They have been modelled by using BEAM elements, while S4R SHELL finite elements have been used to model aluminium plate and the stiffeners. The input load is applied to the top beam of the external lateral reaction frame (Figure 3b). In order to provide major generality to the results of the following parametric study, in the developed numerical model, the eccentricity between external frame members and the internal plate element due to connecting system and also the cross-section dimension of the frame members has been neglected. Namely, the frame member axes have been considered coincident with the plate edges. 3.2 The considered materials In the current numerical, as base material for shear panels, four different metals have been considered. In fact, in addition to LYS steel and almost pure aluminium (AW 1050A) that have been already 668

Table 1. Metal materials considered in the numerical analyses. Material

f0.2 , (fy ) (Nmm−2 )

fu (Nmm−2 )

εu (%)

E (Nmm−2 )

E/f0.2

α = fu /f0.2

Fe360 steel LYS steel Pure aluminium (EN-AW 1050A) Aluminium alloy (EN-AW 5154A)

(235) 86 21.3 75.2

360 254 80 203.6

25 40 45 18

210000 210000 70000 70000

893 2441 3286 931

1.53 2.95 3.76 2.71

Figure 4. Assumed stress–strain curves for the materials under consideration.

assumed as convenient materials to build up dissipative metal shear panels (De Matteis et al., 2003; Nakagawa et al., 1996), a different aluminium alloy (AW 5154A) and a traditional Fe360 steel have been considered as well. The former has been chosen because it is an aluminium alloy with a quite large ductility and reasonably limited strength, easily available on the market, it being commonly used in the marine and shipbuilding industry. The later has been selected for comparative reasons, it being the material more economical and easily available. The above material selection allows the comparison in terms of performance of shear panel made with materials having different yielding strength, elastic modulus and also hardening features. It is worth noticing that the two aluminium alloys considered in this study are commercial materials that have been subjected to heat treatment to improve their mechanical features (reduce the yield stress and increase the ductility). Therefore, the relevant stress–strain relationships have been defined according to material tests carried out at the Laboratory of the Department of Structural Analysis and Design of the University of Naples Federico II. Contrarily, for LYS steel and mild steel Fe360 a Ramberg-Osgood model and a tri-linear constitutive law have been used, respectively. In particular, the former model has been defined on the basis of existing experimental results (Nakashima et al., 1994). The mechanical features of the considered materials are listed in Table 1, where f0.2 indicates either the conventional yield strength (aluminium alloys and LYS steel) or the yield strength fy (Fe360 steel), while the assumed stress–strain relationships are drawn in Figure 4.

4 THE PARAMETRIC STUDY The aim of the study is to define useful design tools for the selection of shear panels made with different base materials in order to ensure a suitable behaviour of the panel under both monotonic and cyclic loading conditions in relation to the shear deformation demand. Based on the panel 669

Table 2. Depth of ribs hst (in mm) for the considered configuration. b/t = 25

b/t = 37.5

b/t = 50

b/t = 75

b/t = 100

b/t = 200

0 10 20 40 60 80 100 110 118 120 150 185

0 10 20 40 50 60 70 80 92 100 120 144

0 10 20 30 40 50 60 70 77 80 100 120

0 10 20 30 40 45 50 55 60 70 80 93

0 5 10 20 30 40 45 50 55 60 70 78

0 5 10 15 20 25 30 32 35 40 45 51

Table 3. Ist,lim values according to EC3 and EC9. b/t

Ist,lim (mm4 )

gst,lim

hst,lim (mm)

25 37.5 50 75 100 200

24.000.000 7.111.111 3.000.000 888.889 375.000 46.875

210 210 210 210 210 210

118 92 77 60 50 32

configuration type F previously defined, the numerical study has been carried out ranging the widthto-thickness ratio of the plate elements and the second moment of area of the ribs. In particular, the stiffness of the ribs has been changed by varying their depth, while their thickness has been assumed equal to the one of the sheeting. The analyses have been carried out by assuming width-to-thickness ratios b/t equal to 25, 37.5, 50, 75, 100 and 200, where the distance between the ribs b is constant and equal to 250 mm, while the thickness of the plate t ranges from 1.25 mm to 10 mm. The b/t ratios and the corresponding values of examined depths of the ribs hst are given in Table 2. It is worth noticing that for each ratio b/t, the case of ribs having a second moment of area equal to Ist,lim has been also considered (bold numbers in Table 2), where Ist,lim is the limit value beyond which the stiffeners may be considered as “rigid”. It is assumed as sum of moments of inertia of the nst intermediate transverse stiffeners placed in√the panel and it has been calculated according to EC3 and EC9 provisions, considering that b/hw < 2, where hw is the width of panel equal to 1000 mm. For each value of b/t ratio, the second moment of area Ist,lim , the corresponding limit values of the stiffener depth hst,lim and the corresponding normalized stiffness parameter γst,lim are listed in Table 3, where the normalized stiffness parameter γst is calculated by the following equation, where a constant value of Poisson’s ratio ν = 0.3 has been assumed:

5 THE OBTAINED RESULTS The monotonic analyses of the several examined configurations of panel have been given in terms of F/F0.2 – γ curves, where F0.2 is the yielding shear force calculated by assuming an uniform 670

Fe360 steel

LYS steel

Pure Aluminium (EN-AW1050A)

Aluminium alloy (EN-AW5154A)

Figure 5. F/F0.2 – γ curves for panel configuration characterised by b/t = 25 and different stiffer depth.

Fe360 steel

LYS steel

EN-AW 1050A

EN-AW 5154A

Figure 6. Typical buckling mechanisms for analyzed shear panels.

distribution of yielding shear stress (evaluated according to either fy or f0.2 , see Table 1) along the width of panel B = 1000 mm and γ is the shear deformation. All the analyses have been worked up to 20% of shear deformation, which can be considered as a limit value for practical applications. In Figure 5, for the sake of comparison, F/F0.2 – γ curves for a value of b/t equal to 25 are presented for each material. They highlight the increased effect of depth ribs on the global plastic response of shear panels. The comparison among different materials, clearly emphasises that the best performance in the plastic range is provided by the AW 1050 aluminium alloy, it being characterised by the highest strain hardening and the most convenient E/f0.2 ratio. On the contrary, the worst performance is exhibited by Fe360, which is characterised by a poor post-elastic behaviour and by a larger value of yield strength, strongly increasing the susceptibility of shear panel to buckling. It is also emphasised that the performance of LYS steel and AW 5154 aluminium alloy are very similar to each other, they being characterised by similar values of yielding strength. 671

4.0 g=7,5%

F/F0,2

3.5 3.0

4.0

b/t=25 b/t=37.5 b/t=50 b/t=75 b/t=100 b/t=200

3.0

2.5

2.5

2.0

2.0

1.5

1.5

1.0 0 4.0

50

F/F0,2

3.5 3.0

100

150 200 Fe360 steel g=7,5%

250

gst

0 4.0

2.0

2.0

1.5

1.5 gst

F/F0,2

3.0 2.5

50 100 150 200 250 Pure aluminium (EN-AW 1050A)

50

3.5

2.5

0

b/t=25 b/t=37.5 b/t=50 b/t=75 b/t=100 b/t=200

1.0 300

b/t=25 b/t=37.5 b/t=50 b/t=75 b/t=100 b/t=200

1.0

g=7,5%

F/F0,2

3.5

100

150 200 LYS steel g=7,5%

250

0

300

b/t=25 b/t=37.5 b/t=50 b/t=75 b/t=100 b/t=200

1.0 300

gst

50 100 150 200 250 Aluminium alloy (EN-AW 5154A)

gst

300

Figure 7. F/F0.2 – γst curves for shear deformation γ = 7.5%.

Anyway, it can be noticed that in all the cases, for each b/t ratio, for increasing values of ribs stiffness, the obtained F/F0.2 – γ curves approach a sort of envelope curve, which correspond to the plastic failure of the system. As far as the stiffener depth decrease, the separation from the envelope curve is representative of global buckling involving the stiffeners. Then, the postbuckling behaviour is characterized by tension field mechanism, which can be either of local or global type, depending on rib stiffness, which can be such to exclude them or not from the buckling mechanism. Also, typical collapse mechanisms of examined shear panels are shown in Figure 6. It can be noticed as for a given shear panel configuration (b/t = 100, hst = 40 mm) and shear deformation (γ = 10%), the deformation mechanism of the system is material dependent. In particular, the obtained out-of-plane displacements of shear panels show as the shear buckling phenomenon (of global and local type) is decreasing going from Fe360 steel to pure aluminium. In Figure 7, the shear strength levels F/F0.2 for all the examined cases are plotted as s function of the normalised stiffness parameter γst for a typical value of shear deformation γ = 7.5%. These curves allow the definition of an optimal value of the stiffness parameter (gst,opt ). In fact, such curves show that the shear strength of the system increases with the rib stiffness up reaching a maximum value identified by the curve plateau. Therefore, gst,opt can be defined as the one corresponding to the attainment of such a maximum value of the shear strength. Obviously, the optimal value of the stiffness parameter (gst,opt ) regarding the material under consideration, depends on the prefixed value of shear deformation level γ , which could be intended as the design deformation demand for shear panels. For each material, the results obtained for different shear deformation levels (γ ranging from 2.5% to 15%) can be summarized into single design charts (Figure 8), where for a given design value of shear strength F/F0.2 and shear strain γ , both the b/t ratio and the optimum value of the stiffness parameter gst,opt are obtained. In the same charts the limits related to the attainments of buckling phenomena are reported as well. In particular, the buckling limits placed on the right side of the diagram (large b/t values) 672

Figure 8. Design chart for shear panels.

are related to the attainment of elastic buckling (τcr ≤ τ0.2 ) (elastic buckling curve), which clearly represents a limit for the use of shear panels as a dissipative device. Obviously, such buckling phenomena could be either of local or global type, depending on the panel configuration. In particular, global buckling is more relevant for reduced shear deformation levels, where the applied ribs have a lower flexural stiffness. Shear panel configurations falling on the right of the above buckling curve can be defined as ‘slender’, meaning that they suffer buckling phenomena before being involved into plastic deformation. Similarly, the buckling curves depicted on the left side of the above charts (small b/t values) are representative of panel configurations where the buckling phenomena occur for shear stress (τcr ) equal or larger than the one corresponding to the attainment of the design deformation demand (τγ ) (plastic buckling curve). Shear panel configurations falling on the left of the above buckling curve can be defined as ‘compact’, meaning that they do not suffer buckling phenomena up reaching the required plastic deformation. As a consequence, shear panel configurations falling between plastic buckling curve and elastic buckling curve can be defined as ‘semi-compact’, meaning that suffer buckling phenomena while developing plastic deformation. From the comparison of the above diagrams in relation to the different materials, the following considerations can be drawn: – LYS and AW 1050 aluminium alloy shear panels clearly allow a better exploitation of the system characteristics, as it appears from the attained F/F0.2 ratios. They allow the use of reduced rib stiffness (γst ) and/or larger local slenderness ratio b/t for a given plastic deformation demand (γ ). It can be also noticed that in such a case all the analysed panel configuration fall into the ‘compact’ and ‘semi-compact’ range, and the plastic buckling curve (τcr ≥ τγ ) is moved towards larger values of b/t ratios. – For a defined shear deformation level and for a specific b/t ratio, AW 1050 aluminium alloy shear panels allow to obtain the maximum strength level F/F0.2 with the minimum value of the rib stiffness in comparison to the other materials. 673

– For a reduced strain hardening factor α, the obtained curves show a significant reduction of their variability range, emphasising the possibility of a reduced exploitation of the system post-elastic resources. 6 CONCLUSIONS In this paper the monotonic behaviour of stiffened shear panels made of four different metal materials has been examined. In particular, Fe360 mild steel, LYS steel, aluminium alloy AW 5154A and an almost pure aluminium (AW 1050A) have been examined. This study represents the preliminary phase of an experimental activity undertaken at the University of Naples ‘Federico II’ and aimed at investigating the possibility to employ pure aluminium shear panels as dissipative devices for passive seismic protection of framed buildings. On the basis of a non-linear finite element model, a wide numerical study, finalised to the evaluation of the influence of the adopted base material and the main geometrical parameters of shear panels, has been carried out. The obtained results allowed the definition of design charts for the selection of the optimum values of the geometrical parameters of the panels, namely width-to-thickness ratio and flexural rigidity of applied stiffeners. By the comparison of four metals, it appears that pure aluminium shear panels provide a better performance, due to the exploitation of their plastic characteristics in terms of both strain hardening and ductility, allowing the use of ribs with a lower flexural stiffness and also of larger b/t ratios. On the contrary, the traditional Fe360 steel exhibited a poor post-elastic behaviour owing to both a larger value of yield strength and a lower strain hardening. Also, aluminium alloy AW 5154A and LYS steel provide a similar performance, which is intermediate between the above ones. Finally, the numerical study allowed to identify for each material the ranges of b/t ratio were shear buckling phenomena take place in the elastic phase, during the plastic phase or after a predefined value of plastic shear deformation γ, allowing the correct definition of ‘slender’, ‘semi-compact’ and ‘compact’ shear panel classes. ACKNOWLEDGEMENTS Current activity has been developed in the framework of the research project “Innovative steel structures for the seismic protection of new and existing buildings: design criteria and methodologies” financially supported by the Italian Ministry of Education, University and Research (MIUR) for the years 2003-2005. REFERENCES Bleich, F. 1952. Buckling strength of metal structures. McGraw-Hill. De Matteis, G., Mazzolani, F.M. and Panico, S. 2003. Pure aluminium shear panels as passive control system for seismic protection of steel moment resisting frames. IV Conferenza internazionale STESSA (Behaviour of Steel Structures in Seismic Areas), Naples (ITALY). De Matteis, G., Addelio, F. and Mazzolani, F.M. 2004a. Sul dimensionamento dei pannelli di alluminio puro per la protezione sismica di telai di acciaio. XI Convegno Nazionale Anidis “L’Ingegneria Sismica in Italia”, Genova (ITALY). De Matteis, G., Mazzolani, F.M. and Panico, S. 2004b. Numerical analysis of pure aluminium stiffened shear panels for design optimisation. Proc. of CIMS ’04-Fourth International Conference on Coupled Instabilities in Metal Structures. Rome, Italy. Nakagawa, S., Kihara, H., Torii, S., Nakata, Y., Matsuoka, Y., Fujisawa, K. and Fukuda, K. 1996. Hysteretic Behavior of Low Yield Strength Steel Panel Shear Walls: Experimental Investigation. Proc., 11th WCEE, Elsevier, CD-ROM, Paper No. 171. Nakashima, M., Iwai, S., Iwata, M., Takeuchi, T., Konomi, S., Akazawa, T. and Saburi, K. 1994. Energy Dissipation Behaviour of Shear Panels Made of Low Yield Steel. Earthquake Engineering and Structural Dynamics, 23, 1299–1313.

674

Panico, S., De Matteis, G. and Mazzolani, F. M. 2003. Numerical investigation on pure aluminium shear panels. XIX Convegno C.T.A., Vol. 2, pp. 459–470, Genova (ITALY). Stowell, E.Z. 1948. Critical shear stress of an infinitely long plate in the plastic region. NACA Tech. Note. Tamai, H. et al. 1991. On hysteretic damper using low-yield stress steel plate installed in K-braced frame, part 1 and 2. Proc. Annu. Meeting of the Arch. Inst. Of Japan, 1447–1450, Tokyo (JAPAN). Tanaka, K., Torii, T., Sasaki, Y., Miyama, T., Kawai, H., Iwata, M. and Wada, A. 1998. Practical application of Damage Tolerant Structures with seismic control panel using low yield point steel to a high-rise steel building. Proc. Structural Engineering World Wide, Elsevier, Paper T190-4- CD-ROM. Timoshenko, S.P. and Gere, J.M. 1961. Theory of elastic stability, 2nd Ed. McGraw-Hill Book Co., New York (USA). Wada, A., Connor, J.J., Kawai, H., Iwata, M., Watanabe, A. 1992. Damage Tolerant Structures. Fifth US-Japan Workshop on the Improvement of building structural design and construction practices, ATC-15, California (USA).

675

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Applicability of variable stiffness seismic isolators based on magnetically controlled elastomer T. Isakovi´c & M. Fischinger University of Ljubljana, Faculty of Civil and Geodetic Engineering, Ljubljana, Slovenia

ABSTRACT: High damping rubber bearings HDRB are frequently used to isolate structures at major earthquakes. However, at lower displacement amplitudes related to more frequent weaker earthquakes the stiffness of HDRB is larger, jeopardizing the isolation effect. Of special concern is possible damage of sensitive equipment at earthquakes with short return period. To overcome this problem HDRB using magnetically controlled elastomers to provide uniform stiffness over the entire range of possible displacements were proposed. Based on the parametric study of the inelastic response of standard HDRB presented in the form of storey spectra the range of applicability of the proposed devices was investigated.

1 INTRODUCTION The main idea of the base isolation is to input into the structure special devices, which have the capabilities either to increase the flexibility of the structure or to increase its capability to dissipate the seismic energy. In both cases the seismic forces in the structure could be significantly reduced. First type of devices, due to its large flexibility increases the natural period of the structure, and consequently reduces the acceleration demand (see Fig. 1). The second type of devices reduces the seismic forces increasing the damping of the structure. Elastomeric high damping rubber bearings (HDRB) are often used for seismic isolation of different types of structures. These devices increase the flexibility of the structure as well as the damping of the structure (the damping of HDRB e.g. is typically 10–20% instead of the standard 5%). Therefore, it seems the HDRB are ideal devices for the purpose of the base isolation. However these devices have several deficiencies. One of those is discussed in the paper. It was observed that at lower displacement amplitudes, related to more frequent weaker earthquakes the stiffness of HDRB is larger, jeopardizing the isolation effect. Of special concern is possible damage of sensitive equipment at earthquakes with short return period. To protect the equipment under the weaker earthquakes, a new device, magnetically controlled elastomeric bearing has been developed. The development of this device is funded by the European Union through the 5th framework project VAST-IMAGE, where the consortium of industry research groups (MAURER Soehne-Germany, the principal investigator, TARRC-Great Britain, BIKANI Spain) has collaborated with the university and institute research groups (ENEL-HYDRO and ENEA, Italy, KTH – Sweden, ULJ-Slovenia, IFW Germany). The main idea of the new device is presented in the next section. To identify the cases where the new devices could be effectively used to protect the installed equipment, a parametric study of simplified structures isolated with the standard HDRB has been performed. Together with the discussion about the proper analytical models for standard HDRB under the smaller seismic load levels, the parametric study is presented in the third section. Results are analysed in the fourth section. 677

Sa

T0 – unisolated structure TI – isolated structure

F

keq,1(δ1)

keq,2(δ2) keq,3(δ3)

reduction of accelerations due to increased flexibility of the structure δ1

reduction of accelerations due to increased damping T0

δ2

δ3

δ

T

TI

Figure 1. Reduction of acceleration due to increase of the period and the damping of the structure.

Figure 2. Effective stiffness of the HDRB under various levels of load (authors: ENEA and ENEL-HYDRO).

Computer controlling mechanism

magnet

rubber + iron particles

Figure 3. Scheme of the new device.

2 MAGNETICALLY CONTROLLED ELASTOMERS 2.1 Theoretical scheme of the new device The increase of seismic demand of the equipment, due to the increase of the stiffness of the HDRB under the weaker earthquakes could be overhauled regulating the stiffness of the bearings depending on the seismic load level. The main scheme of the device, where the stiffness of the isolators would be kept constant regardless the variations of the seismic load level, is presented in Figure 3. Generally speaking, this device would consists of two main parts: the isolator and the control mechanism. Isolator also consists of two main parts. First one includes the rubber, where iron particles are added, and a second one is the magnet. The variation of the magnetic field influences the changes of the stiffness of rubber with iron particles. The variation of the magnetic field is regulated using the control mechanism, based on the level of the seismic load.

