Phebus Fission Product Project: Presentation of the experimental programme and test facility (Publication No. Eur 13520 En of the Commission of the Europe)

THE PHEBUS FISSION PRODUCT PROJECT

Presentation of the Experimental Programme and Test Facility

Proceedings of the Seminar of the Phebus-FP (Fission Product) Project, presentation of the experimental programme and test facility, organised by the “Commissariat à l’Energie Atomique (CEA), Institut de Protection et de Sûreté Nucléaire (IPSN), Centre d’Etudes Nucléaires de Cadarache (CEN) and the Commission of the European Communities (CEC), Directorate General Science, Research and Development, Joint Research Centre (JRC) Ispra, Safety Technology Institute (STI) and held at the Château Cadarache, St. Paul-Lez-Durance, France, 5–7 June 1991. Programme Committee Members:

Co-ordinators:

C.LECOMTE—CEA/CEN Fontenay-aux-Roses M.GOMOLINSKI—CEA/CEN Fontenay-aux-Roses A.MEYER-HEINE—CEA/CEN Cadarache P.VON DER HARDT—CEC/JRC Ispra A.G.MARKOVINA—CEC/JRC Ispra M.C.RUBINSTEIN—CEA/CEN Cadarache W.KRISCHER—CEA/JRC Ispra

Local organisation: M.C.RUBINSTEIN Publication arrangements: D.NICOLAY—CEC Luxembourg Scientific secretary: W.KRISCHER—CEC/JRC Ispra

THE PHEBUS FISSION PRODUCT PROJECT Presentation of the Experimental Programme and Test Facility

Edited by W.KRISCHER CEC Joint Research Centre, Ispra Site, Ispra (VA), Italy and M.C.RUBINSTEIN CEA, IPSN, Centre d’Etudes Nucléaires de Cadarache, St Paul-Lez-Durance, France

ELSEVIER APPLIED SCIENCE LONDON and NEW YORK

ELSEVIER SCIENCE PUBLISHERS LTD Crown House, Linton Road, Barking, Essex IG11 8JU, England 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.” Sole Distributor in the USA and Canada ELSEVIER SCIENCE PUBLISHING CO., INC. 655 Avenue of the Americas, New York, NY 10010, USA WITH 23 TABLES AND 95 ILLUSTRATIONS © 1992 ECSC, EEC, EAEC, BRUSSELS AND LUXEMBOURG British Library Cataloguing in Publication Data Phebus Fission Product Project I. Krischer, W. II. Rubinstein, M.C. 621.48 ISBN 0-203-21351-3 Master e-book ISBN

ISBN 0-203-27038-X (Adobe eReader Format) ISBN 1-85166-765-2 (Print Edition) Library of Congress CIP data applied for Publication No. EUR 13520 EN of the Commission of the European Communities, Dissemination of Scientific and Technical Knowledge Unit, Directorate-General Telecommunications, Information Industries and Innovation, Luxembourg. LEGAL NOTICE Neither the Commission of the European Communities nor any person acting on behalf of the Commission is responsible for the use which might be made of the following information. No responsibility is assumed by the Publisher for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use of operation of any methods, products, instructions or ideas contained in the material herein. Special regulations for readers in the USA This publication has been registered with the Copyright Clearance Center Inc. (CCC), Salem, Massachusetts. Information can be obtained from the CCC about conditions under which photocopies of parts of this publication may be made in the USA. All other copyright questions, including photocopying outside the USA, should be referred to the publisher. All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, without the prior written permission of the publisher.