3 DESCRIPTION OF THE ANALYSED STRUCTURES AND PARAMETRIC STUDY 3.1 Cantilever beam and two types of isolators To identify the range of applicability of the new device, a parametric study of simplified structures isolated with the standard HDRB has been performed. Two sets of data have been used in the 678

200

(b)

M = 1000 t

150 K1

stiffness after "yielding": K1 = 995 kN/m

m << M force [kN]

(a)

initial stiffness: K0 = 8700 kN/m

K0

100 50 0

-0.2

-0.15

-0.1

-0.05

0

0.05

0.1

0.15

0.2

-50 -100 -150 -200

deformation [m] 80

h = 10 m

(c)

initial stiffness: K0 = 1354 kN/m

force [kN]

stiffness after "yielding": K1 = 565 kN/m

K1

60 40

K0

20 0 -0.12 -0.1 -0.08 -0.06 -0.04 -0.02 0 -20

0.02 0.04 0.06 0.08 0.1 0.12

-40

HDRB isolator

-60 -80

deformation [m]

Figure 4. Cantilever structure used in the parametric study.

actual behaviour bilinear model

Figure 5. Behaviour of HDRB at small and large strain rates.

analyses. In both cases the unisolated structure has the fundamental period of 0.5 s. In the first case a structure is isolated with the isolators having the properties presented in the Figure 4b. In this case the fundamental period of the isolated structure is 2 s at the design load level. For the second set of data, isolators with properties presented in Figure 4c have been used, so the isolated structure has the fundamental period of 1.5 s at the design load level. To identify the range of applicability of the new device, the acceleration storey spectra at the top of the cantilever structure (see Fig. 4a) have been studied (more explanation can be found in section 4). These spectra are obtained based on the results of the non-linear time-history analysis. 3.2 Earthquake load For non-linear time-history analysis, five accelerograms with acceleration spectrum equal to the Eurocode 8 elastic spectrum, for the soil type B has been used. The design level peak ground acceleration (PGA) of ag = 0.25 g for soil type A has been assumed. The peak ground acceleration has been varied from 0.025 g to 0.25 g with a step of 0.025 g, for each accelerogram. 3.3 Modelling HDRB under the small deformations Usually the HDRB are modelled using bilinear model presented in Figure 5. This model is suitable for the large strain rates, since the nonlinearities at the initial and at the unloading phase have small influence on the overall response. However, these nonlinearities are more important when the deformations of HDRB are smaller (see Fig. 5). 679

2000

300

100

1500

Improved model

1000

0 -0.015

-0.01

-0.005

0

0.005

0.01

0.015

Force [kN]

force [kN]

200

Common model

Common model Improved model

500 -02

-015 .

-0.1

0 -0.05 5000

0.05

0.1

0.15

0.2

-1000

-100

-1500

-200

-2000

-300 displ. [m]

Displacement [m]

Figure 6. Hysteresis loops for weak and design earthquake obtained with two models. Table 1. Summary of the results, obtained by nonlinear analysis. PGA∗ (g) 0.25 Average acceleration [m/s2 ] Average displacement [cm] ∗ PGA

1.31 12.6

0.225

0.20

0.175

0.15

0.125

0.10

0.075

0.05

0.025

1.12

0.97

0.84

0.71

0.60

0.49

0.39

0.31

0.23

8.8

7.3

5.9

4.6

3.4

2.2

1.3

0.6

10.5

– peak ground acceleration.

To estimate the response of an isolated structure, subjected to lower and moderate seismic load more realistically, an improved model, which takes into account described nonlinearities of HDRB has been studied. A model, which was developed by ENEL-HYDRO/ISMES (Rebecchi 2004) and which was included into program ABAQUS (Hibbitt, Karlsson & Sorensen, Inc., 2002) has been compared with the similar model in the program SAP2000 (Computers & Structures, 2003). Results obtained by the two programmes (ABAQUS and SAP2000) were compared. Since the differences were negligible, the simpler program SAP2000 has been used in the further analyses. Responses, using common bilinear model of HDRB and the improved model have been compared. As it was expected, the larger differences were observed for the weaker earthquakes. The hysteresis loops for the 20% and 100% of the design earthquake are presented in Figure 6. Similar conclusions have been obtained analysing other monitored quantities (e.g. the storey acceleration spectra at the top of the structure, which are described in the next section). 4 ANALYSIS OF RESULTS Since the results for two sets of data (see section 3.1) do not differ drastically, only the results of the nonlinear analysis corresponding to the first type of isolated structure (see Fig. 4b) are presented (see Table 1). The values of average accelerations obtained at the top of the cantilever as well as the average displacements at the top of the isolator are summarised. It is evident that neither the accelerations, nor the displacements corresponding to the smaller seismic intensity exceed the value corresponding to the design earthquake intensity (maximum expected intensity). However, it should be emphasized that decrease of the acceleration as well as displacements is smaller than the corresponding decrease of the load intensity. Based on the absolute accelerations obtained at the top of the investigated cantilevers, with the non-linear time history analysis, the acceleration story spectra have been calculated for each accelerogram and each intensity of the load (see section 3.2). Since the equipment itself can have different capabilities to dissipate the energy, six levels of damping have been considered (0%, 2%, 5%, 10%, 15%, 20% of critical damping). Then the average spectra for each intensity of the 680

Figure 7. Average acceleration storey spectra.

earthquake load have been determined. In Figure 7 the average spectra corresponding to different intensity of load and same damping are compared. Only the results for the first type of isolators (see Fig. 4b) are presented. The storey acceleration spectra represent the accelerations of the mass of the equipment (with different fundamental periods), installed at the top of the main structure (see mass m in Fig. 4a). In general the acceleration storey spectra characterise an amplification of the acceleration of the main structural system’s mass at the level of the installed equipment. Therefore it is reasonable to expect that the largest amplification would occur in the range, where the fundamental period of equipment is similar to that of the main structural system (resonant region). 681

In the analysed case the resonant region of story spectra for larger load intensities is around the period of T = 2.0 s, which is the equivalent period of the analysed structure, corresponding to the maximum design displacement. When the intensity of the load is reduced, the equivalent stiffness of the structure increases and the resonant region gradually moves to the region around T = 0.8 s, which is an approximate period of the structure at lower load level. The transition from the resonant region around 2 s to that around 0.8 s results in quite wide resonant region for medium level earthquakes (e.g. PGA = 0.15 g). However, for almost all periods in this region, the intensity of the load is smaller than that obtained at the maximum expected earthquake intensity. The exception from this rule is equipment with the period similar to the initial equivalent period of the isolated structure (calculated based on the initial stiffness of the isolator). In that region, it was noticed that the spectral acceleration is very close or even larger for very low earthquake intensities than that corresponding to the maximum expected earthquake. This is observed in the case of small damping (between 0% and 2% for the equipment). 5 FINAL REMARKS AND CONCLUSIONS It was observed that the maximum acceleration of the mass of the main structure, subjected to weaker earthquakes was in all cases smaller than that in the case of the design earthquake. However (as expected) the decrease in the acceleration of the mass was smaller than the decrease of the ground acceleration. The average storey acceleration spectra indicated that there were certain cases when the acceleration of the equipment was similar or even greater at weaker earthquakes than at strong earthquakes. This happened for equipment having relatively small damping and having natural period close to the period of the isolated structure at low level of earthquake intensity. The number of such cases was relatively small. The absolute values of accelerations under different earthquake levels were compared. It should be noted that the requirements for the seismic performance at low level (frequent) earthquakes were typically more stringent than in the case of major earthquakes. When this issue is taken into account, the number of cases, when the response of classically isolated structures is not adequate at low-level earthquakes, increases substantially. It was found that the increase in the (equivalent) stiffness at low amplitudes is typically related to the increase in (equivalent) damping. Therefore this two quantities should not be analysed separately, as it has been typically done in the past considering only the stiffness – amplitude relationship. This could make an illusion of extreme influence of varying stiffness on the response, which in fact may be less important, since the equivalent system is highly damped at these amplitudes. It has been found the inelastic model with constant unloading and reloading stiffness, which is typically used in the study of HDRB bearings and improved model presented in section 3.3 give quite different results in the case of weaker earthquakes. The differences between two models are less important, when the structure is subjected to a design earthquake. ACKNOWLEDGMENTS The work presented in the paper has been funded by the European Union through the 5th framework project VAST-IMAGE under the coordination of the Maurer, Germany as the principal investigator. REFERENCES Hibbitt, Karlsson & Sorensen, Inc., 2002, ABAQUS version 6.3, Users and Theory Manuals. Rebecchi Valter, 2004, User guidelines for HDRB UMAT routine working with ABAQUS/STANDARD, Version 1, Report VAST- IMAGE Document VIP-TR-EH-6-02. Computers & Structures, Inc., 2003, SAP2000 Nonlinear 8.2, Structural Analysis Program, Users Manual.

682

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Numerical analyses of the wall structure under conditions mining subsidence L. Szojda Department of Civil Engineering, Silesian University of Technology, Gliwice, Poland

ABSTRACT: A numerical analysis of complex stress state was provided on a base of compression and shear tests of wall specimens. Software package Mafem3D, which uses Finite Element Method, was applied for this analysis. In both cases the parameters of boundary surface have been determined in accompanying laboratory tests. Results of numerical compression and shear tests were compared with findings of laboratory experiments and proved proper conformance, especially on strength under compression, deformation, cracking. The wall structure represented a typical small building. The material parameters both for structure and soil were adopted from own laboratory tests and from tests of other scientists of the Silesian University of Technology. The subsoil deformation was applied on the bottom plane of the soil solid. A horizontal strain ε and vertical radius of curvature R of ground caused changes of stresses in foundation and wall structure. 1 INTRODUCTION Influences of non-mechanical loading on a building construction are introduced during design process. The temperature and shrinkage are the most known influences of this type but certain national standards demand special strengthening of construction against earthquake or mining subsidence like in Upper Silesia (south part of Poland). Damages of building construction increase during intensive coal excavation. First theories of surface deforming prediction under mining excavation were developed in the middle of last century. Parallely an observations of buildings were carried out to answer that additional loading. That made the basis to determine a strengthening of building structure on the threaten region. The aim of this paper is description of numerical analysis of soil-structure interaction. The calculation was made using elasto-plastic material model with isotropic hardening and softening. The analysis was provided by advanced numerical software package Mafem3D. 2 MATERIAL MODEL A software package Mafem3D developed by S. Majewski was used in all numerical analyses. The software is based on the non-linear algorithm of the finite element method and provides analysis of 3D elements. The elasto-plastic material model [3], [4] with associated isotropic hardening or softening rule and smeared crack representation is implemented in the software. The analyses are carried out by incremental-iterative procedure. For brick and mortar the plasticity surface is identical as the boundary surface as no hardening was taken into account for these materials. The boundary surface is formulated in the stress space determined by the mean stress σm (1), stress intensity σi (2) and Lode angle .

683

where: σx , σy , σz = normal stresses; τxy , τyz , τzx = shear stresses; σm = mean stress; σi = stress intensity In an elastic area non-linear elasticity, which was determined by tangential elasticity modulus Et and Poisson’s coefficient νt was assumed. The values of these parameters depend on the effort level (slev ), which is considered as a quotient between distances from mean stress axis to the point, which represents current stress-state and corresponding point on the boundary surface. The graphic illustration of this definition is presented in Figure 1. The plasticity surface for each material is defined separately for three regions. The first, biggest conical region includes mean compression stress zone and small part of mean tension stress zone (region A) and is described in meridian plane by a straight-line. In both remaining regions the meridians were described by circle shape, which are tangent to straight-line meridian from region A. The second region (B) consists of remaining part of mean tension stress zone. The third region (C) is a cap in the great mean compression stress. The above described boundary surface is presented in Figure 1. In meridian (aksiatoric) plane a trace of boundary surface in the region of mean compression stress (region A) is described equation:

where: FA = shape of surface in region A; α, β = coefficients depending on material strength parameters; Y1 , Y2 = softening functions in post-elastic phase of material work; ρ = coefficient, which describes a shape of deviatoric boundary surface section. From tension stress side the conical surface (region A) is closed by the circle cap in meridian plane (region B) and in high compression stress zone (region C). This region were described by equations:

where: FB , FC = shape of surface in region B and C; ct , cc = co-ordinates of cap circles centres on the mean stress axis σm respectively in tension and compression stress zone; rt , rc = radius of cap circle respectively in tension and compression stress zone.

Figure 1. The shape of boundary surface for material, which was used in numerical analysis, in deviatoric and meridian (aksiatoric) plane; where: fccc = strength under three-axial compression; fttt = strength under three-axial tension.

684

In the deviatoric section the plasticity surface consists of three mutually tangential curve segments according to Willam and Warnke proposal [8]. Three ellipses creating the surface in deviatoric section can be considered as a smooth approximation to the Coulomb-Mohr criterion (Figure 1 – trace of the surface in deviatoric plane).

3 DETERMINATION OF BOUNDARY SURFACE Common tests of brick wall specimens according to Polish Standard (PN) are not sufficient for estimation of material failure criteria in complex stress-state. Therefore three-axial compression tests separately for brick and mortar on cylinder specimens with diameter 60 mm and high 120 mm were carried out. The three-axial compression equipment similar to that used in soil tests provides the uniform horizontal (radial) stress σrad on side surface of a specimen. The specimen was destroyed in two ways: • due to vertical stress (σver ) at different levels of constant horizontal compression (σrad ), • due to horizontal (radial) compression (σrad ) at constant vertical compression (σver ). In the first way the stress path in the plane of tension meridian ( = 0◦ ) with stress-state determined as follows:

Loading according to the second way generates the stress is situated in the plane of compressive meridian ( = 60◦ ) as the stress-state is defined as:

Loading paths for brick in lab tests with constant radial stress (σrad = const) and constant vertical stress (σver = const) are shown respectively in Figures 2 and 3. The failure surface with planes of stress-paths is presented in Figure 4 with a view of boundary surface from camera situated in high mean compressive stress zone. Internal planes present location of stress paths, which were defined in three-axial compressive apparatus. This figure is three-dimensional interpretation of Figure 1. Ten cylinders of brick and mortar were destroyed by vertical compression stress (σver > σrad ) and six cylinder specimens were destroyed by radial stress (σver < σrad ). In the first case the maximal value of stress intensity σi at adequate mean stress σm were concentrated along the compression meridian, while in the second case along the tension meridian. All maximal points of each loading path were presented in the region of boundary surface,

Figure 2. Loading paths for brick specimens at dif- Figure 3. Loading paths for brick specimens at ferent levels of constant radial stresses (σrad = const), different levels of constant vertical stresses (σver = according to equation (6). const), according to equation (7).

685

Figure 4. Loading stresses inside the boundary surface for tests.

Figure 5. Main meridians (straight segments) of boundary surface for mortar.

Figure 6. Variability of Poisson coefficients ν for mortar and brick.

which was described by straight line (region A in Figure 1). The strength of uni-axial compression state (fc ) is described by σrad = 0, while in biaxial (fcc ) by σver = 0. In Figure 5 the maximal values of stress intensity σi for all performed tests for mortar have been approximated by straight segments of main boundary surface meridians. This relation was described by the equation (3). On the side surface of cylinder specimens two pairs of electroresistance gauges, registering the longitudinal as well as lateral strain were located. On this base the variability of elasticity modulus E and Poisson coefficient ν (Fig. 6) in elastic phase for both materials were described. Variability of Poisson coefficient ν, which was achieved in tests for mortar and brick on relation to maximal vertical stress, was shown in Figure 6. On this base ν function in elastic phase was determined. In the same way the variability of elasticity modulus E was described. The values, which were assumed in the numerical analyses, were presented in Table 1 (basing on tests, which were presented in [5] and [6]). 4 MODEL OF BRICK WALL SPECIMENS FOR NUMERICAL ANALYSIS Numerical analyses were carried out for a fragment of brick wall structure. Two independent models were performed. The first model imitated a typical wall specimen (according PN), which was used to define a compression strength of wall. The second one was applied to determine a shear strength 686

Table 1. Material parameters determined in three-axial tests. Value Parameter 1 2 3 4 5 6

Strengthening in uni-axial compression Strengthening in biaxial compression Compression meridian Tension meridian

Brick

Mortar

fc [N/mm2 ] fcc [N/mm2 ] αc βc αt βt

11.49 12.46 0.3081 3.0139 0.1833 2.3578

7.39 8.30 0.3116 1.9578 0.1978 1.3281

Eo [N/mm2 ] νo [–]

1988.9 0.062

8555.7 0.161

Initial parameters in elastic phase 7 8

Initial value of elasticity modulus Initial value of Poisson coefficient

Figure 7. Lab specimen and numerical model with Figure 8. Lab specimen and numerical model for finite elements mesh for compression tests. shear tests.

and shear angle of brick wall. For all laboratory tests ‘full brick’ typical for Polish masonry were used. Dimension of these bricks are 25 × 12 × 6.5 cm. Compressive strength for this part of bricks was determined on level 15.5 N/mm2 . All specimens for three-axial tests were prepared from one portion of mortar. Proportions of dry components were about 1:1:6 (Portland cement 250:lime: sand). The same proportions were used in specimens for compressive as well as for shear tests. In Figure 7 the view and dimensions of a laboratory sample according to [1] as well as the adequate numerical model are presented. The numerical model comprised 2660 nodes and 1944 cubic elements. The dimensions of elements in brick wall specimen differed from 5.0 to 16.25 mm, and reached 20.00 mm in elements, which simulated the plate of loading machine. In the first step of loading the own weight of structure was applied, while in next 30 steps the numerical model was loaded by vertical displacement of nodes situated in the upper plane of the strengthening machine plate. The total displacement in last step (31) of loading was 2 mm. In Figure 8 brick-wall specimen, which was used in shear tests and the adequate numerical model with finite elements mesh are presented. In the laboratory test one of vertical edges of a specimen was moved in parallel to the other one. The numerical model was exactly adequate to the lab specimen (1.290 × 1.415 m) and consisted of 5130 nodes and 2464 elements. Dimensions of finite elements differed between 10.0 mm (mortar) and 55.0 mm (brick). In the numerical model the concrete elements connecting a specimen with laboratory equipment were considered too. In laboratory tests the right column was pushed up vertically. In numerical model this effect was simulated by parallel, vertical displacement of the right edge of specimen. The maximal vertical displacement was 5mm. This value was divided in to 49 loading steps (no 2 to 50) and that means that in a single step the 0.102 mm vertical displacement was applied. In the first step only the own weight of structure was regarded. 687

Figure 9. Map of effort level for specimen, last step before destruction.

Figure 10. Directions of main tensile stress (σ1 ) and effort level in finite elements for model.

5 RESULTS OF NUMERICAL ANALYSES Comparison of numerical and laboratory results proves the reliability of the proposed material model. The basic comparative parameters between lab and numerical compression tests were strains observed on the same base points, stress levels on cracking and final compression strength. The first crack of compressive lab model was observed at vertical stress σcr,labAV = 2.88 N/mm2 , while in numerical model σcr,numAV = 4.44 N/mm2 . The highest vertical stress noted down for both models were respectively fc,labAV = 6.21 N/mm2 and fc,numAV = 5.99 N/mm2 . The map of effort level in finite elements in the load step foregoing the destruction of a specimen is shown in Figure 9 (the symmetry planes are located along left and bottom edge of specimen). The destroyed (cracked) elements are signed in black colours. On Figure 10 the picture of effort level for model under shear test in 22 loading step, without averaging between neighbouring finite elements is presented. The directions of main stresses for each element are presented as well. The direction of a crack is always perpendicular to main tensile stress σ1 . The region with highest effort level is formed along the diagonal of the specimen (dark elements). Black colour in the picture indicates these elements whose stress vector has reached the boundary surface. Due to smeared crack representation assumed in the material model only the area threatened by cracking can be revealed. This area as well as the direction of main stresses very well corresponds with the real crack pattern in lab tests. 6 NUMERICAL ANALYSIS OF SOIL-STRUCTURE INTERACTION Soil-structure system consists of soil shape and brick-wall structure. Numerical model was presented on Figure 11. All structure was divided on cuboid 3D finite elements. Volume of each element consists of only one kind of material. Soil-structure system was built from different structure elements and different material. Main part of structure was made from brick wall but foundation and curb-plate were strengthen by reinforced concrete. The roof and ceilings in this structure were built from a wood and their strength and stiffness were very low. In this case, those elements were omitted in calculation. Material parameters for concrete, soil and reinforce were adopted from other work [7]. Numerical model consisted on 4424 nodes, 2895 cuboid elements and 668 bar elements. Due to symmetry only ¼ of all numerical model was analysed. The loadings of numerical model were 688

Figure 11. Numerical model of soil-structure interaction.

Figure 12. Normal stresses σx in the 6th and 16th Figure 13. Normal stresses σy in the 6th step of steps of loading (foundation of the structure). loading (foundation of the structure).

divided into two categories: mechanical (material weigh, programme loading) – first step of loading and non-mechanical – soil deformation – next 20 loading steps. Soil deformation were adopted form mining subsidence. Bottom plane of soil shape was deformed according to horizontal strain ε = 6‰ and radius of curvature R = 6 km. This parameters are counted to III category of mining subsidence in Poland [2]. Deformation were applied along one axe X only. Extreme soil displacement appeared in 6th (extreme elongation of horizontal strain ε and minimal radius of convex curvature R) and 16th step of loading (extreme abbreviation of horizontal strain ε and minimal radius of concave curvature R). 7 RESULTS OF NUMERICAL ANALYSIS OF SOIL-STRUCTURE Numerical analysis allowed to describe stress-state in each finite element of the whole structure, in each step of loading. In the consequence of horizontal strain ε of a soil new horizontal stresses appear in the structure of foundations. Normal stresses σx and σy for extreme steps of loading (6 and 16) are presented on the Figures 12 and 13. Influence of simultaneous soil displacement of horizontal strain and curvature is observed in whole structure. The map of stress intensity σi was presented on the Figure 14. On this picture only half of the external longitudinal wall (to the plane of symmetry) was shown. Main concentration of stress intensity appeared on the connection between brick-wall and concrete elements. Dark fields shall be read as the highest level of stress on this structure. 689

Figure 14. Stress intensity σi on 6th step of loading (external longitudinal wall).

8 SUMMARY AND CONCLUSION The numerical analyses, provided by software package Mafem3D, give promising convergence with laboratory tests. Good convergence was observed in the crack distribution (Figs. 9 and 10) and interdependencies σver −εver of compression model. The area of highest effort level in finite elements is adequate to cracking regions in laboratory models. Precision of structure modelling allowed to build advanced virtual numerical model. Soilstructure system enables to take parameters of soil and structure separately. Non-mechanical loading, which was provided to the bottom plane of soil solid, proved increase of normal stresses in foundation (σx , σy ) and stress intensity σi in different places of the whole structure. This type of numerical analysis interaction gives promising results for prediction of hazardous regions in real structure. Nonetheless the numerical analysis based on well verified, advanced elasto-plastic material models provides useful tool for detailed analysis of masonry structures, supplementing and enriching rather than replacing still necessary, but laborious and expensive laboratory tests. REFERENCES [1] Drobiec Ł., Jasi´nski R., Kubica J., 2000 Preliminary Tests of Different Causes of Cracking Hardening of Compression Walls. (in Polish) Final report BK-259/RB-2/97, Silesian University of Technology, Faculty of Civil Engineering, Department of Building Structures, Gliwice. [2] Knothe S. 1953 The equation of the finally formed hutch subsidence profile, Archive of Mining and Metallurgy, t.1, z.1. [3] Majewski S. 1995 Elasto-Plastic Cooperation Model of Ground-Structure under Mining Subsidence. (in Polish), Exercise Science Book of Silesian University of Technology, No 1271, Civil Engineering Z.79, Gliwice. [4] Majewski S. 1997 Elasto-Plastic Model with Isotropic Hardening/Softening Rule for Cohesive-Frictional Materials, Proceedings of the 5th International Conference of Computational Plasticity COMPLAS, Barcelona, Vol. 2. [5] Szojda L., 2001 Numerical Analysis of Cooperation of Brick Wall Structure under Ground Deformation (in Polish), (doctoral thesis) Gliwice. [6] Szojda L. 2001 Possibilities of a Numerical Analysis of Brick Wall Structure (in Polish), XLVII Science Conference KILiW PAN i KN PZITB Krynica , t3. [7] Wandzik G. 1999 Numerical analysis of punching of reinforce concrete plate (in Polish), (doctoral thesis) Gliwice. [8] Willam K. J., Warnke E. P. 1975 Constitutive Models for the Triaxial Behavior of Concrete, IABSE Seminar Bergamo in IABSE Proc. Vol.19.

690

Session 12: Research and development concerning mixed building technologies

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Composite joints – rotational capacity Ulrike Kuhlmann & Martin Schäfer Institute of Structural Design, University of Stuttgart, Germany

ABSTRACT: Until now composite joints are designed as pinned or as continuous joints. A continuous joint requires stiffeners in the column and high degrees of reinforcement, whereas a pinned joint requires some minimum reinforcement for crack width control that however is not considered to provide any moment capacity. Optimised solutions are partial-strength joints that consider the moment resistance and that allow easy and cheap producible joints with small degrees of reinforcement and without stiffeners in the column. Using partial strength joints, in the ULS in structural systems a moment redistribution occurs resulting in a certain rotation in the joint, the so-called required rotational capacity. It has to be taken care for that the available rotational capacity of the joint is sufficient in comparison to the required rotation. In this publication means are given in order to calculate the required and the available rotational capacity. KEYWORDS:

Composite Joints, Rotational Capacity, Semi-Rigid Joints, Partial-Strength Joints.

1 INTRODUCTION Figure 1b shows an optimised joint solution by using partial-strength joints that are easy and cheap producible with low degrees of reinforcement and without stiffeners in the column: these joints have a certain moment resistance which is in general smaller than the moment resistance of the beam, so they are partial-strength joints. Using partial strength joints, in the ULS in structural systems a moment redistribution occurs resulting in a certain rotation in the joint, the so-called required rotational capacity j,req , see Figure 2. It has to be taken care for that the available rotational capacity j,avail of the joint is sufficient in comparison to the required rotation j,req .

additional web plate

a)

b)

Figure 1. Continuous (a) and partial-strength (b) composite joint.

693

j,req j,req  j,avail

Figure 2. Verification of the rotational capacity. Table 1. Typical required rotational capacities in braced frames according to (Kattner 1999) where L = span; h = beam height; g = dead load; q = life load. q g

unpropped during construction L/h = 15

L/h = 30

j,req shear conn.

η[%]

1,00

0,95

0,90

0,85

0,80

1,00

0,95

0,90

0,85

0,80

27

18

14

10

7

44

32

25

20

17

S355

monolithic

&

fully conn.

100

37

22

17

13

8

86

54

41

30

24

C35/45

partial conn.

70

28

20

15

10

7

70

38

28

21

16

S235

monolithic

22

14

9

8

5

32

26

21

17

14

&

fully conn.

100

29

18

13

10

7

57

43

34

26

21

C25/35

partial conn.

70

23

17

13

9

6

44

31

23

18

14

In this publication means are given how to calculate the required rotational capacity j,req depending on the structural system and the loading. Secondly the available rotational capacity j,avail of composite joints is also evaluated. The available rotational capacity depends on the geometry of the joints that has an influence on the ductility as can be seen in the following.

2 REQUIRED ROTATIONAL CAPACITY 2.1 Composite frames with bracings (Kattner 1999) calculated the required rotational capacity of composite frames that are fixed horizontally, see Table 1. The required rotational capacity mainly depends on the degree of plastification ω in the area of the maximum sagging moment. To activate the full plastic moment in the beam Mpl,Rd partial plastifications occur in the beam (yield zones in Figure 3b). This leads to large curvatures and large declinations α at the ends of the beam where it is connected by the partial-strength joints to the columns. Provided that the columns are not inclined, which is usually the case for braced frames, the required rotation j,req is identical to the declination α. So large rotational capacities j,req are demanded by the system when plastifications of the beam in the sagging moment zone appear. 2.2 Composite frames without bracings In composite frames with partial-strength joints a plastification of the joint only on one side of the column occurs. On the other side of the column only a small elastic rotation within the joint appears, see Figure 4. As the curvature of the beam caused by vertical loads in the unbraced system 694

q1 1 Mj,pl,Rd

Mj,pl,Rd

EIb

a) ELASTIC

q2 Mj,pl,Rd

2

Mj,pl,Rd

b) YIELD ZONES Mj,pl,Rd

Mj,pl,Rd

Case a) ELASTIC: Msag,d (q1)  Mel,Rd

Case b) PLASTIC: Msag,d (q2)  Mel,Rd c) MOMENT DISTRIBUTIONS

Figure 3. Elastic (a) and plastic (b) curvature of a beam and the respective moment distribution (c).

plastification of the joint j,L  αL

j,L

only small elastic joint rotation  j,R

R

j,R

L L 



(a) Mj,L = Mj,pl,Rd M Mj,R

Mj,pl,Rd

Mc,d

Mc,d  M Msag,d

(b) Figure 4. Composite frame without bracings.

is similar to that of the beam in a braced frame, and as the total rotation is concentrated on only one joint, so roughly the double rotation is required. Further, the small rotations j,R of the elastic joint lead to an inclination ψ in the column, that is of similar size to the declination of the beam αR under vertical loads, see Figure 4. Caused by these large deformations when increasing the load, due to second order effects a collapse by instability 695

required joint rotation
j,req [mrad]

70

unpropped propped during construction

60 50 40 30 20 10 0 0

2

4

6 8 life load p [kN/m²]

10

12

14

Figure 5. Required total joint rotations j,req for a several typical composite frames without bracings (3 bays, 3, span varies from 8 to 16 m), provided that no plastification of the beam in the sagging moment area occurs.

may even occur. In general at the maximum load the full plastic moment Mpl,Rd of the composite beam in the sagging moment zone is not achieved and very large rotations occur. As the sagging moment Msag,d exceeds the elastic moment Mel,Rd of the beam, see Figure 3b partial plastification of the beam (yield zones) cause these large overall deformations. The sagging moment Msag,d should therefore be limited to 1.1 Mel,Rd , see (Lahmeyer 2003), (Kuhlmann & Schäfer 2003) where Mel,Rd is the elastic moment just before the first yield zones in the lower beam flange occur. By this limit stability failure of the unbraced frame is prevented. In fact only a small moment capacity of the partial-strength joint is required to form a frame and to carry the horizontal loads that are due to wind, imperfections and second order effects.

2.3 Influence of the construction method The construction method also influences the rotations in the joint. For the load of the wet concrete sometimes additional temporary columns (props) are used to support the steel beam. For a beam that is unpropped during construction, large deflections of the beam occur under the load of the wet concrete due to the relatively small inertia of the steel beam in comparison to the composite beam and due to a small restraint moment of the pure steel joint. These deflections lead to large inclinations α at the support of the steel beam and to large rotations in the steel joint. Some data are given in Table 1 (braced frame) and Figure 5 (non-braced frame, without plastifications in the beam). It can clearly be seen that independent of the load, a propped construction requires between 10 and 20 mrad whereas unpropped systems require rotation capacity values of up to 60 mrad and more. For an unpropped structure these high rotations mainly occur during the construction in the steel joint whereas the required rotations of the composite joint are reduced as only the life load is acting on the composite system.

3 AVAILABLE ROTATIONAL CAPACITY 3.1 General The joint properties stiffness Sj,ini and moment capacity Mj,pl,Rd may be calculated according to (ENV 1993/A2 1998) or (prEN 1993-1-8 2003) for steel joints, (ECCS Doc. 109 1999) and (prEN 1994-1-1 2003) for composite joints. Concerning the verification of the rotational capacity only some regulations are given that are yet not satisfactory. 696

Components: RFT

KB

RFT reinforcement in tension KB slip of shear studs

BT EPT

EPT endplate in bending

CWT CFT CWS

BT bolts in tension CFT column flange in bending CWT column web in tension

BFC

C CWC C

CWS column web in shear BFC beam flange in compression CWC column web in compression

Figure 6. Component model of a composite joint according to (ECCS Doc. 109 1999).

3.2 Component model To analyse the behaviour of the composite joints, the joint is divided into its components, see Figure 6. The available rotational capacity is dependent on the components’ properties and on the geometry of the components. As could be seen in the tests, e.g. (Kuhlmann & Schäfer 2003), the plastic behaviour of the joint concerning its rotational capacity is governed by the behaviour of the failing component. The other stronger components usually remain elastic and only give small contributions to the rotational capacity. In this case the available rotational capacity j,avail may be simplified to:

where i = index for the failing component; wu,i = deformability of the failing component [mm]; zi = lever arm [mm] of the failing (tensile/compression) component to the centroid of the components. In the following paragraphs the ductile behaviour of the components is surveyed. 3.3 Components 3.3.1 Reinforcement in tension (RFT) (ECCS Doc. 109 1999) gives rules how to calculate the deformability of the component reinforcement in tension. Nevertheless (ECCS Doc. 109 1999) does not sufficiently account for the influence of the material properties of the reinforcement as quoted below: The ductility of the reinforced concrete is determined by the strengthening factor fu /fy of the reinforcement and by the ultimate strain εu of the steel material, (Kuhlmann & Schäfer 2003). Analysis of various component tests show that the strengthening factor fu /fy should be above 1,13 for ductile reinforcement steel. As tests on composite joints (Kuhlmann & Schäfer 2003) have shown, even small degrees of reinforcement (0,45%) may then lead to large deformabilities of the reinforced slab and to high rotational capacities of the composite joint. 3.3.2 Bolts in tension (BT) For high grade bolts the ductility is mainly dependent on the strengthening factor fu /fy . For 10.9 bolts with small strengthening factors a plastification occurs only in the thread area and not in the shank. The failure mode “thread stripping” exhibits very small deformabilities. Thus full threaded bolts provide higher ductility, avoiding thread stripping. Detailed investigations can be found in (Steurer 1996 &1999). 697

3.3.3 Endplate in bending (EPB) (ENV 1993/A2 1998)Annex J or (prEN 1993-1-8 2003) give recommendations to limit the thickness of the endplate in order to prevent brittle failure of the bolts and to obtain ductile joints by yielding of the endplate in bending. These rules do not guarantee a ductile failure and large rotational capacities as the bolt may be forced to bend resulting in an early brittle failure of the bolt, see Figure 7 (left). A thicker bolt in this case changes the failure mode and enlarges the rotational capacity, see Figure 7 (right). Here further research is needed especially concerning high deformable joints. 3.3.4 Column web in compression (CWC) (Kühnemund 2003) analysed the force deflection curve of the component column web in compression: After reaching the maximum load a long descending branch (Figure 8) is following.

Figure 7. Brittle failure (left) and ductile failure (right) of a steel joint, (Kuhlmann & Schäfer 2003). Fc,wc

Fc,wc,u Fc,wc,dec; wc,wc,decr

Fc,wc,pl Fc,wc,el

wc,wc,el

wc,wc,pl

elastic

strengthening

wc,wc

wc,wc,u decending branch

plastic

Figure 8. Typical force deflection curve of the component “column web in compression” (CWC).

698

This descending branch is caused by the buckling of the web. Thus a relatively large ductility may be taken into account, if the calculative compression resistance is limited. In (Kühnemund 2003) analytical equations based on mechanical models and describing the load deflection behaviour are given. Rotation capacities of steel joints are analysed under consideration of under- and overstrength effects. 3.3.5 Column web in shear (CWS) According to (ENV 1993/A2 1998) the component column web in shear is ductile. Nevertheless in the load introduction zone the web panel is also subjected to compression forces so that a similar behaviour to that of the component column web in compression may occur. 3.3.6 Beam flange in compression (BFC) (Kuhlmann 1986) et al. analysed the requirements for the application of the plastic hinge theory for continuous beams. The buckling behaviour of the lower beam flange in the hogging moment zone may limit the moment capacity of the beam. The slenderness flange width/flange thickness of the beam is limited to allow for a full plastic moment and a plastic moment redistribution in the continuous beam. 4 VERIFICATION According to (Kühnemund 2003) and (Steenhuis, M & van Herwinen, F & Snijder, H 2000) an independent verification of the rotational capacity may be conducted as follows:

where j,req = required rotation, calculated with design load [mrad]; j,avail,R = available rotation of the joint (characteristic value) [mrad]; γM = partial safety factor for the rotational capacity [–] Another possibility for a simultaneous verification of the moment and the rotational capacity of the joint is the beamline method. By varying e. g. the moment resistance of the joint Mj,pl,Rd , the required rotation of the joint j,req that refers to the moment of the joint Mj,d is calculated. The herewith gained results form a line, the so-called beamline, see Figure 9. In general, the beamline forms a straight line. This is even the case for unbraced frames where second order effects occur, see (Kuhlmann & Schäfer 2003). The beamline deviates from the straight line when a change of the structural system occurs, e.g. yielding of the beam in Figure 9.

Joint moment Mj,d [kNm]

800 700 600 Yielding of the beam in the sagging moment area

500 400

Beamline Joint configuration 1 Joint configuration 2

300

Joint configuration 3

200 100 0 0

5

10

15

20

25

30

35

Joint rotation j [mrad]

Figure 9. Typical beamline of an unbraced composite frame calculated according to the yield zone theory and typical moment rotation curves of composite joint configurations (Kuhlmann & Schäfer 2003).

699

If an intercept point of the beamline and the moment rotation curve of the joint exists, both resistance and rotation requirements may be satisfied. In the case of joint configuration 2 no interception point occurs due to the lack of rotational capacity whereas for the joint configuration 3 a lack of moment capacity of the joint causes the failure of the system. The beamline allows for a direct optimisation of the joint within a system. The third possibility is to ensure a sufficient rotational capacity of the joint by construction rules, the so-called deemed-to-satisfy criteria. See e. g. some basic results in (ENV-1993/A2 1998).