FOREWORD

Severe, hypothetical accidents in light water reactors may lead to the release of radioactive material, mainly fission products, from the containment barriers. The knowledge of that release and of its evolution with time and in function of certain depletion phenomena is a major point for an efficient accident management and emergency planning, as it allows to estimate the Source Term, i.e. quantity and physical-chemical quality of radioactive releases to the environment. Computer codes have been developed and several experimental programmes performed in an attempt to understand and quantify the complex sequences of events involved in core degradation and fission product transport. Based upon the scientific and technical information made available internationally by experiments and computations, the French Commissariat a l’Energie Atomique proposed in 1985 a series of in-pile integral experiments to study the release and transport of fission products under the most representative conditions: the principle was to study phenomena taking place in core, primary circuit, containment building, and the release to the environment in an experimental facility where the coupling between phenomena could be also investigated. Consequently, the Phebus Fission Product (FP) programme has been settled with the main objective to improve the understanding of FP physical and chemical behaviour, i.e. their emission, transport and evolution in the primary circuit and containment. The experiments are designed to study also the degradation of high burn-up fuel, particularly during the later phases of the transient. The Commissariat à l’Energie Atomique and the Commission of the European Communities agreed in 1988 to carry out the Phebus-FP project as a joint effort, and to open it to international collaboration. A new facility has been designed over the past two years using the existing Phebus test reactor at Cadarache, France. Fission products from overheated high burn-up fuel will be swept, by hot steam and hydrogen, through simulated reactor cooling system components into a tank simulating the reactor containment. Kinetics and results of FP transport, chemical reactions, and depletion will be measured by appropriate instrumentation and by post-test analysis. The first in-pile test is planned for autumn 1992, and five subsequent tests will follow in yearly intervals. The seminar on the Phebus-FP project was organized by the French Commissariat à l’Energie Atomique and the Commission of the European Communities. The objective of the seminar was to present the state of the art in LWR (Light Water Reactor) Source Term evaluation, the needs for safety analysis and the lessons learned from relevant projects. The presentations outlined the contributions and limitations of the Phebus-FP in solving identified problems and in providing a qualified database for the validation of code systems for the Source Term estimation during a postulated severe accident. Essential project features such as the instrumentation of tests and analytical support were illustrated. The seminar was closed by a panel discussion where a summary of the main points under discussion during the different sessions was presented and the speakers gave their recommendations on the Phebus-FP programme and, in general, severe accident research. A visit of the Phebus plant was included. P.Fasoli-Stella

CONTENTS

FOREWORD

v

ABBREVIATIONS

ix

WELCOME ADDRESS M.LIVOLANT , Director of Research of the Protection Safety and Nuclear Institute (IPSN)

1

J.P.CONTZEN , Director General of the Joint Research Centre, Commission of the European Communities

3

SESSION I SURVEY OF LWR SEVERE ACCIDENT SOURCE TERM RESEARCH LWR SEVERE ACCIDENT SOURCE TERM RESEARCH IN EUROPE M.LIVOLANT , CEA/IPSN Fontenay-aux-Roses, France

6

LWR SEVERE ACCIDENT SOURCE TERM RESEARCH IN THE USA T.P.SPEIS , R.Y.LEE , L.SOFFER and R.O.MEYER , US Nuclear Regulatory Commission, Washington

10

SURVEY OF SEVERE ACCIDENT EXPERIMENTS AND ANALYSES IN JAPAN M.AKIYAMA , University of Tokyo, K.TAKUMI , Nuclear Power Engineering Center, Tokyo and K.SODA, JAERI , Tokai-mura, Japan

23

SUMMARY OF DISCUSSION S.FINZI , CEC Brussels

35

SESSION II STATE OF THE ART DEDUCED FROM PREVIOUS LARGE EXPERIMENTS CORE DEGRADATION AND FISSION PRODUCT RELEASE R.W.WRIGHT , US Nuclear Regulatory Commission, Washington and S.J.L.HAGEN , KfK, Karlsruhe, Germany

37

FISSION PRODUCT TRANSPORT J.O.LILJENZIN , Chalmers Tekniska Högskola, Göteborg, Sweden and W.SCHÖCK , KfK, Karlsruhe, Germany

46

FISSION PRODUCT CHEMISTRY IN SEVERE REACTOR ACCIDENTS: REVIEW OF RELEVANT INTEGRAL EXPERIMENTS A.L.NICHOLS , AEA Technology Winfrith, UK and C.HUEBER , CEA/IPSN Cadarache, France

63

vii

PHEBUS-CSD PHEBUS SEVERE FUEL DAMAGE PROGRAMME: MAIN EXPERIMENTAL RESULTS AND INSTRUMENTATION BEHAVIOUR C.GONNIER , C.REPETTO , CEA/IPSN Cadarache and G.GEOFFROY , CEA Saclay, France

79

REVIEW OF B9+ BENCHMARK RESULTS B.ADROGUER , CEA/IPSN Cadarache, France and P.VlLLALIBRE , CSN Madrid

91

SUMMARY OF DISCUSSION A.MEYER-HEINE , CEA/IPSN Cadarache, France

104

SESSION III CORE AND FP-BEHAVIOUR SAFETY ANALYSIS NEEDS AND MAIN PHENOMENA TO BE STUDIED J.GAUVAIN , CEA/IPSN Fontenay-aux-Roses, France and H.M.VAN RIJ , CEC/JRC Ispra, Italy

106

PHEBUS-FP OBJECTIVES, TEST MATRIX AND REPRESENTATIVITY OF THE PHEBUS-FP EXPERIMENTAL PROGRAMME A.ARNAUD , CEA/IPSN Cadarache, France and A.MARKOVINA , CEC/JRC Ispra, Italy

110

PHEBUS-FP TEST FACILITY PH. DELCHAMBRE , CEA/IPSN Cadarache, France and P.VON DER HARDT , CEC/JRC Ispra, Italy

121

PHEBUS-FP INSTRUMENTATION P.VON DER HARDT , CEC/JRC Ispra, Italy and G.LHIAUBET , CEA/IPSN Fontenay-aux-Roses, France

132

CEA ANALYTICAL ACTIVITIES: HEVA, PITEAS, MINI-CONTAINMENTS C.LECOMTE and G.LHIAUBET , CEA/IPSN Fontenay-aux-Roses, France

148

CEC SUPPORT ACTIVITIES: EC SHARED COST ACTIONS AND OTHERS P.FASOLI-STELLA and A.MARKOVINA , CEC/JRC, Ispra, Italy

159

PHEBUS-FP: ORGANISATION OF THE PROJECT AND INTERNATIONAL COLLABORATION A.TATTEGRAIN , CEA/IPSN Cadarache, France and P.VON DER HARDT , CEC/JRC Ispra, Italy

168

SUMMARY OF DISCUSSION P.FASOLI-STELLA , CEC/JRC Ispra, Italy

172

viii

SESSION IV ANALYTICAL ACTIVITIES SURVEY OF SOURCE TERM CODES M.R.HAYNS , AEA Technology Harwell and S.R.KINNERSLY , AEA Technology, Winfrith, UK

174

ESCADRE AND ICARE CODE SYSTEMS M.REOCREUX and J.GAUVAIN , CEA/IPSN Fontenay-aux-Roses, France

183

ESTER—A EUROPEAN SOURCE TERM EVALUATION SYSTEM A.V.JONES and I.M.SHEPHERD , CEC/JRC, Ispra, Italy

198

FPT0 TEST PRECALCULATIONS A.MAILLIAT , CEA/IPSN Cadarache, France, A.V.JONES and I.M.SHEPHERD , CEC/JRC Ispra, Italy

208

SUMMARY OF DISCUSSION A.TATTEGRAIN , CEA/IPSN Cadarache, France

238

PANEL DISCUSSION VALUE OF THE PHEBUS-FP AND RELATED SOURCE TERM STUDIES FOR THE SAFETY OF LWRs Chairman: H.F.HOLTBECKER (CEC) Members: M.AKIYAMA (JAP), R.E.VAN GEEN (B), M.PEZZILLI (I), C.LECOMTE (F), M.R.HAYNS (UK), T.P.SPEIS (USA), M.BANASCHIK (D)

240

CLOSING REMARKS

249

LIST OF PARTICIPANTS

250

INDEX OF AUTHORS

260

Abbreviations

AEA CEA CEC CSN IPSN JAERI JRC KfK NRC Univ.

Atomic Energy Authority (GB) Commissariat à l’Energie Atomique (F) Commission of the European Communities Consejo de Seguridad Nuclear (E) Institut de Protection et de Sûreté Nucléaire (F) Japan Atomic Energy Research Institute Joint Research Centre of the CEC Kernforschungszentrum Karlsruhe GmbH (D) Nuclear Regulatory Commission (USA) University

WELCOME ADDRESS M.Livolant Director of Research of the Nuclear Safety and Protection Institute (IPSN)