5 FUTURE DEVELOPMENTS Further results are demanded to obtain reliable values for the deformability wu,i of all of the components of the joint. Overstrength of ductile components may prevent a planned yielding of the ductile component and may lead to an unforeseen brittle failure of another component. Thus the rotational capacity of the joint may be reduced. Further probabilistic investigations like (Borges 2003) are needed to cover that effect and to obtain reliable values for the available rotational capacity j,avail,R , for the safety factor γM and for deemed-to-satisfy criteria. These aims form part of a general strategy to develop adequate ductile joints for structures in seismic regions and ductile joints for robust structures to withstand catastrophic scenario. REFERENCES CODES ENV 1993-1-1/A2. Eurocode 3 1998. Design of steel structures, Part 1-1: General rules and rules for buildings, Amendment A2, Annex J. prEN 1993-1-8 2003. Eurocode 3. Design of steel structures. Design of joints. prEN 1994-1-1 2003. Eurocode 4. Design of composite steel and concrete structures. Part 1-1: General rules and rules for buildings. ECCS Document No. 109 1999. Anderson, D & Aribert, J. -M. & Bode, H. & Huber, G. & Jaspart, J.-P. & Kronenberger, H.-J. & Tschemmernegg, F.: Design of composite joints for buildings. 1st Edition. RESEARCH Borges, L. 2003: Probabilistic evaluation of the rotation capacity of steel joints. Master thesis, Faculdade de Ciências e Tecnologia, especialidade de Esterturas. Coimbra. Kattner, M. 1999: Beitrag zum Entwurf von Rahmen mit Verbundknoten im Hochbau. Thèse MK No. 2055, Ecole Polytechnique Fédérale de Lausanne. Kuhlmann, U. 1986: Rotationskapazität biegebeanspruchter I-Profile unter Berücksichtigung des plastischen Beulens, Ruhr-Universität Bochum, Institut für Konstruktiven Ingenieurbau, Mitteilung Nr. 86-5, Dissertation. Kuhlmann, U. & Schäfer, M. 2003: Research report: “Innovative verschiebliche Verbundrahmen mit teiltragfähigen Verbundknoten”. Funded by Forschungsvereinigung für Stahlanwendung. Institute of Structural Design, University of Stuttgart. Kühnemund, F. 2003: Zum Rotationsnachweis nachgiebiger Knoten im Stahlbau. PHD thesis, Institute of Structural Design, University of Stuttgart, Proceedings No. 2003-1. Lahmeyer, J. 2003: Untersuchungen an verschieblichen Verbundrahmenelementen mit teiltragfähigen Verbundknoten. Diploma thesis, Institute of Structural Design, University of Stuttgart, Proceedings No. 2003-5x. Steenhuis, M. & van Herwinen, F. & Snijder, H. 2000: Safety concepts for ductility of joints. International AISC/ECCS workshop on connections in steel structures IV, Roanoke, Virginia, USA. Steurer, A. 1996: Trag- und Verformungsverhalten von auf Zug beanspruchten Schrauben. IBK-Bericht Nr. 217. Institut für Baustatik und Konstruktion, ETH Zürich. Birkenhäuser Verlag Basel. Steurer, A. 1999: Tragverhalten und Rotationsvermögen geschraubter Stirnplattenverbindungen. IBK Bericht Nr. 247, Institut für Baustatik und Konstruktion, ETH Zürich. Birkenhäuser Verlag, Basel.

700

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Characterization and qualification of reinforcement by composite of the reinforced concrete labs D.T. Nguyen, J.F. Georgin, A. Limam & J.F. Reynoard National Institute of Applied Sciences of Lyon, Lyon, France

ABSTRACT: The objective of this study is to evaluate the behavior of the reinforced concrete (RC) two-way slabs pre-damaged under membrane and flexural loads, then repaired and retrofitted with TFC (Tissue of Fibers of Carbon). This paper presents the result of an experimental test. In the first time, the large-scale reinforced concrete shell element was pre-damaged under membrane force and bending moment. After, the shell element was repaired by the joining of TFC on the tensile face. Then, the slab retrofitted was tested until the ruin. The failure mode was the detachment of the external TFC strips from the slab. KEYWORDS: retrofit.

cooling tower, two-way slabs, shell, reinforcement, TFC composite, repair,

1 INTRODUCTION In the past few years, several major international research programs was lunched to investigate the feasibility of using technologies of polymer composite for repair and retrofitting of structural systems. One of these successful structural applications is the external composite strengthening for repair and upgrade of the structural capacity and rigidity of axially loaded concrete members (column, for example) in both seismic and corrosive environment. Recently, this application of composite has been extended to other RC structural elements: beams [1] and slabs [2]. In this process, several laminates of composite are bonded to the finished concrete surface in the hoop or longitudinal direction for enhancing the flexural capacity of these members. The experimental program supported by EDF (Electricity de France) focuses on investigation of the behaviors of large-scale RC slabs retrofitted by TCF for the purpose of the application of this reinforcement process to the RC shell of hyperbolic cooling towers. In fact, several cooling tower of the power station of Electricity de France (EDF) exhibit a state of mechanical degradation which is characterized by presence of geometrical defects or by presence of shell’ cracking. An experimental investigation was undertaken involving the testing of large-scale RC shell element. The objectives were to observe the phenomenological behavior of shell elements subjected to combined membrane and bending conditions and to provide data useful in checking the accurateness of non-linear finite element models [3]. The shell element tester developed at INSA Lyon is used now to study the reinforcement in TFC of square RC slabs. 2 EXPERIMENTAL PROGRAM 2.1 Test specimen and material properties The test specimen is two-way RC slabs. The overall dimensions are 2 m by 2 m and the thickness is 10 cm. There are 48 holes of 32 mm of diameter which are provides in order to connect the loading devices on the specimen (Fig. 1). 701

Y

Parts for handling Supplementary reinforced zone

445

1

50 50

100100

2

50 50

3 X

1

4

250

5 250

110

2000

110

70110 70

100100

7011070

390

2000

390

110

250

6

250

445

250

250 125

8 250

250 125

1

750

7

250

100

Section 1-1

2000

Figure 1. Detail of two-way reinforced concrete slab. 700

STRESS (MPa)

600 500 400 300 200 100 0 0

0,02

0,04

0,06

0,08

0,1

0,12

0,14

STRAIN

Figure 2. Stress–strain curves for reinforcing steel. Table 1. Material properties of steel reinforcement bars. E (GPa)

fy (MPa)

εy (10−3 )

fu (MPa)

εu (10−3 )

218

547

2,51

616

30,95

The steel reinforcement is composed of two layers of 6 mm diameter bars spaced at 10 cm on two orthogonal directions parallel to the edges of the slab. Within a distance of 35 cm from the edges, the supplemental steel reinforcement bars were placed in order to prevent local failure due to the stress concentrations caused by the loading devices. Steel reinforcement bars exhibited ductile behavior with a bilinear stress–strain curve as shown in the Figure 2. The material properties of the steel bars, obtained from tests tubes are summarized in Table 1. 702

Table 2. Composition in weight for 1 m3 of concrete. Cement (kg)

Water (l)

Sand (kg)

Gravel (kg)

320

190

765

1110

Table 3. Parameters of the TFC [4]. Ply thickness Longitudinal tensile strength Transversal tensile strength Longitudinal modulus of elasticity

0,43 mm 1400 MPa 580 MPa 105 GPa

Figure 3. General view of the test set-up.

The concrete used was designed with cement Lafarge CPA 55, sand of 0–5 mm and gravel of 0–12,5 mm. The composition of the concrete is given in Table 2. The characteristics of the concrete were defined by tests of characterization in compression on the cylindrical test-tubes of 11 × 22 with a press of 5000 kN controlled in force. The resistance of compression is about 39 MPa. The resistance of traction is 3MPa and Young’s modulus (E) has a value of 35 GPa. The TFC consists 70% of fiber in the direction of chain (longitudinal direction) and 30% of fiber in the direction of weft (transverse direction). The characteristics of the composite are presented in Table 3. The behaviour’ law of the TFC is elastic perfect until the rupture. The TFC is conditioned in rollers. The standard widths available are 40, 75, 150, 200 and 300 mm. We used the strips of 300 mm of width. 2.2 Test set-up The test was conducted using a test rig which has been developed at the URGC-Structure of INSA of Lyon (Fig. 3). The descriptions of the test set-up were detailed in [5] and [6]. In the following, only a brief description is given: Two layer of in-plane actuators are used to apply loads along the edges of the test element, allowing a wide range of loading combinations of bending moments and in-plane forces. The ratio between in-plane forces and bending moments can be chosen freely. 703

Girder Bearing Load sensing pin N1

N2 M

N

Specimen

200 100 200

2000 2480

Loading bar

N

M

N1

N2 2d Bearing Laboratory floor

Figure 4. Principle of the test set-up.

The principle of the system is similar to exterior prestressing (Fig. 4). There is no need reaction frame. However, a support frame is used to maintain the specimen in vertical position. There are twelve steel supports which are fixed with high strength bolts on the two faces of the test specimen. Hydraulic jacks and loading bars are fixed to these supports by pinned connections. The forces in the loading bars are balanced by reactions forces in the test specimen. Pinned joints ensure that the loading bars work only in tension or compression. The prescribed axial force and bending moment is obtained by including suitable forces in the pairs of loading bars: in fact, equal forces produce only membrane loads, opposite forces produce only bending moment and two different forces produce a combination of membrane and bending. 2.3 Instrumentation Measures are taken of the forces applied to the specimen, displacements normal to the plane of the slab, strains in the concrete, reinforcing steel and TFC. Forces are measured by 12 forces sensors of type dynamometric pins Strainsert SPA force transducers placed in the joints connecting the loading bars to the supports (Fig. 4): SPA 50 for the jack of capacity of 15T and SPA 100 for the jack of capacity of 40T. Transversal displacements are measured in 8 points by 8 LVDT (Linear Variable Differential Transformer). Concrete strains are measured by 15 stain gauges HBM LY41-50/120, 9 on the tensile face and 6 on the opposite face of the specimen. These gauges are placed in the central zone of the slab. Steel strains are measured by 20 stain gauges Vishay CEA 06-125 UN 120. Composite strain are measured by 9 stain gauges Vishay CEA 06-125 UN 120. There are a total of 55 channels connected to a Central data acquisition unit for the test of RC slabs and 60 channels during the test of retrofitted RC slab with two TFC composite strips. 704

Figure 5. Position of two TFC strips on tensile face of RC slab.

2.4 Process of reparation of RC slab pre-damaged with two TFC composite strips In the first time, the RC slab is pre-damaged by horizontal bending moment and verticals forces. Then this slab is repaired and retrofitted with TFC strips. The reinforcement consists of 2 TFC bands of 30 cm of width symmetrically placed from the axis of symmetry of the specimen, on the tensile face caused the bending moment. The clear span between two strips is 30 cm. The position of reinforcement is presented in the Figure 5. 2.5 Process of experiment • The RC slab is pre-damaged under vertical in-plane forces and bending moment. • Application of two TFC strips on the fissured face of the specimen. • Testing the rehabilitated slab until the ruin with vertical in-plane forces and bending moment.

3 TEST RESULTS During the test, the slab was subjected to vertical compression forces and bending moment in the horizontal direction. The value of vertical forces is 120 KN which corresponds to a stress of 0,6 MPa (cross-section of slab is 2000 cm2 ). For the test of RC slab (slab no FC composite strip), the cracks are appeared at deflection of about 1/1500 of the span. The RC slab is loaded until 16 KNm/m where the deflexion in mid-span is 10 mm, and then it’s unloaded, the residual deflexion rest 3,4 mm. Cracks were parallel and regularly spaced about 15 cm, and developed deeply in the depth of the shell on the tensile face. After apparition of the crack, the rigidity of the specimen decreases a lot, the stiffness of the cracked shell rest only 15–20% of the no cracked shell (Fig. 6). Repair of damaged shell with composite strips increased his stiffness of above 60%. The nonstrengthened presents more ductility than the strengthened one. The ultimate capacity of repaired slab is 21,5 KNm/m, a profit of about 30% compared to that of the RC slab. The ultimate mode of failure was a shear failure at the strips edge near the un-strengthened portion of the slab (Fig. 7). There are also large cracks around the supports of jacks noted. 705

25 RC slab in load Retrofitted RC slab with TFC in first load Retrofitted RC slab with TFC until the ruin

Moment (KNm/m)

20

15

10

5

0 0

2

4

6 8 Deflexion (mm)

10

12

14

Figure 6. Relations between Moment–deflexion of RC slab and repaired slab with TFC.

Figure 7. Debonding of TFC strips from the specimen.

The tension strain in the concrete was about 1200E-6. The gauges fixed on the tensile face have the strains over 4000E-6 and were broke caused due to apparition of cracks. The strains of reinforcing steel were 3950E-6 during the test of RC slab and were 4700E-6 during the test of RC slab strengthened. However, the strains of TFC rest about 2800E-6 (Fig. 8). The finite element method is used for analyzing the two way reinforced concrete slabs with and without TFC composite strips. For this purpose, in the code finite element CASTEM2000, the elements DKT (Discrete Kirchhoff Triangle) multilayers are used for simulating the concrete and the eccentric elements DKT simulate the reinforcing steel. The results of numerical simulation will be presented in the near future. 706

5500 steel-8 steel-9 TFC-40 TFC-42 TFC-43

4500

Strain (E-6)

3500

2500

1500

500

-500

1

351

701

1051

1401

1751

2101

2451

2801 3151

Time increment

Figure 8. Evolution of steel strains and composite strains during the test.

4 CONCLUSIONS The evaluation of full-scale test conducted on reinforced concrete slabs specimens strengthened with TFC composite strips appreciates the TFC system in upgrading the structural capacity of reinforced concrete slabs. For the repair application, RC slab pre-damaged, test results indicated that the composite system restored not only the original capacity of the damaged slab but also an appreciable increase of the strength of the repaired slab.

REFERENCES [1] Chafika DJELAL. et al. Renforcement des poutres en béton armé à l’aide de lamelles composites: étude expérimentale – prédimensionnement. Analyses du bâtiment et des travaux publics. [2] Ayman S. Mosallam & Khalid M. Mosalam. Strengthening of two-way concrete slabs with FRP composite laminates. Construction and Building Material., 2003, Vol 17, pp 43–54. [3] Pascu I. Jullien J.F. Contribution à l’analyse d’éléments en béton armé sollicités en membrane et flexion biaxiale. Thèse de doctorat, INSA Lyon, France, 1995. 218p. [4] FREYSSINET International & Cie. “Fiche technique Ref : FT F 0021, Révision: A” Fiche technique 21 mai 2001. 4p. [5] Pascu I., Malczik A., Edwards A.D. & Jullien J.F.: “Etude expérimentale du comportement de plaques en béton armé soumises à un chargement biaxial en membrane et en flexion. Définition des essai”. Rapport 201.135/01, 18 juin 1991, INSA Lyon [6] Pascu I. et al. Setting up biaxial membrane and bending tests on reinforced concrete panels. Proceeding of the 10th International Conference on Experimental Mechanics, Lisbon, 18–20 july 1994. Vol. 2, p. Proc. 10th Int.Conf. on Experimental Mechanics: 1061–1066. Rotterdam: Balkema.

707

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Component model for steel to concrete joints D. Gregor & F. Wald Czech Technical University in Prague, Czech Republic

ABSTRACT: The paper describes the procedure for developing of component prediction model of end plate joints under cyclic loading for the mixed building technology. Three components of the steel to concrete end plate connection were observed experimentally under repeated loading: the concrete in compression and end plate in bending, the bolt in tension and end plate in bending and end plates in bending and bolts in tension. A component model for repeated loading was developed taking into account the existing knowledge of base plate behaviour as well as published tests. The model was verified on published test results and is ready for application in practice using a complex prediction as well as for developing the simplified one taking into account the static loading only as an envelope of the behaviour. KEYWORDS: Steel structures, Mixed building technology, Steel to concrete connections, Structural joints, Experimental observations, Component method, Repeating loading.

1 INTRODUCTION Mixed building technology is commonly used in modern structures. Steel, concrete, composite steel and concrete, glass, and timber elements (beams, columns and walls) are developed. European structural standards (Eurocodes) facilitate design of elements made from different materials with required safety. Connection of structural elements produced by different technology is today’s challenge of structural engineers. The traditional approach for prediction under cyclic loading uses sophisticated curve fitting procedure based on description of the major joint parameters. Accuracy of the model is limited to the range of experimentally observed values. Extending the application outside of the boundaries for even one parameter is not possible. Therefore, other prediction methods were developed, one of them is component method based on work [1]. The application of component method requires three basic steps: identifying the components of the joint, evaluation of force-deflection diagram of the individual components, and assembling of the components in view of the evaluation of characteristics of the whole joint. The component method is based on analytical prediction of component behaviour. The model of the components differs with the required accuracy of joint prediction. The stiffness, resistance and deformation capacity of the joint are assembled separately [2]. The application of component method for cyclic loading brings good opportunity for application of both sophisticated and simple design models, see [3]. In general, two types of components may be distinguished, one-directional component (representing contact between two surfaces) and twodimensional component (resisting in tension and compression, i.e. plate in bending), see Figure 1. The knowledge developed for base plates [4] and applied in European structural steel practice may be used for connections between steel frame and reinforced concrete part of the structure. Two approaches may be distinguished: concrete in compression under rigid plate and effective area of flexible plate. Stiffness and resistance of the concrete in compression is limited by crushing of the concrete surface. Other factors influencing the behaviour are the concrete quality, the thickness and area of the plate, the grout quality and thickness, the location of the plate on the concrete structure, the size of the concrete structure its reinforcement [5]. The component plate in bending and anchor 709

Force Fc,T Vj

Mj

a)

Deformation

Fj Fc,c

d) Force

b) Deformation

Fd,bV

c)

e)

Figure 1. End plate joint of mixed building technology; a) anchor bolt in tension and plate in bending, b) concrete in compression and plate in bending, c) bolt in shear; d) one directional component behaviour (contact); e) two directional component behaviour (bending of plate).

Figure 2. a) Concrete in compression and plate in bending, position of transducers, b) threaded bar cast in the concrete block test set-up.

bolt in tension is solved by T-stub analogy. The stiffness and resistance are affected by elongation of the anchor bolts, which usually prevents development of prying forces. 2 EXPERIMENTS 2.1 Concrete in compression and end plate in bending The test set-up of experiment concrete in compression and end plate in bending is shown in Figure 2 [6]. The tests were carried in two configurations. Three specimens were attached to upper horizontal surface of the concrete block (representing concrete slab) and three specimens were tested on vertical side of the block (representing concrete wall). The surface of the concrete was cleaned and levelled by a thin (less than 1 mm thick) layer of high strength grout to achieve smooth surface. The steel plate with nominal dimensions 200 × 100 × 10 mm was placed on the fresh grout layer. The steel bar, nominally 10 × 10 × 220 mm, was centred on the plate. The concrete block was positioned under the head of the hydraulic actuator and a layer of plaster was made under the block to ensure the level of the top surface and good support on the laboratory floor. The experiments C1/3-V-10, 710

Force kN C1/2-V-0

600

C2/2-V-0

Prediction fy

Prediction fu

C3/2-V-0

500

F 400

tw = 10 mm

P 10 x200-100

300

 200

100 Displacement, mm 0 0

-0,5

-1

-1,5

-2

-2,5

-3

-3,5

-4

-4,5

Figure 3. Displacement of centre of the plate, test C1/2-V-0, C2/2-V-0 and C3/2-V-0, displacement calculated as mean value of transducers I16 and I17.