Ladies and Gentlemen, I have to excuse Mr. J.RASTOIN, Director of the Nuclear Safety and Protection Institute, who intended to open the seminar and welcome the participants. A last minute obligation at the CEA Headquarters made it impossible for him to come. This gives me the pleasure to welcome you in Cadarache for this first international Phebus Fission Product seminar. As you know, the Phebus-FP project was basically an enterprise between the Commission of the European Communities and the French Commissariat à l’Energie Atomique, but soon other organizations showed their interest for the project. The last Steering Committee was, for example, attended by representatives of NRC for the US, NUPEC (MITI) for Japan and Candu Owners Group (Ontario Hydro and AECL) for Canada. I have also to mention that the Korea Atomic Energy Research Institute (KAERI) very recently became a member of the project. May I take this opportunity to announce that this first seminar will be followed up periodically by meetings where the results of the tests will be presented in detail. These meetings will be open exclusively to the members of the project. You have received the programme of the seminar. Let me focus on the highlights of the meeting. After a survey of LWR severe accident source term research throughout the world, a state-of-the-art deduced from previous large experiments will be presented concerning the three main aspects of severe accident source term: - core degradation and fission products release; - fission products transport; - fission products chemistry. This will give some ideas about what is known on the subject. Then, safety analysis requirements will be discussed, to give some thoughts to what else is needed on the subject. Subsequently a description of the Phebus project will be given including supporting analytical activities on source term carried out mainly in Europe. Ideally, this should allow to show how this project satisfies safety analysis requirements. In practice, it is well known that the connection between safety analysis requirements and research activities is not easy to establish and maintain. Finally, as Director of Research of the Nuclear Safety and Protection Institute (IPSN), I would like to make some remarks about the team in charge of the preparation of the tests. As you might know, prior to the Phebus-FP project, two other projects were realized in the same installation: - the test series Phebus LOCA (Loss of Coolant Accident), and - the test series Phebus CSD (Severe Fuel Damage). The same team has been, and still is, in charge of the following experiments concerning the reactivity of cooling accidents in fast breeders: - SCARABEE, - CABRI 1 and CABRI 2. CABRI 1 has been terminated whereas the other two experiments are continuing activities. The success of these large-scale experiments gives us some confidence that, even if the Phebus-FP project is difficult from the experimental point of view, the team in charge has a profound experience in the field, which is one of the strong points concerning this project. The Phebus-FP is presently the largest in-pile experiment in the field of LWR severe accident source term research. I am glad that this experiment will take place in Cadarache and that it gives us the occasion of a wide international cooperation. I hope that the seminar will be sufficiently persuasive to convince other countries to join in and become members of the Phebus-FP project.

2

THE PHEBUS FISSION PRODUCT PROJECT

Ladies and gentlemen, I thank you for your attention.

WELCOME ADDRESS J.P.Contzen Director General of the Joint Research Centre, Commission of the European Communities

Mr. Chairman, Ladies and Gentlemen, In these introductory remarks, I would like to stress the genuine interest of the Commission of the European Communities for the project which is the subject of our discussions during the seminar and in the name of the Commission, I wish to convey our best wishes for the success of this meeting. We should also thank the CEA/CEN Cadarache for its hospitality. I will refrain from making the usual comments or jokes about the weather and how good our hosts are at weather control, I shall limit myself to the brief comment that meeting in Cadarache is definitely more pleasant than in industrial environments such as Duisburg, Roubaix Tourcoing, Pittsburgh or downtown Kobe. The European Community interest in the Phebus-FP project stems from two types of considerations: one of organisational and one of technical origin. From the organisational point of view, this project is fairly unique according to Community practices. Indeed it is the only case where the European Community as such, i.e. its 12 Member States, participates technically and financially, through its own Research Centre, in a project of a significant size which has been initiated and largely implemented by one of its Member States. This reflects, I feel, an evolution in the approach followed by the European Community for the fulfilment of its scientific management. Decentralisation, closer coordination with Member States are key words in this respect. The rather centralised concept embedded in the Euratom Treaty—a Napoleonic view if I refer to the terms used by our own President Jacques Delors —should evolve into a concept where decentralisation and delocalisation have more space: if Member States offer substantial capabilities in terms of human resources and facilities which could be used for reaching Community goals, these capabilities should be fully utilised at a Community level using innovative schemes of cooperation. Important is that objectives responding to the needs of the entire Community are attained, more important than how they are implemented. In this sense, the Phebus project could be a model for other projects in the nuclear or in other fields of Community activities. Allow me, as Director General of the JRC, to express to our colleagues of the CEA my appreciation for the good collaboration established between our respective staff. What has been put in place here can be described as the first JRC outpost in a national research centre. This concept of scientific outposts is currently under discussion for a wider implementation and again Phebus can be quoted as a model case. The Phebus project, the approach which has been adopted for its implementation, responds to the needs dictated by the current situation in Europe: shrinking credits, a fact which constitutes a strong incentive for task sharing and a growing concern about building a consensus on nuclear safety issues, an element which leads to associate in nuclear safety research as many partners as possible—both those who pursue the nuclear option and those who have not opted for it or have abandoned it for the time being. This requirement for an as wide as possible consensus on safety issues leads naturally to seek partners beyond the borders of the European Community and we welcome the efforts which have been made to associate Korea, Japan, the United States and hopefully further partners, in the Phebus project. All the arguments I briefly evoked in favour of European cooperation equally apply at international level. From a scientific and technical point of view, the Phebus project fits very well to the general objectives of the European Community action in the nuclear safety field. Since the implementation of the EURATOM Treaty, the Commission has sought to promote, notably through research actions, a harmonised approach to nuclear safety in general and to nuclear reactor safety more particularly. During the last 20 years it consistently has made available substantial resources for improving our understanding of potential consequences of hypothetical accidents.

4

THE PHEBUS FISSION PRODUCT PROJECT

Because of the complexity of phenomena occurring during large hypothetical accidents, in-pile tests, reproducing realistic conditions and making use of real materials have been executed for some considerable time. I would like to mention in this respect Phebus-CSD, PBF/NRU and LOFT. In these facilities the phenomena of fuel degradation and melting were investigated under loss of coolant and special transient conditions. What remained to be more thoroughly investigated was the chain of events leading to fission product release to the outer containment during substantial melting of irradiated fuel: the so-called source term problem. In 1985, the Commission initiated an action with its Member States with a view to demonstrate the state-of-the-art in this problem area. In this context the discussion with the French authorities on our possible participation started around the end of 1985. The merits of the project itself, but also the preoccupations created by the Chernobyl accident in April 1986, created the conditions for removing obstacles to our participation in the CEA proposed project. I had the pleasure of signing, for the Commission, the agreement associating us with the CEA in July 1988. In spite of the drastic cuts in funds available for nuclear safety research in the Community framework programme 1990– 1994, we have the firm intention to pursue our collaboration. We feel that the technical goals set initially have kept their validity and that a comprehensive and difficult test programme such as Phebus-FP will contribute to the formulation, notably in Europe but also at a world level, of the best possible methods of analysis, by bringing together, around the experimental facility, analysts to develop and verify common tools. One problem that I would like to raise in closing these remarks is the issue of our credibility vis-à-vis our political masters, vis-à-vis those who have to provide the funding for such a project. With the final Phebus-FP report foreseen for 1999, we span at least the duration of 2 even 3 legislatures. Even if it has not the very long term character of the Fusion programme where politicians cannot hope to see a practical application during their own life, this project requires a standing effort over a long period of time. How can we ensure continued support? Meetings such as the seminar which begins today are a good mechanism to monitor at technical level the progress of work, to adjust programmes in order to achieve the most significant results, to verify the adequacy of the objectives with respect to the needs of the users, notably the regulatory authorities. All this is necessary and I welcome Mr. Livolant’s proposal to transform this seminar into a periodic exercise for those participating effectively in the project. But how to carry the message to the political level? How to demonstrate in simple terms that effective progress is made and that the money is well spent, that scientists and technicians are not indulging in an exercise of self-satisfaction? It is an essential problem, to which I have no immediate answer to offer. I can only tell you that we need an answer. In wishing you every success in your work, I beg you to devote some consideration to the question I just raised. Thank you for your attention.

SESSION I SURVEY OF LWR SEVERE ACCIDENT SOURCE TERM RESEARCH

LWR severe accident source term research in Europe M.Livolant , CEA/IPSN Fontenay-aux-Roses LWR severe accident source term research in the USA T.P.Speis , R.Y.Lee , L.Soffer , R.O.Meyer , NRC Washington Survey of severe accident experiments and analysis in Japan M.Akiyama , University of Tokyo K.Takumi , Nuclear Power Engineering Center, Tokyo K.Soda , JAERI Tokai-mura Summary of discussion S.Finzi , CEC Brussels

LWR SEVERE ACCIDENT SOURCE TERM RESEARCH IN EUROPE M.LIVOLANT Commissariat à l’Energie Atomique DRSN, CE/FAR, BP 6–92265 FONTENAY-AUX-ROSES CEDEX, France