C2/3-V-10, and C3/3-V-10 were performed with 15 mm thick layer of grout with strength 10 MPa, see Figure 3. The grout oversized the steel plate by 50 mm at each side. For the set labelled C1/4-V-50, C2/4-V-50, and C3/4-V-50 were design the strength of the grout 50 MPa. Inductive transducers measured vertical displacements of the plate and concrete block, see Figure 2. The transducers I10, I11, I12, and I13 were placed at corners of the cube on the cupreous plates glued to the concrete surface. These transducers measured displacements relative to the laboratory floor. The transducers I14, I15, I18, I19 were located at the corners of the steel plate. The transducers I16, I17 were connected to the steel bar at distance 20 mm from its ends. The transducer I20 was placed on a cantilever glued on side of the steel cylindrical head of the hydraulic jack, above transducer I16, (not installed during tests C1/1 and C2/1). The load history followed the modified ECCS curve to reach comparable results, see [9]. All specimens failed by crushing of the concrete. Tension failure of the concrete block, induced by shear, was not observed. The tests were stopped after the head overturning caused by non-homogeneity of the damages at the concrete surface. The complete tests data may be found in [7]. 2.2 Threaded bar cast in the concrete block Threaded bar cast in the concrete block loaded in tension was tested separately to learn a local behaviour of the tension part of the connection, see Figure 2b. Threaded bar M20 was concreted in concrete block 500 × 500 × 500 mm. Two hydraulic jacks were placed on the plaster layer to ensure their vertical position and good transfer of their reactions into the concrete block. The beam made from two UPN 140 profiles with web stiffeners was used for transfer of forces to the threaded bar. The bar was fixed to the beam by washer plate with thickness 20 mm and hole diameter 22 mm. Three specimens were tested. The tests A1 and A2 failed by a rupture of the steel bar in tension. The test A3 was stopped before failure, the rupture of the bar developed to close to the concrete surface. The displacements at the top of the bars are shown in Figure 4. 2.3 End plate in bending and anchor bolts in tension Two threaded bars M20, 540 mm long were cast into the concrete block. The T-stub was positioned at the concrete surface four hours after the casting to ensure proper contact with the concrete surface, 711

Force, kN

Prediction

120

fu F

100 Prediction

Ø 20

fy

80

I20 +δ

60

I21

I20

30

I21

40 20

Displacement, mm

0 0

0,2

0,4

0,6

0,8

1

1,2

1,4

1,6

1,8

Figure 4. Deformation at the bar surface, transducers I20 and I21; test A1.

Figure 5. End plate in bending and anchor bolts in tension; a) test set-up; b) failure mode of end plate.

no grout was used. The nuts were tightened before the test (after 120 days from concreting) by torque of 40 Nm to simulate the hand tightening. Two hydraulic jacks were placed on the block on grout layer. The beam of two UPN 140 profiles transfers the tensile force, see Figure 5. Three specimens were tested, marked TC1, TC2 and TC3. Tests of three specimens TC1, TC2 and TC3 were carried out with this set-up. The failure of all tests was caused by rupture of the anchoring bar in tension with large plastic deformations of plate, see Figure 5b. The deformation on the top of the set up for tests TC1, TC2 and TC3 is shown in Figure 6. 3 MODELLING In compression may be the strength predicted by equation

Stiffness prediction was carried out using equation

712

Force, kN P 30 x100-170

F

120 +δ

Prediction fy

100

Prediction fu

80 60

TC1

40

TC2 TC3

20

Displacement, mm 0 0

2

4

6

8

10

12

14

16

Figure 6. Displacement of the top of the T-stub, tests TC1, TC2 and TC3.

where, kj is concentration factor, see [2], fc is concrete compressive strength based on cylinders, tw is side dimension of steel bar cross-section, c is effective width as defined in [2], L is plate width, and Ec Young modulus of concrete. The embedded depth and the concrete block size are designed in order to prevent the concrete failure. The steel failure load is predicted as

and for the initial stiffness the following equation is used

where As is tensile stress area of the anchorage threaded bar, E is modulus of elasticity, Lb is superposition of the length from the concrete surface to the measurement point and effective length equal to eight times the bar diameter. The model based on prediction by [8] is included using the measured values of material properties and taking into account the effective length of the bolt embedded in concrete as eight bolt diameters. Strength is determined as

Stiffness is calculated based on following formula

where n is distance between anchor bar axis and the end of the T-stub flange, m is distance between anchor bar axis and the T-stub web, dw is average outer diameter of the nut, Leff is T-stub effective length, in this case equal to total length of the T-stub, tp is thickness of the T-stub flange, E is 713

SP1 1

54 2

3

2

3

4x53

1

54 16

300

16

35

80

35

a)

b)

Figure 7. a) Test set-up of the joint assembly; b) component model. Moment, kNm

Moment, kNm

120

120 Test

80

Test

80

Model

Model 40

40

0

0 -20 -15 -10 -5

0

5

-20 -15 -10 -5

10 15 20

0

-40

5

10 15 20

-40

-80

-80 Rotation, mrad

Rotation, mrad

-120

-120

a)

b)

Moment, kNm

120 Test 80

Model

40 -25

-20

-15

-10

-5

0 0

5

10

15

20

25

-40 -80 Rotation, mrad c)

-120

Figure 8. Comparison of the model to test; a) first cycle, b) second cycle, c) sixth cycle.

modulus of elasticity of steel, Lb is superposition of the length from the concrete surface to the middle of nut height and effective length equal to eight times the bar diameter, As is stress tension area of the anchor bar. The deformation capacity is calculated from the measured ductility of material. The prediction in Figures 3, 4 and 7. was prepared based on the measured values of yield strength fy and ultimate strength fu of the steel and average value of concrete strength. The simple step by step procedure with constant increment of 1/1000 of deformation was applied to achieve 714

the description of the working diagram of each component as well as of the whole assembly. The unloading of each deformable component was based on the initial stiffness. 4 EXPERIMENTAL EVALUATION The prediction modeled developed based on the presented and published tests, see [4], of components, was compared to the available test of the whole assembly. The Figure 7. exhibits the test performed by Dunai et al [10], the bars transferred the tensile forces and the studs the shear as well as tensile forces. The results of the calculations show on Figure 8. the significant influence of the 1 represents shear studs by creating round 20% of the bending stiffness of the joint. The springs 2 the bars, and 3 the header studs. The model was loaded the compressed part behaviour, springs by cyclic actions based on the test records. The measured values of the material were applied in presented simulation. The first, second and sixths cycle shows a good agreement of the prediction by model to the test. 5 CONCLUSIONS The tests confirm the observations published by Steenhuis [5], which supports today’s design practice [2]. No basic influence of the grout quality was found. The shape of the unloading part of the force-deformation curve conforms to the linear assumption without softening. The comparisons to the predicted values for static loading, using measured material characteristics, show that the quality of prediction is good and the stiffness and resistance are reduced by cyclic loading. The modeling by component model for cyclic loading brings higher understanding the connection parts influence. The method exhibits good quality of prediction based on input data quality as well as chosen accuracy in detailing of model. The calculation is handicapped by the step by step procedure for the each component as well as whole joint assembly. Based on analytical nature the method enables to go to prediction of new developments. ACKNOWLEDGEMENT The work has been supported by grant COST C12 of Czech Ministry of Education, Youth and Sport. REFERENCES [1] Zoetemeijer P.: Proposal for Standardisation of Extended End Plate Connection based on Test results and Analysis, Rep. No. 6-83-23, Steven Laboratory, Delft 1983. [2] EN 1993-1-8, Eurocode 3: Design of Steel Structures, Part 1.8: Design of Joints, European Standard, CEN, Brussels 2004. [3] Rassati G.A., Noe S., Leon R.T.: PR Composite Joints Under Cyclic and Dynamic Loading Conditions: A Component Modeling Approach, in Connections in Steel Structures IV, Steel Connections in the New Millenium, Roanoke, AISC, Chicago. pp. 213–221. ISBN 1-56424-053-3. ˇ [4] Wald F., Sokol Z.: Connection design, CVUT, Praha 1999, p. 145, ISBN 80-01-0273-8. [5] Steenhuis, C.M., Bijlaard F.S.K.: Tests on column bases in compression, Published in the Commemorative Publication for Prof. Dr. F. Tschemmernegg, ed. by G. Huber, Institute for Steel, Timber and Mixed Building Technology, Innsbruck 1999, pp. 285–295. [6] Gregor D., Wald F., Eliášová M., Jírovský I.: Joints for mixed building technology with view to experiments of component steel plate in bending and concrete in compression, in Eurosteel 2002, Coimbra 2002, pp. 977–986, ISBN 972-98376-3-5. [7] Gregor D., Wald F., Sokol Z.: Experiments with End Plate Joints for Mixed Building Technology, in Experimental Investigation of Building Materials and Technologies, ed. Konvalinka P., Luxemburg F., ˇ CVUT, Praha 2003, pp. 65–82, ISBN 80-01-02835-6.

715

[8] Sokol Z., Wald F.: Experiments with T-stubs in Tension and Compression, Research Report, CTU, Prague 1997. [9] ECCS TC1 TWG 1.3 Recomended testing procedure for assessing the behaviour of structural steel elements under cyclic loading, European convention for constructional steelwork, Brussels 1986. [10] Dunai L., Ohtani Y., Fukumoto Y.: Experimental Study of Steel-to-Concrete End-Plate Connections under Combined Thrust and Bending, Technology Reports of Osaka University, Vol. 44, No. 2197, Osaka 1994.

716

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

New technique of improving the cracking resistance of concrete walls in early age S. Wolinski Rzeszow University of Technology, Rzeszow, Poland

ABSTRACT: The contribution deals with problems concerning the new technique of elimination early ages cracking in reinforced concrete walls of water treatment tanks. Slight prestressing, or rather sustained compressive stress (0.5 ÷ 1,0 N/mm2 ) applied to young concrete after a few hours of setting prevents concrete walls from cracks caused by exothermal reactions and shrinkage of hardening concrete. Different technologies of prestressing: with bonded and unbonded, permanent and temporary, internal and external tendons are discussed. The results of investigations concerning the influence of the prestressing start time on the ability to regeneration of the microstructure of young concrete and on its strength and stiffenes necessary to withstand the prestressing force are presented and discussed in this paper. KEYWORDS: early ages concrete, thermal cracking, regeneration of concrete structure, aviodance of cracking, prestressing of early age concrete.

1 INTRODUCTION Early ages thermal and shrinkage cracking is a widespread problem in reinforced concrete walls of buildings and particularly of water treatment tanks. The hardening of concrete is the result of the hydration of the cement which is exothermic reaction. The heat produced by hydration can cause a rise of temperature in the interior of a concrete wall. If the concrete in a wall cannot deform freely and strains are imposed on it, cracks may arise. Various precautions are employed in order to prevent early age cracking in medium-sized reinforced concrete walls, for example: to apply an extra “skin-reinforcement”, to use less quantity of cement or/and to replace it partially by fly ash, to use low-heat cement, to use angular aggregate with low coefficient of thermal expansion, to provide the proper curing conditions, to cool down the mix ingredients or the fresh and young concrete, to control the temperature throughout the concrete mass and between the base slab and walls by water run through the pipes embedded in the walls and base slab, to reduce restrains by including an extra movement joints within the walls, etc. Unfortunately, the efficiency of these methods of prevention of early ages cracking in reinforced concrete walls is rather limited. In the past much research has been devoted to the formulation of the reliable methods for predicting early age strength, stresses and the risk of cracking in hardening concrete. Both the practice-oriented empirical formulae (Emborg & Bernander 1994, Rostasy & Laube 1990) and computer programs for numerical simulation of temperature effects and strength development in young concrete (Garboczi 1992, Breugel 1994) are however unreliable in practice (Mangold & Springenschmid 1994). The suggestion that prestressing, or rather sustained compressive stress applied to young concrete after a few hours of setting is an efficient method of prevention of early age cracking in medium-sized concrete members like walls in water tanks, is presented and discussed in the paper. In the Department of Building Structures, Rzeszow UT we are working on the problems related to the prestressing technologies of early age concrete (Kus & Wolinski et al. 1998). 717

Figure 1. Early age cracking in the reinforced concrete wall of waste water tank.

2 CASE STUDY For many concrete structures especially for the reinforced concrete water and waste-water tanks crack control plays an important role. Cracking of concrete elements in these structures shall be limited to an extent that water-tightness will be ensured. In practice most reinforced concrete walls show unacceptable cracking just after the completion of the construction works and need difficult and expensive repairs. According to conservative estimation (Emborg 1998) about 15 × 106 m3 of young concrete built into structures in the Europe for each year is repaired in order to ensure the proper functioning or/and durability of structures. Approximate expenses of these repairs reach as far as one billion Euros per year. Several site projects of reinforced concrete waste-water tanks built recently in the Podkarpacie Region (Poland) were investigated to check the reasons of early age cracking and to work out an efficient method to prevent the young concrete from cracking. A typical example of cracking in the walls of RC tank is presented in Figure 1. The external walls of tanks were divided into segments of lengths 15 ÷ 20 m, with the vertical expansion joints. The base slab are usually made several weeks before walls. Most elements are made of the ready mixed concrete strength class B20 and B25. Walls were reinforced with the grade fyk = 350 N/mm2 ribbed vertical bars and smooth horizontal reinforcing bars of the grade fyk = 240 N/mm2 . Mesh-reinforcement is applied underneath the full external surface of walls. Walls elements were cast in the steel formworks which were stripped off after five days . Then the entire surface of concrete was sprinkled with water continuously during next seven days. Nearly the half of the external wall segments of the investigated tanks were cracked before removal of frameworks and cracks in other segments appeared during next two-three days. Typical pattern of cracks is shown in Figure 1. Vertical cracks of the width in the range from 0.05 mm to 0.45 mm had the maximum width at the level 0.8–1.5 m over the base slab. The spacing between adjacent cracks were from 1.5 m to 3 m. All cracks of the maximum width greater than 0.15 mm were through-cracks. Restrain stresses caused by thermal and strains due to the heat of hydration and other random temperature changes as well as shrinkage strains restrained by the base slab and reinforcing steel, are suggested to be the main causes of early age cracking in the walls. An analysis of temperature development and risk of cracking proves that the tolerable temperature differential between the base slab and the walls should be limited to about 10◦ C and the tolerable temperature differential between the final peak temperature of early ages concrete inside a wall and ambient temperature should be limited to about 15◦ C during first four-five days after placing, in order to avoid cracking in the walls (Kus & Wolinski et al. 1998). Unfortunately, such limitations of temperature are hardly obtainable in practice. 3 CONCEPT OF PRESTRESS OF EARLY AGE CONCRETE The prestress or rather sustained compressive stress applied to early ages concrete after a few hours of hardening may be used to produce the desired state of precompression in walls of reinforced 718

concrete water tanks which prevents early age cracking (Kus & Wolinski 1999). Prestressing of early ages concrete requires special measures due to low strength and stifness and time-dependent effects which are very difficult to control and to describe. Two methods of prestressing, a permanent one with internal, bonded tendons and a transient one with unbonded, external or internal tendons can be applied. In both cases the prestressing force is imposed to concrete by movable rigid front bulkheads of a formwork. The steel formwork should be designed to withstand the stresses resulting from the pressure of fresh plastic concrete and from the prestressing force imposed to the young concrete. Two stages of permanent prestressing by means of multi-wire strands or by hot rolled processed solid bars embedded in concrete may be distinguished. In the first stage a slender bond between the concrete and strands or bars is destroyed during prestressing of early ages concrete because the initial deformation of young concrete is about twice as large as the deformation of steel immediately after prestressing. But the early ages concrete can continue to hydrate under the sustained stresses and the bond can be rebuilt. Thus in the second stage, after a sufficient strength of concrete is attained, front bulkheads of the formwork are released and the prestressing force is imposed to concrete by bond. The magnitude of the prestressing force and the imposed stresses in concrete decreases significantly in course of time due to prestress losses, but afterwards the prestressing steel serves as an additional normal reinforcement. The concept of transient or temporary prestressing may be executed by posttensioning with sheated strands without bond or with external tendons. Action of the prestressing force is limited in this case and can be kept until the formwork is finally removed. Corrections of the imposed compressive stress in the early age and young concrete can be introduced in accordance with volume changes of hardening concrete. Any additional prestressing steel is used in this method, however after the formwork is removed the strands inserted in the inner ducts may be used for traditional prestressing by post-tensioning.

4 OBJECTIVES, SCOPE AND RESULTS OF INVESTIGATIONS To judge whether the suggested method of prestressing the early ages concrete will be effective and to verify assumptions and procedures necessary for preliminary calculations and execution of prestress in practice, an extensive research program was carried out in the Department of Building Structures, Rzeszow UT (Kus & Wolinski et al. 1998). The research were concentrated on the following main tasks: – To determine the temperature, strain and strength development in the interior of concrete hardening under conditions typical of medium-sized reinforced concrete walls. – To assess whether cracks will arise as a consequence of hydration heat development, taking into consideration external and internal restrains. – To determine the influence of prestress on fresh and young concrete longitudinal and lateral deformations and stresses development. – To find the optimum prestressing start time in the context of the early ages concrete strength and ability of its microstructure to regeneration. – To assess the average values of prestressing force during the hardening of young concrete taking into consideration different losses of prestress. – To settle the technical possibility and conditions of efficient prestressing the early age concrete during hardening in walls of waste water tanks. 4.1 Properties of concrete at early ages Approximately two to four hours after the addition of water to the mix, the fresh concrete starts to set. A period of setting may extend over several hours when gradual transition to the state of hardened concrete takes place. Due to an exothermal process of the hydration of cement the temperature of 719

the young concrete and its volume increases which if restrained leads to low compressive stresses in concrete. As soon as less heat is developed than is dissipated to the surrounding atmosphere or to adjacent parts or elements of a structure (e.g. to the foundation slab), the temperature of the young concrete and its volume decreases again. The volume reduction is of much more significance, because, if the concrete element is restrained, then internal and restraining stresses occur which result in severe cracks across the entire section. Also the ultimate tensile strain capacity of the young concrete decreases during the setting period and for elements of a medium massivity passes through a minimum at an age between 5 and 10 hours. Analysis of the development of restraining stresses requires a number of input data which are difficult to determine experimentally for the concrete at early ages. Review of technical literature data and verifying test results (Kus & Wolinski et al. 1998) permitted the author to formulate simplified semi-empirical formulas to define the temperature T (t) and the mean values of mechanical properties of the young concrete:

where Tp = initial temperature of concrete mix; M = massivity of a member (for walls M = d/2); t = time of hardening in hours; tc = 2T −20/10 t = corrected time of hardening in hours; T = ambient temperature; k1 and k2 = empirical coefficients,

where fctm (tc ), fcm (tc ), Ecm (tc ) = mean values of the tensile strength, the compressive strength and the modulus of elasticity of concrete after tc hours of hardening; fctm , fcm , Ecm = mean values of these properties of concrete after 28 days (672 hours) of hardening; ω(tc ) = degree of hydration after tc hours of hardening; ω0 = threshold of hydration (about 0.2). During the hardening of concrete when the temperature course is well defined, the stresses development σct (tc ) in a wall which longitudinal deformation are restrained can be calculated as follows:

where α = coefficient of thermal expansion; r = restraining factor, 0 ≤ r ≤ 1; Ti = temperature change in corrected time of hardening tc = tc,i − tc,i−1 (“+” for rise and “−” for fall in temperature); cr = relaxation factor, 0 ≤ cr ≤ 1. To prevent cracks, the inequation σct (tc ) ≤ fctm (tc ) must be fulfilled. However, it is rather difficult to judge theoretically whether cracks will arise as a consequences of hydration heat development. 4.2 Prestressing start time The optimum age for prestressing of young concrete in order to prevent cracking at early ages depends on two conflicting factors. The first factor is the ability to regeneration of the microstructure of hardening concrete and the second factor is the strength and the modulus of elasticity of young concrete necessary to withstand the prestressing force. Theoretical analysis of the optimum 720

1.0 (a)

fct,fl / f'ct,fl

0.8 0.6 0.4 0.2 0.0 0

2

4

6

8

10 12 14 16 prestressing start time [h]

18

20

22

24

10 (b)

strain x 103

8 6 4 2 0 0

2

4

6

8

10

12

14

16

18

20

22

24

prestressing start time [h]

Figure 2. Results of tests on prestressing start time. The mean value of relative flexural strength (a) and longitudinal strain (b) of early ages concrete.

prestressing start time is complex and rather unreliable due to uncertainities associated with the prediction of the early ages properties of concrete and stresses caused by thermal strains. Therefore the laboratory investigations were undertaken in order to study these problems (Kus & Wolinski et al. 1998). Tests were performed using specimens of dimensions 150 × 150 × 500 mm with artificial cracks in the half-length, prestressed from 3 to 16 hours after compacting the concrete mix in steel moulds. The ability to regeneration of the microstructure of early ages concrete was evaluated by means of the ratio of the flexural tensile strength measured on prestressed samples with initial crack to the flexural strength and longitudinal strain measured on prestressed samples without  after 28 days of hardening (Fig. 2). The test results show that the optimum initial crack fct,fl /fct,fl prestressing start time of concrete strength classes B20 and B25 is about 8 hours after placing and compacting of the concrete mix and that cracked specimen after about 16 hours of concrete hardening didn’t have the ability to regeneration of concrete microstructure. 4.3 Prestressing force and deformations of early ages concrete For the permanent prestressing with internal bonded strands Ø12.5 mm, the initial bond between strends and young concrete is destroyed and after prestress it has to be restored. Such a process may be considered as the transformation of post-tensioned prestress concrete into pre-tensioned prestressed concrete. The average value of the imposed compressive stress in young concrete necessary to prevent early ages cracking in reinforced concrete walls of a medium massivity, calculated theoretically, varies from 0.5 to 1.0 N/mm2 depending on the conditions. In order to assess the mean value of prestressing force during the hardening of early ages concrete experiments on two types of wall segments were carried out. Dimensions of the first wall segments without a base slab were 500 × 1200 × 3000 mm. The second wall segment with the dimension 500 × 1500 × 9000 mm 721

was fixed in reinforced concrete base slab of the dimensions 500 × 1200 × 9000 mm, executed two months ealier. Each segment was prestressed using 6 strends Ø12.5 mm with the characteristic breaking force of 155 kN, spacing uniformly in three levels of the cross sections. Wall segments were cast using the standard formworks with additionally stiffened latteral sides and steel movable rigid front plates where prestressing tendons were anchoraged. The prestressing of wall segments were started eight hours after setting the fresh concrete in formworks, by successive stretching pairs of tendons from one side of the wall. During the prestressing and next twenty eight days, the tensile forces in tendons, forces in ties of formwork, the longitudinal deformations of young concrete, lateral displacements of the formwork sides were measured. Twenty eight days after prestressing the tendons were released from the anchorages and drive-in of tendons was measured. Then the bond between prestressing tendons and concrete was checked by means of in situ pull-out tests. In accordance with experimental results the average tensile force in tendons and consequentially the average imposed compressive stress in concrete appears to stabilize within 1.5 to 2 hours after prestressing start time. The total loss of the prestressing force in tendons was about 40% in the 3 m long walls and about 50% in 9 m long wall. Due to the friction between tendons and concrete the difference (about 10 ÷ 20%) between tensile forces in tendons at the tensioning side and the anochorage side were observed. The longitudinal deformations of early ages concrete compressed between rigid front plates which move inside formworks were measured by means of front plates displacements Figure 3a and Figure 3b.