1. NATIONAL SOURCE TERM POSITIONS AND PRACTICES The positions on source terms and the practices concerning emergency plans are different from one country to another in Europe. A presentation of some of those practices is useful before presenting the corresponding research work. 1.1. Switzerland Severe accident source terms are primarily used for the planning of emergency countermeasures, considering both the amount and the time scale of the postulated releases during severe core melt accidents. The time scale is especially important in Switzerland where nuclear power plants are often situated near highly populated areas. In the past, emergency planning has been based on adapted WASH 1400 source terms, ranging from PWR 2 (with iodine release reduced by a factor 10) for the fast alarm system to PWR 5 or 6 for countermeasures against ground contamination. A new reference source term for the purpose of emergency planning has been defined by the Swiss Federal Nuclear Safety Inspectorate on the basis of probabilistic studies for the two newest Swiss plants, the PWR Gösgenand and the BWR Leibstadt. The approach chosen to define the reference source term is to consider that emergency planning does not need to cover all imaginable accident situations but can be based on reasonably conservative best estimate assumptions. Moreover, it appears better to define one reference source term for all plants. So, excluding accident sequences with extremely low probabilities, and under realistic assumptions concerning accident event timing, containment break, engineered safety features and operator intervention, the following source term is used as a general basis for all Swiss nuclear power plants:

-

Start of release: Duration of release: Released radioactivity:

4 hours after shutdown 4 hours 100% noble gases (3×108 Ci) 1×106 Ci iodine (0.3%) 3×105 Ci aerosols, including 3×104 Ci Césium. 1.2. Sweden

In Sweden, the use of source terms is primarily considered as forming part of a complete safety analysis of the nuclear power plants. A variety of estimated source terms and corresponding environmental consequences may be deduced from the study of various types of accidents. Regulatory requirements for mitigative measures (guidelines given by the Government by decree of February 1986) are the following: -

Land contamination which could impede the use of large areas for a long period shall be prevented; Fatalities in acute radiation should not occur; Extremely improbable scenarios have not to be considered for meeting the requirements; The specified maximum release of radioactive substance shall apply to all reactors irrespective of site and power.

LWR SEVERE ACCIDENT SOURCE TERM RESEARCH IN EUROPE

7

Practically, the corresponding releases to fulfill the requirements on land contamination are estimated as at most of 0.1 % of the inventory of the Cesium isotopes 134 and 137. However, the off-site emergency planning applied in practice, based on source terms at higher levels defined precedently, has not changed, because most of the various aspects of emergency preparedness, like information to the public, plans for measurements, means for communication, etc, are rather insensitive to the level of source term. 1.3. France The rationale of the source term estimation in France is as follows: Three levels of source terms, named S1, S2, S3, have been defined. S1 would correspond to a containment rupture in a short time, and of fission products release without filtration, S2 to a delayed rupture with no filtration, and S3 with a delayed rupture with filtration. The order of magnitude of release is some ten percent for S1, some percent for S2 and some per thousand for S3. The accident sequences conducting to a S1 term are generally of explosive nature, and considered very improbable. All accident sequences which may terminate by a S2 level term are studied in detail and mitigation consequences are taken, including eventually some new equipment, like sand bed filters, so as to reduce the corresponding source term level to S3. So practically, S3 is the source term used for emergency planning. As already stated, it corresponds to a delayed release through a filtered channel. The main S3 characteristics are as follows: Noble gases: I131 total: Cs137:

75% 0.9% 0.35%

The main part of the release is supposed to begin approximately 20 hours after core melting. 1.4. Other European countries The other european countries have more or less similar views, but, in general, source terms for emergency planning are not stated on a national base, but estimated on a case by case evaluation of accident sequence studies made in the frame of probabilistic safety analysis. The extreme case is Italy where there are no more nuclear reactors in exploitation. For the future plants, it is foreseen to achieve a reduction of the source terms, evaluated in a realistic manner, to such values that there is no need of preplanned evacuation plan and extensive land decontamination. 1.5. General trends The general trend, specially for the future reactors, is to impose a reduction of the level of the source term considered as realistic and used in the emergency planning. One of the main reason for that is certainly in connexion with the fact that the consequences of a nuclear accident with a source term of the order of magnitude of those actually in use, like S3 in France, will not be limited to the strict health and economical direct effects, but will be largely amplified by mediatic effects. So, there is a general agreement to consider that presently used source terms have to decrease in the future. When one considers how source terms are established, it appears that there are still many incertitudes in their evaluation. In the perspective of reduced source terms, a strong research program is necessary to avoid excessive and costly conservatism. Such a program is in progress in Europe, under the financing of national organisations, with a support from the European Communities. 2. SOURCE TERM RESEARCH IN EUROPE The research work on source term in Europe is made partly in national laboratories, partly in European Research Centers (mainly Ispra). The funding is also a mixed one. The ECC plays a role in this field by its very important participation to the PHEBUS FP program, by direct research work in Ispra JRC, and by its contribution to the shared cost actions.

8

THE PHEBUS FISSION PRODUCT PROJECT

2.1. The Commission Source TERM Activities The initiation of an activity related to severe accident analysis, and in particular to the evaluation of the “source term” started in 1984. At that time, following reviews made by various institutions in the world, it was estimated that it was useful to reassess systematically the potential source terms related to severe accidents, taking into account the new information available and performing in a short time a series of well defined tests and analytical development. A Reactor Safety Shared Cost actions program was set up in 1985, focussed on Fission Products and aerosols behaviour in the reactor containment, an area where it appeared possible to make substantial progress with limited resources. Simultaneously, an Ispra team began also to work on code assessment and test analysis. At the end of 1985, the French CEA asked the Commission to participate in the funding of the in pile test program PHEBUS-FP, which was in a very preliminary stage of discussion in France. The Chernobyl accident emphasized the need for research on severe accident mitigation, and the Commission effort in the field was largely increased. In the two years 1987–1988, the JRC launched a new group of experimental and analytical SCAs in the source term area, part of them with the specific objective of preparing in collaboration with the Member States the participation of the Community to the FP project. The agreement for that participation was finally signed in July 1988, with an increase of the direct JRC effort: a team was detached to Cadarache and directly involved in the project work, the Ispra team contributing mainly to test analysis and code development validation. At the same time, the Member States sent experts to the Scientific Analysis Working Group and the Technical Group of the PHEBUS-FP program, and the ECC consultative group CGC 5 set up an adhoc working group on source term where the work performed by the Member States and the shared cost actions are periodically discussed. The main operation directly driven in the field of source term research by the JRC is the ESTER system: the main idea of such a system is to make possible the use of codes made by various organisations in the same system, in order to built in a relatively short time a code able to calculate a whole accidental sequence from modules made for partiel calculations. A data processing structure has been established, and some modules like ICARE, FIPREM, JERICHO and VICTORIA are implemented, or in course of implementation. 2.2. The French program Due to the large number of reactors in France, it was estimated that a good comprehension of the physical bases of the severe accident phenomena was necessary on a national base. So a large experimental program including analytical and in-pile experiments was defined some time ago, and is still proceeding, in parallel with the ESCADRE system for calculation of accidents. The ESCADRE System is able to calculate a complete sequence of a severe accident in a PWR up to radiological consequences. Each phenomenum is represented by models currently validated on analytical experiments carried out in France and abroad, with the needed simplification to allow sufficiently quick parametrical studies. The first part of the HEVA program has been completed. The release rates of fission products and the aerosol characteristics are determined by using pre-irradiated fuel rod sections heated in a furnace to a temperature of approximately 2000°C, in a steam jet with or without hydrogen. The additional program SOPHIE allows to examine the kinetics of the deposit and revaporation of certain selected fission products, like Cs I, Cs OH, I2 Te. It is planned to extend the HEVA program to the examination of fission product release at higher temperature (>2300°C) in a new loop called VERCORS. Experiments carried out as a part of the PITEAS aerosol physics program consist in injecting dry aerosols in a 3 m3 tank containing air and steam with variable saturation rates and monitoring the changes in them. The TUBA loop, designed to examine the retention phenomena of aerosols in small pipes representing steam generator tubes, began operation in 1988. An extension of that type of study to large pipes is being studied now (TUBA GROS TUYAU, i.e. TUBA “BIG PIPE”). Naturally, the top of the program is the PHEBUS-FP program, which will simulate a complex set of phenomena, from fission products emission to deposition in containment. 2.3. UK research program An extended research is made in UK on the nature and behaviour of fission products in the primary circuit and the containment, with a particular attention to chemical effects. Experiments are conducted in the FALCON facility, to study the interaction of real or simulated fission products emitted from fuel samples heated up to 2500°C with aerosols. The transport of the released material is followed through a complex pathway simulating the reactor core, the upper plenum, hot leg