25 displacement [mm]

(a) 20 15 10 5 0 0

8

16

24 time [h]

32

40

48

Figure 3a. Average displacements of front plates in time for 3 m long wall segment.

5 displacement [mm]

(b) 4 3 2 1 0 0

8

16

24 time [h]

32

40

Figure 3b. Average displacements of front plates in time for 9 m long wall segment.

722

48

Lateral pressure of early ages concrete on the longitudinal vertical sides of formworks were measured by means of tensile forces in steel ties stiffening the sides of formworks and their lateral displacements. The maximum values of lateral pressure were rather small and didn’t exceed 110 kPa for 3 m long walls and only 10 kPa for 9 m long wall, and this means that even for short segments of walls standard formworks can be used in casting concrete walls provided for prestressing in early ages. Twenty eight days after prestressing the strands were released from the anchorages and their drive-in at the anchorage was measured during this operation. The drive-in of strands measured at the active (tensioning) side were 2.5÷4.9 mm for short walls and 1.0÷2.5 mm for long wall, and at the anchorage side: 1.5÷3.2 mm and 1.0÷2.0 mm respectively. In situ pull-out tests were applied to check the bond between prestressing strands and concrete after more than twenty eight days of prestressing the young concrete. In all tests failure was reached by tension of prestressing tendons. It means that the initial bond destroyed during prestressing of early ages concrete had been fully restored.

5 CONCLUSIONS Based on the analysis and investigations presented in the paper the following more important conclusions can be drawn: – The concept of prestressing of early ages concrete can be recommended as the effective and practically feasible method for prevention of the thermal and shrinkage cracking in reinforced concrete walls of water and waste-water tanks. – The optimum age for prestressing the early ages concrete of the low and medium strength classes is about six to eight hours after casting and compacting the concrete mix in the formwork. Required corrections of the prestressing force in the internal bonded tendons should be finished before the next two to four hours elapsed. – To control and keep the designed value of prestressing force during the initial ten to twelve hours after prestress, posttensioning with the sheated strands or with the external tendons may be recommended, especially in case of short wall segments. – Tensile forces in tendons prestressed six to eight hours of concrete hardening appears to stabilise within two to three hours. The total loss of tensile force in prestressing tendons depends on the prestressing start time and the length of wall segment and can be assessed as forty to fifty percent of its initial value. – The longitudinal deformation of early ages concrete prestressed after six to eight hours of hardening increases rapidly during a few dozen of minutes after prestressing and during next twenty eight days is limited up to about five to ten percent. – Small drive-in values of tendons released from the anchorages and results of pull-out tests shows that initial bond between tendons and concrete destroyed during prestressing procedure was restored afterwards. – The maximum values of lateral pressure of early ages concrete on the formwork sides didn’t exceed ten to fifteen percent of the longitudinal pressure. It means that even for tubular formworks and short wall segments standard formworks can be used for casting concrete walls provided for prestressing in early ages. REFERENCES Breugel, van K. 1994. Numerical solution of the development of concrete properties and risk of cracking in early age concrete. In Proc. First Slovak Conference on Concrete Structures, Bratislava, Slovakia, September 1994. Bratislava TU. Emborg, M. & Bernander, S. 1994. Thermal stresses computed by the method for manual calculations. In R. Springenschmid (ed.), Avoidance of thermal cracking in concrete; Proc. Intern. RILEM Symp., Munich, October 1994. Munich TU.

723

Emborg, M. 1998. Avoidance of thermal cracking in order to achieve durable concrete structures – New experience. Proc. of the 12th European Ready Mixed Concrete Congress (vol. 2), Lisbon, June 1998. Garboczi, E.J. 1992. Computer-based models of the microstructure and properties of cement-based material. In Proc. 9th Inern. Conference on Chemistry of Concrete, New Delhi 1992. New Delhi. Kus, S. & Wolinski, S. et al. 1998. Prestressing of early age concrete as the method of avoidance of thermal and shrinkage cracking in walls of reinforced concrete tanks (in Polish). Report, Rzeszow TU, grant No. 7S10302107 from Polish National Research Committee, Rzeszow, 1998. Kus, S. & Wolinski, S. 1999. Prestressing of concrete at early age as the method of elimination of thermal and shrinkage cracks (in Polish). In˙z ynieria i Budownictwo No. 6/1999: 327–330. Mangold, M. & Springenschmid, R. 1994. Why are temperature – related criteria so unreliable for predicting thermal cracking at early ages? In R. Springenschmid (ed.), Avoidance of thermal cracking in concrete; Proc. Intern. RILEM Symp., Munich, October 1994. Munich TU. Rostasy, F.S. & Laube, M. 1990. Experimental and analytical planning tools to minimize thermal cracking of young concrete. In Testing During Concrete Construction; Proc.RILEM Workshop, New York 1990.

724

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Investigations of the lap-joints with blind bolts of cold-formed sections W. Wuwer Silesian University of Technology, Gliwice, Poland

ABSTRACT: In order to recognize the behavior of joints joined by means of special blind bolts, in thin-walled structures, the author has tried to solve this problem analytically in connection with an experimental verification of exemplary calculations. Basing on his own investigations he has set up a regression curve for one single blind bolt, describing the relation between load and displacement. The regression function was introduced into a set of equations, which permits to calculate any arbitrary five-blind-bolt joint, simultaneously loaded with the bending moment and the shearing force. The results of numerical calculations were checked experimentally. By means of the suggested set of equations the interactive influence of the shearing force on the load-bearing capacity and rigidity of the investigated joint has been illustrated. Besides that, boundary curve of the load-bearing capacity of the joint has been plotted. KEYWORDS: Flexible lap-joint, blind bolt, instantaneous center of rotation, boundary curve, initial and instantaneous rigidity. 1 INTRODUCTION In the case of structures consisting of cold-formed sections of much importance are joints warranting a fast and efficient assembly of the elements on site. Such requirements are met by a special kind of blind bolts, type BOM R16, produced in the USA. They permit to join walls with open or half-closed cross-sections, including economical closed cross-sections. Overlapping blind-bolt joints for sheet iron, 4.0 and 5.0 mm thick, have been investigated experimentally. The joints were axially and eccentrically stretched (Wuwer 2004). The results of tests have made it possible to determine the load-bearing capacity to pressure of the applied blind bolts, in accord with Recommendations 1990. Functions of the regression curves have been derived, describing the relation between the load of a single blind bolt and the mutual displacement of the joined steel-sheets. Equations have been quoted describing the work of any arbitrary lap joint loaded with the bending moment M and the shearing force W as the resultant of the component forces H and V . The adequacy of this set of equations was experimentally verified. Problems concerning the work of eccentrically loaded lap joints have been dealt with already by many authors, e.g. Crawford & Kulak 1971, Foti & Dunai 2002, Salmon & Johnson 1990. A set of equations describing the behaviour of a joint with five blind bolts, stretched eccentrically by the arbitrarily oriented shearing force W has been solved numerically. The boundary curve, of the load-bearing capacity, was plotted in a rectangular system of dimensionless co-ordinates M /Mlim and W /Wlim . The rigidity of a five-blind-bolt joint loaded with the forces M and W was compared with the rigidity of such a joint loaded only with the momentum M . 2 LOAD-BEARING CAPACITY OF THE BLIND BOLT The relation existing between the load of a single bolt S1 and the entire displacement δL+E , i.e. the elastic δE and constant “slack” δL , occurring between the joined walls, 5.0 mm thick, is marked in 725

Figure 1. Relation S1 –δL+E when walls 5.0 mm thick are joined in five sample elements “I”.

Figure 1 by broken lines. They are envelopes of the load paths obtained in the course of experimental investigations on five sample elements “I” (Wuwer 2004). The joints were cyclically loaded and relieved to zero, which was repeated at least four times on at least four levels, according to (Recommendations 1990). The joints in the elements “I” consisted of two blind bolts, type BOM R16-6, with a diameter of d = 13.6 mm. The blind bolt is designed as a mandrel ø10.25 mm with a bushing whose wall is about 1.7 mm thick. The blind bolts were mounted in holes ø14.0 mm. With the growing load, the blind bolts were gradually tilted, and than they began to operate in a complex state of stresses, i.e. they were sheared, stretched and bent. Their destruction consisted in the shearing of the bushes and the breaking of the mandrel, the forces amounting to 76.0÷77.0 kN, together with mutual displacements of the walls by up to 10 mm. According to the guidelines (Recommendations 1990), however, the boundary destructive forces acting on one single blind bolt jointly with S1 = Slim = Pm have been determined as associating the mutual displacements of the sheets in the axis of the bolt, which amounts to δlim = 3.0 mm. The values of the forces Pm in five investigated joints amounted to 56.8 kN, 56.7 kN, 56.0 kN, 54.0 kN and 55.8 kN, respectively. Thus, the average boundary value could be calculated, amounting to Pm,med = 55.85 kN. The loads Pm were taken as the basis for the statistical determination of the load-bearing capacity to pressure of the blind bolt, which was Pd = 48.6 kN. The calculated load-bearing capacity of the blind bolt to shearing SRv = 89,5 kN, quoted in manufacturer’s catalogue is by 84% higher than the determined value of the load-bearing capacity to pressure SRb = Pd . The tested blind bolt satisfies, therefore, the standard requirements of similar bolt connectors, so that its load-bearing capacity to shearing would be at least by 20% greater than the load-bearing capacity to pressure. The regression curve in Figure 1 is described statistically by the exponential function:

Figure 1 quotes the values s = of the standard deviation and rK = of the correlation index. The load-bearing capacity Pd corresponds to the global displacement of the joined sheets amounting to δL+E,d = 209.3 · 10−2 mm. In order to find out how the investigated joints behave when subjected to alternate loads, two experimental elements were tested, viz. I-100,5.1/2;n and I-100,5.2/2;n (Fig. 2). 726

Figure 2. Diagrams S1 –δ plotted for alternately loaded joints in the experimental elements “In”.

The total mutual displacements of the joined walls amounted to as much as 346.8 · 10−2 mm and 356.9 · 10−2 mm, the forces exerted on the blind bolts amounting to S1 = Pd = ±48.6 kN, which means that they exceeded the boundary value of δlim = 300 · 10−2 mm. Displacements in the positive or negative semi-cycles, however, were smaller then the admissible values δL+E,d = 209.3 · 10−2 mm (cf. Fig. 1). 3 THE EFFECT OF A SINGLE-CUT JOINT ON THE BLIND BOLTS The object of consideration is an arbitrary lap joint of two sheets of the thicknesses tg i td . The loads H , V and M , exerted by the “upper” sheet on the “lower” one will cause their mutual displacements (Fig. 3). These sheets will undergo two component shifts with respect to each other, i.e. horizontally u = ug + ud and vertically v = vg + vd , as well as a rotation, i.e. an angular displacement φ = φg + φd . The quantities ug , vg , φg – are displacements of the “upper” shield, whereas ud , vd , φd – are displacements of the “lower” shield, with respect to the assumed center point O of the coordinate system (x, y). The pressure of the mandrels of the blind bolts to the walls of the bored holes was replaced by the resultant forces Si , situated in the designed centres of the boreholes. The direction of every resulting force Si is determined by the angle βi between it and the axis x. The set of equations describing the behaviour of any arbitrary joint, taking into consideration the unknown forces Si and displacements ug , vg and φg , may generally be expressed by 3 equations of equilibrium and m physical equations, for each link “i” one equation, assuming i = 1, 2, . . . m, where m denotes the number of links in the joint. In the set of equations (3 + m) the conditions of equilibrium may be presented as follows (Wuwer 2004):

727

Figure 3. Forces and displacements in the axis of the blind bolts “i” in a single-cut lap joint transferring the loads H , V and M .

whereas the physical equations, connecting the load of the links Si with the global displacement δi = δi,g + δi,d , take the form:

where: f (Si ) – denotes the reverse function to the relation (1). The bending moments in equation (4) were calculated with respect to the centre O of gravity of the joint. In the publications Crawford & Kulak 1971, Foti & Dunai 2002, Salmon & Johnson 1990 the equations of equilibrium were calculated with regard to the unknown instantaneous centre of rotation, introducing an additional unknown value as the difference between the center of gravity of the fasteners and the instantaneous center of rotation. The solution suggested below permits to calculate the values of the co-ordinates of the centre of rotation R, in each arbitrary joint. The occurrence of the instantaneous centre of rotation in the joint was confirmed by the results of experimental investigations presented in section 4. 4 INVESTIGATIONS CONCERNING JOINTS IN AN EXPERIMENTAL ELEMENT “V” In two identical experimental elements “V” a lap joint with five bolts of the type BOM R16-6 was cyclically loaded and relieved at various levels of loading F (Fig. 4a). To this joint were transmitted the shearing force V = F and the bending moment M = F · e = V · e. The results of measurements of the displacements have made it possible to determine for each level of the load F the values of the coordinate xR (cf. Fig. 3) of the instantaneously “wandering” center of rotation. The behaviour of a five-blind-bolt joint, subjected to the same sequence of the paths of static equilibrium, i.e. to 26 cycles of loading and relieving (Wuwer 2004), is illustrated by the paths 1 and 2 in Figure 5. The characteristic irregularity occurring in the paths of static equilibrium S1 − δL+E and M − φ (cf. Figs 1, 5) results both from unavoidable errors in the preparation of the elements to be joined and the imperfect technique of inserting the blind bolts in the holes, and also from the fatigue of the material of the walls, due to cyclic loads of the blind bolts. 728

Figure 4. Five-blind-bolt joint: a) experimental element “V”, b) scheme of a joint loaded with the bending moment M and the shearing force W inclined at the angle αW to the axis x.

Figure 5. Relation M − φ: broken lines 1 and 2 for joints in the experimental elements “V”, curve 3 – complying with the regression curve (1), concerning only the bent joint, curve 4 – acc. to numerical calculation.

The equations (2)÷(5) concerning a five-blind-bolt joint, respectively supplemented by geometrical relations, were solved numerically by means of the programme “Mathematica”. The obtained results of calculations permitted to plot the curve 4 in the diagram M − φ (cf. Fig. 5). Curve 3 illustrates the behaviour of a five-blind-bolt joint, analogical to the joint tested in the experimental element “V”, but loaded only by the moment M . Curve 3, based on the exponential function M −φ, was plotted making use of the regressive curve (1). The reduction of the rigidity of the simultaneously sheared and bent joint in the element “V” in relation to an analogical bent joint is 729

characterized by the coefficient of the reduction of rigidity υV :

The coefficient υV realised within the range of 10,0 kN ≤ V = F ≤ 100 kN varied accordingly from 0,995 to 0,663. 5 BOUNDARY CURVES FOR A FIVE-BLIND-BOLT JOINT The set of equations (2)÷(5) may be used to plot boundary curves for any kind of joint. For example, boundary curves have been plotted for a five-blind-bolt joint, loaded with the bending moment M and the shearing force W , inclined at any arbitrary angle αW to the axis x (Fig. 4b). The boundary values of the load components were calculated by means of the following relations:

corresponding respectively to the three considered boundary states in the joint. In the boundary state of the load-bearing capacity – I it has been assumed that at least one link in the joint reaches the force Pd = 48.6 kN. In the boundary state of displacements – II the mutual displacements of the joined walls reach, at least in one link the boundary value δlim = 3,0 mm, which complied with II loading the blind bolt by the force Si ≤ Pm,med = Slim = 55.85 kN. However, in the boundary state of destruction – III the load of all the links in the joint approach theoretically the boundary value equal to aS = Pdestr = 58,58 kN, resulting from the constitutive model assumed for the link described by the function (1). In the case of each one of the three boundary states the set of equations (2)÷(4) was determined after the substitution of the respective boundary values Wlim and Mlim , calculated in compliance with (7); the values of H , V and M must be expressed by the quantity Slim , the quotients W /Wlim and M /Mlim and of the angle αW . Basing on the results of numerical calculations, the boundary curve I in Figure 6 was plotted for αW = 0 or π /2 in dimensionless orthogonal coordinate systems (M /Mlim , W /Wlim ).

Figure 6. Boundary curve I of a joint loaded with the forces M and W when αW = 0 or π /2 and the path D I . of the load realized in the elements “V”, as well as the contours of the coefficients of reduction υW

730

Figure 7. Initial and instantaneous rigidities of a five-blind-bolt joint: a) loaded with the moment M (curve 3), b) with the moment M and force W when αW = π /2 (curve 4).

The equations derived in section 3 were used to generalize the reducing influence of any arbitrarily oriented shearing force W on the rigidity at bending a five-blind-bolt joint. For this purpose the value of rigidity at bending, reduced due to the force W , was compared with the value of the rigidity of a joint loaded merely with the bending moment M . It has been found that in order to take into account all the possible orientations of the operating force W – inclined at the angle αW towards the axis x of the coordinate system within the range 0◦ ≤ αW ≤ 2π – we need only assess its effect in the range of variation 0◦ ≤ αW ≤ π/4. If the force W changes its orientation within this range, the highest effort will be exerted on the blind bolt No. 4 (cf. Fig. 4a), reaching at first subsequent values of Pd , Pm,med and approaching to Pdestr . The value of the coefficient of the reduction of rigidity must then be calculated in compliance with (6), introducing the denotations υW and φM +W instead of υV and φM +V . The influence of the shearing force W on the rigidity at bending can be illustrated by means of a contour schedule prepared for changing values of υW in Figure 6. As we have to do with a single-stage interpolation, the contour scheme seems to be reliable and convenient when reading off quickly the coefficient υW . 6 RIGIDITY OF FIVE-BLIND-BOLT JOINT The diagram M − φ in Figure 7 provides a comparison of the rigidity of an eccentrically loaded five-blind-bolt joint, type “V” (curve 4) with the rigidity of a joint loaded only with the moment M (curve 3). As far as the curve 3 is concerned, the relation between the initial rigidity Kri and instantaneous rigidity Krt of the joint may be expressed as follows:

where ωM denotes, as in (Rabotnow 1968), the parameter of the degradation of rigidity, expressing the global plastic destruction in the joint loaded with the moment M . Basing on formula (6), the instantaneous rigidity in the curve 4 may finally be expressed, as follows:

731

7 CONCLUSIONS The presented set of equations which permits to calculate susceptible joints loaded simultaneously with the bending moment M and an arbitrarily oriented shearing force W , has made it possible to determine numerically the values of forces exerted on the individual blind bolts. In the equations it has been assumed that both loads cause a plastic deformation of the joints, determined experimentally by the regression function. In the joint instantaneously occurs the centre of rotation, versus which the forces are distributed to the individual links. According to the presented procedure three boundary curves, I, II and III, can be plotted in the system of adequate dimensionless orthogonal coordinates M /Mlim and W /Wlim in the case of any joint loaded with the bending moment M and arbitrarily oriented shearing force W . The boundary curves permit to check quickly the load-bearing capacity of the given joint without having to carry out arduous numerical calculations. Besides that, formulae may be derived which allow to calculate the rigidity of selected joints, basing on the results of experimental investigations and the contour plan of the coefficients υW obtained by solving the set of equations presented in section 3 of this paper in a somewhat simplified way. REFERENCES Crawford, S. F. & Kulak, G. L. 1971. Eccentrically Loaded Bolted Connections, Journal of Structural Division, ASCE, 1971, 97, ST3, pp. 765–783. Foti, P. & Dunai, L. 2002. Test Based Design Method of Moment Resisting Joints in Cold-Formed Structures, Stability and Ductility of Steel Structures (SDSS 2002), Memorial Session, edited by M. Ivanyi, Akademiai Kiado, Budapest, pp. 211–218. Rabotnow, J. N. 1968. Creep rupture, Proc. 12 Int. Congr. Appl. Mech., Stanford, pp. 342–349. Recommendations for Steel Constructions, ECCS-TC7, The Design and Testing of Connections in Steel Sheeting and Sections. Constrado, 21/1990. Salmon, C. G. & Johnson, J. E. 1990. Steel Structures: Design and Behavior, Emphasizing Load and Resistance Factor Design, Harper Collins Publishers. Wuwer, W. 2004. The behaviour of eccentrically loaded lap-joints of thin-walled sections (in Polish), Scientific conference on “Problems concerning the boundary states of steel structures”, Cracow, Conference proceedings, pp. 317–326.

732

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Evaluation of behaviour of hybrid composite cable in the saddle-shaped roof D. Serdjuks & K. Rocens Institute of Structural Engineering and Reconstruction, Riga Technical University, Latvia

ABSTRACT: Hybrid composite cable with increased, in comparison with CFCC, breaking elongation and decreased, in comparison with the steel cables, dead weight could be elaborated for prestressed cable net on the base of carbon and steel. Hybrid composite cable contains three layers: carbon fiber reinforced plastic (CFRP) core, glass fiber reinforced plastic (GFRP) distributional layer and steel wire strands. All layers of hybrid composite cable take up tension stresses, acting in the cable during exploitation. But GFRP has second function – distribution of transversal pressure of steel wire strands at CFRP core. The dependence of external pressure per unit of the surface area of the GFRP (due to the pressure of steel wire strands) of hybrid composite cable on the axial force and angle of steel wire strands twisting was estimated by the engineering method of calculations. Tangential and radial stresses for GFRP and CFRP components of hybrid composite cable were obtained. It was shown, that the angle of steel wire strands twisting should not exceed 20 degrees due to the ultimate strengths of GFRP and CFRP components in radial and tangential directions. Opportunity to decrease the displacements of composite saddle-shaped cable roof by the using of cable trusses made of hybrid composite cable, as a supporting contour structure, was investigated.

1 INTRODUCTION Hybrid composite cable with increased, in comparison with carbon fiber composite cable CFCC, breaking elongation and decreased, in comparison with the steel cables, dead weight could be elaborated for prestressed cable roofs on the base of carbon and steel. Hybrid composite cable contains three layers: carbon fiber reinforced plastic (CFRP) core, glass fiber reinforced plastic (GFRP) distributional layer and steel wire strands. All layers of hybrid composite cable take up tension stresses, acting in the cable during exploitation. But GFRP has second function – distribution of transversal pressure of steel wire strands at CFRP core. This pressure has significant value and should be taken into account during the cable design. Perpendicular to the direction of axial force action pressure causes radial and tangential stresses in GFRP and CFRP. Volume fractions of steel and carbon were evaluated basing on the assumption, that in an emergency, when the strain of carbon fiber exceeds the ultimate value and these fibers are disrupted, strands of steel wire must be able to take up significantly decreased tension stresses. The decrease of the tension forces, acting in the cable, is joined with the growing of deflection after excluding from the work of GFRP and CFRP components (Figure 1, stages 2 and 3). Saddle-shaped cable roof is an example of the structures, where hybrid composite cable can be used. The saddle shaped cable roofs with the compliant supporting contour are structures where nearly all load-bearing elements are tensioned. It means that the modern high strength structural materials could be used in the full scale for the saddle shaped cable roofs supported by the tensioned cables. At the same time, the most significant disadvantage of the saddle-shaped cable roofs supported by tensioned cables is the increased compliance. Basing on the just obtained results and the above mentioned information we can suppose, that the best method to decrease the displacements of the saddle shaped cable roof is to use the cable 733

Figure 1. Scheme of hybrid composite cable work: 1 – steel wire, GFRP and CFRP work commonly; 2 – GFRP is excluded from the work; 3 – GFRP with CFRP are excluded from the work and steel wire works alone; q – design vertical load, acting at the cable; f1 – deflection of the cable, which corresponds to the stage, when steel wire, GFRP and CFRP work commonly; f2 – deflection, which corresponds to the stage, when GFRP is excluded from the work and steel wire works commonly with CFRP; f3 – deflection, which is corresponds to the stage, when GFRP and CFRP are excluded from the work and steel wire works alone; l – span of the cable.

trusses made of the materials with the increased moduli of elasticity as structures of the supporting contour. So, the purpose of this study is to evaluate perpendicular to the direction of axial force action pressure of steel wire strands at GFRP distributional layer and that of GFRP distributional layer at the CFRP core in the hybrid composite cable. Radial and tangential stresses in the GFRP and CFRP also should be determined and compared with compression strengths of the GFRP and CFRP. Effectiveness of cable truss application in supporting contour structure as a method to decrease the displacements of the composite saddle-shaped cable roof also should be evaluated. Rational geometrical characteristics of the cable truss shell be estimated.

2 EVALUATION OF MECHANICAL INTERACTION BETWEEN COMPONENTS IN HYBRID COMPOSITE CABLE 2.1 Approach to the solution of the problem Hybrid composite cable is considered as a system of two cylinders (see Fig.2). Steel wire strands are replaced by the external pressure pb per unit of external surface area of the GFRP distributional layer. The GFRP distributional layer is considered as a hollow cylinder inside which another cylinder, i.e., CFRP core is situated. The GFRP distributional layer has constant internal and external radiuses: a and b, respectively. The CFRP core has constant external radius, which is equal to a. Interaction between the GFRP distributional layer and CFRP core is considered as a pressure pa at the unit of the surface area of the CFRP core or at the unit of internal surface area of the GFRP distributional layer. Pressure pb at the unit of external surface area of the CFRP distributional layer could be determined by the following equation:

where n = part of axial force N , which takes up steel wire strands of the cable; α = angle of steel wire strands twisting; a = radius of the CFRP core; R = radius of the cable. The equation (1) was obtained for the case, when GFRP distributional layer limits the displacements of the steel wire strands in the radial direction. Pressure pa per unit of the surface area of the CFRP core and per unit of internal surface area of the GFRP distributional layer could be 734

Figure 2. Scheme for determination of pressure at the CFRP core of hybrid composite cable: 1 – GFRP distributional layer; 2 – CFRPcore; σr – radial stresses; σθ – tangential stresses; pb – external pressure per unit of the surface area of the GFRP (due to the pressure of steel wire strands); pa – external pressure per unit of the surface area of the CFRP (due to the pressure of GFRP); a – radius of the CFRP core of the cable and internal radius of GFRP distributional layer; b – external radius of the GFRP distributional layer.

determined by the equation (2). The equation (2) is obtained due to the equal radial deformations of CFRP core and GFRP distributional layer:

where EGr , ECr = modulus of elasticity for GFRP and CFRP, respectively, in the radial directions; r = coordinate of the point, where deformations are determined; νGrz = Poisson’s ratio of GFRP; νCrz = Poisson’s ratio of CFRP; a = radius of the CFRP core of the cable and internal radius of GFRP distributional layer; b = external radius of the GFRP distributional layer; δGr = radial deformations of GFRP due to the part of axial force, acting in the GFRP component of the cable; δCr = radial deformations of CFRP due to the part of axial force, acting in the CFRP component of the cable. The left and right parts of the equation (2) are radial deformations of GFRP and CFRP components, respectively, due to the pressures pb and pa . Radial deformations of GFRP and CFRP components δGr and δCr due to the parts of axial force, acting in the components also are taken into account. Values of radial deformations of GFRP and CFRP components δGr and δCr were determined basing on the consumption that components work in the elastic stage. Radial and tangential stresses act in the GFRP and CFRP due to the pressure pb . The values of radial and tangential stresses could be determined by the equations, which were obtained for the cylinder with the hole in the center, which is loaded by uniformly distributed by the internal and external surfaces pressures pa and pb , respectively.

where σGr and σGθ = stresses acting in the GFRP component of hybrid composite cable in the radial and tangential directions. For determination of radial and tangential stresses acting in the CFRP component of the hybrid composite cable the following equation could be used:

735

where: σCr and σCθ = stresses acting in the GFRP component of hybrid composite cable in the radial and tangential directions. Equation (5) was obtained from the equations (3) and (4) when the internal radius of the cylinder (CFRP core) is equal to zero, external radius of the cylinder is equal to a, and external pressure per unit of the surface area of the CFRP (due to the pressure of GFRP) is equal to pa .

2.2 Determination of pressures on the components of hybrid composite cables The dependence of pressure at the CFRP core of hybrid composite cable on the axial force N and angle of wire twisting α was developed by the example of tension cable of saddle shape cable roof with dimensions in plan 30 × 30 m. Cable, which is loaded by the uniformly distributed load, is considered as a scheme for analysis. The cable has rational from the point of view of materials consumption initial deflection f1 = 5.7 m. The uniformly distributed load with intensity q = 21 kN/m loads the cable. Mechanical properties of hybrid composite cable components are given in Table 1. The values of moduli of elasticity correspond to the elastic stages of the materials work. Volume fractions of fibers in GFRP and CFRP are 0.6. The fibers are oriented in the direction of axial force action. Volume fractions of steel wire, GFRP and CFRP are 0.4; 0.2 and 0.4, respectively. Total area of cross sections for hybrid composite cable was equal to 0.00097 m2 . The value of the axial force N , acting in the cable due to the uniformly distributed load, could be determined by the equations of cable calculation without taking into account elastic elongation of the cable. The dependence of pressure pa on the axial force N and angle of steel wire strands twisting α is shown in Figure 3.

Table 1. Mechanical properties of hybrid composite cable components. Components of hybrid composite cable

Ez , MPa

νzr

Ruz , MPa

Er , MPa

νrz

Rtr , MPa

Steel wire strand GFRP CFCC

130000 75000 137000

0.3 0.19 0.3

1568 1765 1000

– 9200 8670

– 0.05 0.014

– 78 186

In Table 1 Ruz are the limits of strengths of components in the direction Z; Rtr are the compression strengths of components in the radial direction.

Figure 3. Dependence of pressure pa on the axial force N and angle of steel wire strands twisting α.

736

The value of the axial force N , acting in the cable, changes within the limits of 550 to 750 kN. The following values of the axial forces are considered because with the growing of the axial force value exceeding 750 kN, the GFRP component of the cable is excluded from the work. 2.3 Determination of radial and tangential stresses, acting in the components of hybrid composite cable The dependence of maximum radial σGr and tangential stresses σGθ , acting in the CFRP component of the hybrid composite cable on the axial force N is shown in Figure 4. The values of the radial and tangential stresses are equal for the CFRP component of hybrid composite cable. Comparison of the maximum radial stresses, acting in the CFRP components of the hybrid composite cable with their strengths shows, that the stresses are 1.65 times less, than the strength values, but the maximum value of tangential stresses are 13.79 times less, than the strength value. Maximum values of radial stresses were compared with the compression strength of CFRP in the direction perpendicular to the direction of fiber orientation, which is equal to 186 MPa. Tangential stresses were compared with the compression strength of CFRP in the direction corresponding to the direction of fiber orientation, which is equal to 1558 MPa. The maximum values of radial σGr and tangential σGθ stresses acting in the GFRP component of hybrid composite cable are given in Table 2 in depending on the angle of steel wire strands twisting α. The maximum values of radial and tangential stresses were obtained when the axial force N , acting in the cable, was equal to 750 kN and r = a. Comparison of the maximum tangential and radial stresses stresses, acting in the GFRP components of the hybrid composite cable with their strengths shows, that the maximum angle of steel wire twisting for the considered case is 20 degrees, when the radial stresses are equal to 45.02 MPa, which are 1.73 times less than the strength value.

Figure 4. Dependence of the radial σGr and tangential σGθ stresses acting in the CFRP on the pressure pa and angle of steel wire strands twisting α. Table 2. Maximum radial and tangential stresses, acting in the GFRP component of hybrid composite cable. Angle of steel wire strands twisting α, degrees.

σGr , MPa

σGθ , MPa

10 20 30

10.77 45.02 114.81

21.29 44.85 93.32

737

3 DECREASE OF DISPLACEMENTS OF COMPOSITE SADDLE-SHAPED CABLE ROOF A saddle-shaped cable roof 50 × 50 m in the plan was investigated. The existence of two symmetry planes allows us to regard, as a design scheme, a quarter of the cable net of a saddle-shaped cable roof with a compliant supporting contour in the shape of the cable truss, which is subjected to the prestressing and vertical design load (Figure 5).Three quarters of the cable roof are replaced by the bonds imposed on its one-quarter part. Hybrid composite cables with an elastic modulus of 1.32 · 105 MPa were taken as a material of cable truss elements, since they are most loaded ones. Relatively low elastic modulus of hybrid composite cable is joined with the necessity to use CFRP with the maximum limiting strains (1.6%). Steel cables with an elastic modulus of 1.3 · 105 MPa were assumed as a material for the suspension and stressing cables. From the viewpoint of material consumption, the saddle-shaped cable roof has rational geometrical characteristics: the initial deflection of the contour cables was 8.6 m, the initial deflections of suspension and stressing cables 20 m, and the step in plan of the latter ones was 1.414 m. The structure was calculated for the basic combination of loads – the dead weight of the structure (0.27 kPa) and the weight of snow (1.12 kPa) – evenly distributed on the horizontal projection of the roof. The design load in the form of point wise forces was applied to the nodes of the cable net. The roof had the following layers: a glass net coated with polymer resin (2 mm), foam plastic, reinforced with a glass net (120 mm), and saddle-shaped plywood sheets (6 mm). The cable net was prestressed by applying tension forces to the suspension and stressing cables, such that the residual tension forces in the stressing cables were equal to 20% of their initial values under the vertical design load. The relations between the initial deflection of the top chord of the cable truss, distance between the nodes of the cable truss and the volume of the material of the cable net per unit of the covered area (relative volume) and maximum vertical displacement of the cable net were determined in the form of second power polynomial functions using the method of experimental design. The coefficients of the second power polynomial functions were found from the results of a numerical experiment, which was joined with the determination of forces in net cables, which are necessary to select the cable cross-section and calculate the relative volume of the material of the cable net and maximum vertical displacements of the cable net. The numerical experiment was conducted with the values of initial deflection of top chord of the cable truss, changing from 2.15 to 6.45 m and values of the distance between the nodes of cable truss, changing from 2.5 to7.5 m.

Figure 5. a) Quarter of cable net, supported by compliant supporting contour: 1 – top cable of supporting contour, 2 – tie-bar, 3 – bottom cable of supporting contour, fk1 – initial deflection of top chord of compliant supporting contour, al – distance between the support points of tie-bars. b) Shape of prestressed cable net after vertical design load application.

738

The area, covered by the roof was found with regard to the initial deflections of top chord of the cable truss. It means that the area, covered by the roof includes areas, which are covered by the cable trusses. Using a computer program “ANSYS/ED 5.3” for WINDOWS the numerical experiment was carried out. The program enables to calculate values of the tension forces acting in the cables of the net and maximum vertical displacements of the cable net. In calculating a cable net, the program uses the Newton-Raphson iteration method, which consists of the division of the applied vertical design load into several parts in an ascending order. The cable net was modeled by finite elements of LINK10 type, with three degrees of freedom for each node. Each finite element was divided into two parts of the same length. The judicious values of the basic geometrical characteristics of the cable truss were found from the system of equations:

where fk1 = initial deflections of top chord of the cable truss, al = distances between the nodes of the cable truss. The first and second equation of the system was obtained by taking of partial derivations from the second power polynomial functions by the initial deflection of top chord of the cable truss and distance between the nodes of the cable truss respectively. Coefficients of the equations were equal to: ξ0 , ξ1 and ξ0 are equal to −0.1637, 0.5257 and 0.4682 respectively; χ0 , χ1 and χ2 are equal to 0.0705, 0.2710 and 0.5257 respectively. The values of the initial deflection of top chord of the cable truss and distance between the nodes of the cable truss are equal to 3.13 and 7 m respectively. Maximum vertical displacements of the cable roof for all combinations of the main geometrical characteristics of the cable truss were determined as a maximum difference in the vertical coordinate of the cable net nodes before and after application of design vertical load. It was stated, that the minimum values of vertical displacements of cable net were obtained, when the initial deflection of top chord of cable truss were equal to 3.13 m and distance between the nodes of the cable truss was equal to 7 m. Application of hybrid composite cable as a material of supporting contour instead of steel enables to decrease by 1.3% maximum vertical displacements of the cable net. The using of the cable truss as a structure of support contour enables to decrease by 8% the maximum vertical displacements of the cable net in the case, when the tension cables of the net are hybrid composite cables, but suspension and stressing cables of the net are made of steel.

4 CONCLUSIONS The dependence of external pressure per unit of the surface area of the GFRP (due to the pressure of steel wire strands) pb of hybrid composite cable on the axial force N and angle of steel wire strands twisting α was obtained. It was shown, that increasing of angle of wire twisting α from 10 to 30 degrees causes growing of external pressure per unit of the area of CFCC by 14.61 times when the axial force increases from the 550 to 750 kN. Tangential and radial stresses for GFRP and CFCC components of hybrid composite cable were obtained. It was shown, that the maximum angle of steel wire strands twisting α is equal to 20 degrees for the considered hybrid composite cable. It was shown, that maximum radial stresses σGr acting in the GFRP component of hybrid composite cable, when the angle of the steel wire twisting α was equal to 20 degrees, and the axial force N was equal to 750 kN, was 1,73 times less than the strengths of GFRP. 739

Opportunity to decrease the displacements of composite saddle-shaped cable roof by the using of cable trusses as a supporting contour structure was investigated. It was shown by the numerical experiment, that the rational initial deflections of top chord of the cable truss and distance between the nodes of the cable truss for the cable roof with dimensions in plan 50 × 50 m are equal to 3.13 and 7 m, respectively. It was shown, that the using of cable truss as a structure of supporting contour enables to decrease by 8% the maximum vertical displacements of the cable net in the case, when the tension cables of the net are hybrid composite cables but suspension and stressing cables of the net are made of steel. REFERENCES Bengtson, A. 1994. Fatigue Tests with Carbon-Fiber-Reinforced Composite Cable as Nonmetallic Reinforcement in Concrete. Göteborg: 1–14. Peters, S.T. 1998. Handbook of composites. London: 758–777. Kumar, K. & Cochran, Ir.I.E.1997. Closed form analysis for elastic deformations of multilayered strands, Journal of Applied Mechanics, ASME Vol.54: 898–903. Costello, G.A. 1997. Theory of wire rope, second edition. NewYork : Springer. Serdjuks, D. & Rocens, K. Hybrid Composite Cable Based on Steel and Carbon, Materials Science, Vol.9, No1, ISSN 1392–1320: 27–30. Pakrastinsh, L. & Serdjuks, D. & Rocens, K. 2001. Some structural possibilities to decrease the compliance of saddle shape cable structure. Proceedings of 7th International Conference Modern Building Materials, Structures and Techniques: 18–24. Serdjuks, D. & Rocens, K. & Pakrastinsh, L. 2000. Utilization of Composite Materials in Saddle-Shaped Cable Roof, Mechanics of Composite Materials, Vol.36, No5., ISSN 0191 – 5665.: 385–388. Serdjuks, D. & Rocens, K. 2003. Evaluation of mechanical interaction between components in hybrid composite cable, Architecture and Constructional Science, Scientific Proceedings of Riga Technical University, ISSN1407-7329: 208–215. Rocens, K. & Verdinsh, G. & Serdjuks, D. & Pakrastinsh, L. 1999. Structure of Composite Roof. Patent No 12191 of the republic of Latvia. Serdjuks, D. & Rocens, K. & Mitrofanov, V. 2002. Behavior of Hybrid Composite Cable in Saddle Shape Roof, Scientific proceedings of RigaTechnical University: Vol.1, Architecture and constructional science:162–169.

740

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

From tree trunk to tube or the quadrature of the circle P. Haller Technische Universität Dresden

ABSTRACT: Generally spoken, difficulties in wood construction arise from four sources: the bad material efficiency compared to technical materials, the low strength spectrum, anisotropy as well as preservation. This paper deals with several technologies coping with these drawbacks. A new technology that transforms raw wood into profiles which considers the material as a cellular solid is being presented. Technical textiles turn out as a versatile technique that improves structural performance and serviceability.

1 INTRODUCTION The importance of wood as a building material and its role to be played in a sustainable development will decisively depend on the amount to which it will guarantee an economic and high-quality solution of our tasks. Therefore it is not enough to possess a renewable raw material it also has to come up to today’s and future expectations. Engineers decide on the use of certain materials according to technical and economic points of view. In the course of their academic education they deal with a great variety of building and engineering materials that still will increase in the future. In contrast to craftsmen engineers are not bound to use a special material, nevertheless wood is excluded from use in many fields of engineering right from the beginning. Only in civil engineering there is the application of wood still seriously taken into account. Despite ecological advantages a decision in favour of wood always requires technological and economical arguments. For that reason science and technology should create the preconditions for an efficient and more frequent use of this resource. The forest is not only one of the greatest but also one of the cheapest producers of material in the world. It is difficult to believe that a material produced on one third of the area of our country with the help of solar energy is inferior with regard to its price to materials that are produced using large amounts of fossil energy and capital. One significant reason for this can be seen in the fact that the forest is not considered as producer of material but as producer of cross sections. We will have a closer look at this later. Moreover, we have to ask what further disadvantages prevent the use of wood for technical applications and whether they can be removed or not. In particular these are 1. the low strength spectrum as compared to structural materials 2. the directional dependence of the mechanical properties, the so-called anisotropy, and 3. the low weather resistance.

2 DENSE, DENSER, DENSEST Wood has a well-balanced profile of properties, but nearly always specialised materials surpass certain properties of wood. It is without question that wood is ecologically beneficial as long as this is not ruined by additional treatment. Also the price per unit mass, that even allows a thermal use, is cheaper than of most other materials of our time. 741

Figure 1. Spruce squared timber cross section before and after densification.

Figure 2. Strength values and classes of today’s building materials and newly developed building materials.

The mechanical parameters play a central role for load bearing structures and mostly depend on density and growth structure. The differences among different kinds of wood amount to approximately one order of magnitude. The comparison between structural building timber and timber with parallel fibres and without knots and irregularities in growth presents additional differences so that the unused strength potential increases to a total of a good order of magnitude. The densification of wood (see fig. 1), especially hard wood, using heat and pressure is a technique in wood production known since long. Also nowadays resin-bonded veneer panels, e.g. for electric installations, are produced in this way. The precondition for the densification is the cell construction of the wood that allows densification by means of a press after the softening temperature of the lignin was reached. Thanks to this thermo-mechanical treatment strength and stiffness can be increased proportionally to densification. Further heating above 200◦ C leads to an increase in biological resistance, so that the heat influences two important properties, i.e. strength and durability. Figure 2 presents strength classes of different building materials and their compounds. In this regard wood shows low increments that can be significantly increased by the use of timber with parallel fibres and densification (see fig. 3). 742

Figure 3. Strength values of soft wood (spruce) with parallel fibres, untreated; oil-heat treated; densified; densified and oil-heat treated.

3 FROM A TRUNK TO A CROSS SECTION The growth of a tree and its cutting in the saw mill on the one hand lead to a lot of waste and on the other hand to full cross sections that as compared to technical profiles reach low area moments. The forest as producer of material belongs to the most low-priced sellers, but its competitiveness gets lost while the raw material is transformed into cross sections. Therefore it is absolutely necessary to check all possibilities of material economy in the production of cross sections. The techniques in the saw mill present the first and most important step in the production of cross sections. The relation between output and waste significantly determines the processing and thus the price margin of other partly competing wooden products. This technique considers only one dimension and favours “one-dimensional” kinds of trees. So that in case of reforestation coniferous soft wood, especially spruce, is preferred to hard wood typical for the region as e.g. oak-trees or beeches with their widespread tree-tops. Wood is said to be worked easily, but the opposite is true. Wood is transformed into cross sections by cutting and joining with synthetic bonding agents afterwards. This does not demand any knowledge of the microstructure. But just this presents a great potential for the development of new techniques and products that has not been paid sufficient attention to in science and technology until now. The saw mill delivers a squared rectangular cross section that – as compared to technical profiles made of metal or plastic – has a low efficiency of material. If one adds the bad output of wood by sawing one receives a quite dramatic result. Starting from round timber figure 4 demonstrates the output of material and the area moment reached in the production of different kinds of cross sections. At first sight we are tempted to assume especially good qualities for bearing structures there where we find high strength. But this has to be looked at more closely. What do engineers do when they are planning bearing structures? They transfer forces and moments with the help of the product of a material factor, i.e. the strength, and a geometrical factor, i.e. the cross-sectional area or the moment of area. In simple words: if a material is only half as strong its cross-sectional area will be doubled. But it cannot be more than doubled because with area moments the distance between cross section and neutral fibre is raised to a power. Therefore structural components are easier to be dimensioned by varying the dimensions of the cross section but by changing the strength class. The way of choosing round or square solid cross sections in timber engineering hides the fact that the resource productivity is low. In this respect a comparison between squared timber and technical profiles shows a relation of approximately 1:15, what on the one hand results from the 743

Figure 4. Output of wood with reference to round timber and flexural rigidity EI for different techniques of cross section production.

Figure 5. Comparison of area moment I of the square solid cross section with profiles of the same area.

losses in the saw mill and on the other hand from the low moment of area of the solid cross section (see fig. 5). Since timber does not directly depend on the cross section it has to be optimally placed there according to mechanical considerations and has to fulfil the following three conditions: 1. the cross section must not be limited by transverse or longitudinal dimensions of the tree 2. it has to be efficient, i.e. it has to have a great area moment for a given area 3. a cheap production of large quantities must be guaranteed. Squared timber does not meet condition 1 and 2; glued timber does not meet condition 2 and 3. Only the shaped timber profile shown in figures 6 and 7, based on a new understanding of the material, has the potential to meet all tree conditions. 4 TIMBER IN TOP FORM As far as production techniques are concerned timber construction relies on two basic processes: dividing, i.e. sawing, planing, shredding etc., and then joining by synthetic or metallic fasteners. Already nowadays there is a great variety of possible constructions based on each of these basic processes and their combinations. Imagine this variety could still enlarged by one or two additional ones. Domestic soft wood has a porosity of about 60%. Its polymeric structure allows slight plastic deformation transversally to grain at a temperature of 140◦ C and a pressure of 5 MPa. Thus the dimension of the cross section can be approximately halved (see fig. 1), whereby the microstructure of the wood folds up. This possibility to improve mechanical properties was already mentioned in the preceding paragraph. It is also important to know that it is possible to nearly completely reverse and fix the compression without causing any damage to the microstructure if a suitable process is applied. 744

Figure 6. Process of production of shaped wooden profiles made of square or round timber.

Figure 7. Ring-shaped cross section made of densified half-round timber.

Its great porosity allows to consider the wood in a completely new manner as a foam-like, cellular material that now indeed becomes a material easily to be processed. Thus fracture elongation transversally to grain increases from one to 100 per cent, i.e. by two orders of magnitude. Soft and hard wood are both suited for this. Starting from these thoughts at the Institute for Steel andTimber Structures there were made glued laminated timber boards and densified in the direction of the plane. Afterwards under certain heat and humidity conditions there were produced prismatic cross sections reversing the compression by completely folding up the cells. The bending radius of the deformation depends on the preliminary densification. Depending on the production technique the minimum bend corresponds to about twice the thickness of the board. This way basically all open and closed prismatic cross sections of any length can be produced. According to this method, which meanwhile was patented, tubes of structural dimensions have been successfully produced. Figure 6 shows an example that begins with the densification of round 745

Figure 8. Cross sections of tubes with textile reinforcement; left – carbon fibre; center – without reinforcement; right – glass fibre, varnished.

timber. The division in the direction of maximum density and subsequent gluing lead to a solid panel that can be transformed into a tube by means of thermo-mechanical treatment. As compared with the round timber material economy amounts to about 80 per cent. 50 per cent of it can be saved by avoidance of waste in the saw mill and the rest by an efficient placement in the profile. 5 THREAD MEETS FIBRE When timber is used for bearing structures not only mechanical and biotical behaviour are of great importance but also its anisotropy. The first-mentioned can be improved by sorting and thermal and/or thermo-mechanical procedures whereas the directionality of strength is met by different measures in design. Strength and rigidity can very efficiently be compensated in the course of dimensioning the cross section in longitudinal direction. But even experienced structural engineers face problems dealing with shear and transverse stresses. Meanwhile a lot of different solutions and design methods are available that led to complex special knowledge. Therefore it is desired that the problems connected with anisotropy shall be met by a universal technology. A look at nature could teach a lot of things because many natural constructions meet mechanical stresses by optimally directed fibres: as e.g. crotches of a tree, blades of straw or muscles. Fibre reinforced plastics present a technical application according to this example. The connection of threads to flat or three-dimensional structures is a subject of textile technology. The Collaborative Research Centre (SFB, Sonderforschungsbereich) 528 “Textile Reinforcement for Structural Strengthening and Retrofitting” at the Faculty of Civil Engineering examines their application in civil engineering. This Collaborative Research Centre also elaborates the fundamentals of textile reinforcement of timber structures. The cooperation with the Institute of Textile and Clothing Technology enables the timber engineers in Dresden to apply fully fashioned stress related textile reinforcements made of glass, carbon, aramide or natural fibres that are glued on by synthetic resins afterwards. Technical textiles help to build a bridge between timber engineering and light weight construction what is thought to lead to a completely new quality in the use of this renewable resource. Besides the mechanical behaviour of the construction the low durability of organic building materials proves to be a decisive disadvantage for exterior application that nowadays is answered by modified wood properties and structural design. But in both cases will arise additional costs. 746

The complete reinforcement of whole building components in connection with surface treatment as in light weight construction will not only provide structural reinforcement but also an effective protection against weathering. This is an important advantage not only what concerns humidity but also with regard to a corrosive environment.

6 CONCLUSION The presented developments deal with all shortcomings of present technical applications of wood and in the author’s opinion fundamental solutions are offered. This concerns the efficient use of the raw material that leads to low material prices; the densification of wood that surmounts the limits of the strength classes; textile reinforcement as technology that completely solves the problem of anisotropy at a favourable price and also provides weather protection; and the shaping of efficient cross section profiles as probably the most far-reaching innovation. These new developments can be applied everywhere where cross sections are needed. These may be bearing elements in civil engineering as columns and girders, in light-weight and equipment construction, but also non-bearing parts for furniture or interior work. Moreover a lot of things with an open or closed prismatic cross section can be produced this way, e.g. cable drums, poles, barrels, tanks, rotor blades or hulls. Wood will become of greater technical importance if its properties, cross sections and production techniques can come up to the expectations of engineers more properly. Old constructions always are bound to meet old reservations. So it is easier to apply new methods as astonishingly as this may sound. Wood has the potential for innovations based on material and techniques. That there are few innovations is not to be explained by the wood itself but by structures impeding its development. REFERENCES Haller, P. 2004. 53 Heft 1-2 Vom Baum zum Bau oder die Quadratur des Kreises. in: Wissenschaftliche Zeitschrift der Technischen Universität Dresden, S.100–104, 53 (2004) Heft 1-2 Haller, P. 2003. Design and Optimization in Wood Construction – Further Developments. Proceedings: Workshop on Optimal Design of Materials and Structures; (Hrsg.) Zarka, J.; Laboratoire de Mécanique des Solides, Ecole Polytechnique, Palaiseau, France Haller, P., Birk, T. 2003. Der Einsatz von multiaxialen Nähgewirken und Biaxialgestricken zur Verstärkung von Holzkonstruktionen. Tagungsband, 2nd Colloquium on Textile Reinforced Structures, Institut für Massivbau (Hrsg.), Technische Universität Dresden Haller, P., Putzger, R. 2003. Charakterisierung der Verbundfestigkeit und Dauerhaftigkeit von textilbewehrtem Holz. Tagungsband, 2nd Colloquium on Textile Reinforced Structures, Institut für Massivbau (Hrsg.), Technische Universität Dresden Haller, P., Wehsener, J. 2003. Entwicklung innovativer Verbindungen aus Pressholz und Glasfaserarmierung für den Ingenieurholzbau. Forschungsbericht AiF, Nr. 11164 B, (Hrsg.) Fraunhofer IRB Verlag, Stuttgart

747

Improvement of Buildings’ Structural Quality by New Technologies – Schaur et al. © 2005 Taylor & Francis Group, London, ISBN 04 1536 609 7

Author index Airumyan, E.L. 431 Aktuglu, Y.K. 417, 421 Altay, G. 165 Antimovski, A. 115 Apon, L. 503, 519 Aras, F. 165 Aste, C. 3, 65 Badea, C. 377 Bálint-Major, S. 125 Barecchia, E. 81 Bliudzius, R. 573 Blok, R. 73 Bob, C. 377 Boelman, E.M. 503, 519 Boiko, O.I. 431 Bouchair, A. 311 Bozinovski, Z.Lj. 649 Bragança, L. 495, 527, 555, 565 Brunner, M. 219, 233 Bujnak, J. 441 Butz, C. 481 Byfield, M.P. 333 ˇ Cajka, R. 285, 369 Calderoni, B. 301, 607 Coccia, S. 207 Cvetkovska, M. 277 De Luca, A. 341, 629 De Martino, A. 301 De Matteis, G. 91, 135, 267, 607, 665 De Nardin, S. 473 Della Corte, G. 81, 105, 359 Dienes, E. 547 Dinu, F. 291 Doveika, V. 193 Dubina, D. 125, 291 El Debs, A.L.H.C. 473 Faggiano, B. 267, 359, 587 Feldmann, M. 455 Feng, J. 201

Ferrer, M. 447 Fiorino, L. 105 Fischinger, M. 623, 677 Formisano, A. 135, 301, 665 Fülöp, L.A. 125 Fumo, M. 393 Galvonaite, A. 193 Georgin, J.F. 701 Gervásio, H. 527 Gesella, H. 455 Giordano, A. 341 Giubileo, C. 607 Glatzl, A. 3 Gramatikov, K. 649 Grande, E. 341, 629 Grecea, D. 291 Gregor, D. 709 Haddad, Y.M. 201 Haller, P. 599, 741 Hechler, O. 481 Heiduschke, A. 599 Herwijnen van, F. 73 Hobusch, T. 401 Huber, G. 3, 65 Huovila, P. 425 Ianniruberto, U. 207 Isakovi´c, T. 677 Kaliske, M. 21 Kamynin, S.V. 431 Kantchev, V.K. 225 Kante, P. 623 Kasal, B. 599 Kind, S. 387 Köber, H. 177, 323 Kokalevski, M. 115 Kosteas, D. 537, 547 Koukkari, H. 425, 495 Kozłowski, A. 35, 639 Krstevska, L. 657 Kruger, S. 401 Kubiszyn, W. 57 749

Kudzys, A. 149, 615 Kuhlmann, U. 239, 247, 463, 693 Landolfo, R. 105, 267 Langone, I. 91 Lazarov, L. 277 Limam, A. 701 Mandara, A. 45 Marimon, F. 447 Marzo, A. 587 Mateckova, P. 285 Mateus, R. 495, 565 Mazzolani, F.M. 11, 45, 81, 91, 267, 359, 587, 607, 665 Mele, E. 341, 629 Mendonça, P. 555 Mistakidis, E.S. 135 Monstvilas, E. 573, 579 Naponiello, M. 393 Nguyen, D.T. 701 Odrobinak, J. 441 Pahl, B. 401 Pakrastinsh, L. 185 Panico, S. 665 Partov, D.N. 225 Pisarek, Z. 639 Plewako, Z. 213 Pluto, C. 401 Radev, D. 155 Radeva, S. 155 Radlbeck, C. 537, 547 Ravesloot, C.M. 503, 511, 519 Rensburg van, B.W.J. 351 Reynoard, J.F. 701 Rinaldi, Z. 207 Ristic, D. 657 Rocens, K. 185, 733 Roure, F. 447 Rybinski, M. 463 Rybka, A. 409

Šadauskiene, J. 579 Schäfer, M. 693 Schänzlin, J. 239, 247 Schlinz, M. 537 Schmidt, J. 21 Schnüriger, M. 219, 233 Serdjuks, D. 733 Silva da, L.S. 527 Simkus, R. 193, 615

´ e
czka, L. 35 Sl Snarskis, B. 193 Sokol, Z. 259 Stankevicius, V. 573, 579 Stef ¸ a˘ nescu, B. 177, 323 Stratan, A. 291 Szojda, L. 683

750

Tashkov, Lj. 115 Trumpf, H. 481 Tsirnovas, S.I. 135 Wald, F. 259, 709 Wolinski, S. 717 Wróbel, K. 57 Wuwer, W. 725