LWR SEVERE ACCIDENT SOURCE TERM RESEARCH IN EUROPE

9

structures and the containment. An important modelling work on vapour-aerosol interaction and iodine chemistry is in progress and the corresponding models are incorporated in the INSPECT and VICTORIA codes. Another topic well studied in UK concerns pool scrubbing, for which the BUSCA code is developped. 2.4. Other European countries Some other experimental and theoretical studies made in European countries are relevant of the subject. In Germany, the VANAM tests are run in the Battelle model containment, with a multicompartiment geometry roughly modelling PWR situation. Aerosols are injected at a high level together with steam and conditions of natural circulation and stratification are established. The aerosols distribution and decay are monitored. The results are used to validate coupled thermohydraulic and aerosols calculation codes, like FIPLOC. A research program is in preparation on pool scrubbing with the possibility of tests in Spain and in Italy and the improvement and validation of calculation codes, like the UK BUSCA code, already mentioned. 3. CONCLUSION Such a review of research program on the source term in Europe may give the impression of some dispersion of efforts in various directions. In fact, for each country, the coherence of the program is estimated on a national base, but, the large extend of exchanges under bilateral agreements or under the auspices of organisations like ECC or OECD allows some specialization for the research teams, at least for the large experiments which are expensive. It is one of the interests of a program like PHEBUS-FP to give opportunity for the best experts in the field to meet and work together, which is probably the most effective way for international collaboration.

LWR SEVERE ACCIDENT SOURCE TERM RESEARCH IN THE USA T.P.Speis, R.Y.Lee, L.Soffer, R.O.Meyer U.S. Nuclear Regulatory Commission Office of Nuclear Regulatory Research Washington, DC 20555

SUMMARY Fission product releases to the environment, or source terms, arise as a result of a highly diverse group of phenomena involved in any particular severe accident sequence. For light water reactors (LWRs) these include core heatup, fuel element degradation and melting, pressure vessel attack and failure, possible high pressure melt ejection, interaction of core debris with concrete, retention of fission products within the reactor coolant system, effects of hydrogen burns or detonations, retention of fission products by suppression pools or ice beds, late revolatilization of fission products from surfaces, and, clearly, the effect of containment integrity or containment bypass and time and location of containment failure, if it occurs. Because of the multiplicity of accident sequences that can occur for a given plant as well as the diversity of the, as yet, imperfectly understood severe accident phenomena, it is not surprising that PRAs such as, for example, those documented in NUREG-1150 have indicated large uncertainties in source terms which represent a significant contribution to the uncertainty in the absolute value of risk. Because of the difficulty and expense involved in performing prototypic experiments, substantial reliance has been placed on the development and validation of detailed mechanistic computer codes for analyzing severe accident phenomena and the source terms associated with them. This paper discusses the extensive research and other efforts that have taken place over the last decade to address the technical issues which have a bearing on being able to describe quantitatively the source term(s) and its characteristics. It also summarizes our present state of knowledge and points out areas where additional research will add further to our understanding. In this context the paper discusses the information that could be provided by the PHEBUS-FP program and its use to assess severe accident integral evaluation codes such as VICTORIA and CONTAIN. Finally, this paper discusses the NRC’s efforts to revise the licensing source term (TID-14844) and the implications of this revision, especially for siting and design of future power plants. 1. INTRODUCTION AND BACKGROUND Radionuclide releases to the environment, that is, the type, quantity, timing and energy characteristics of the release of radioactive material from reactor accidents (“source terms”) are deeply embedded in the regulatory policy and practices of the U.S. Nuclear Regulatory Commission (NRC). For almost thirty years the NRC’s reactor site criteria (10 CFR 100) have required for licensing purposes that an accidental fission product release from the core into the containment be postulated to occur and that its radiological consequences be evaluated assuming that the containment remains intact but leaks at its maximum allowable leak rate. Evaluation of the consequences is used to assess both plant mitigation features such as fission product cleanup systems as well as the suitability of the site. The characteristics of the “source term” into the containment, which must be distinguished from a release to the environment, is contained in Regulatory Guides 1.3 and 1.4, but is derived from the 1962 report TID-14844 (Ref. 1), and consists of 100% of the core inventory of noble gases and 50% of the iodines (half of which are assumed to deposit on interior surfaces very rapidly). Regulatory Guides 1.3 and 1.4 also specify that these are instantaneously available for release, which has significantly affected containment isolation valve closure times; and also specify that the iodine is predominantly (91 percent) in elemental (I2) form. The regulatory applications of this release cover a wide range in addition to plant mitigation features and site suitability and include the basis for (1) the post-accident radiation environment for which safety-related equipment should be qualified, (2) post-accident habitability requirements for the control room, and (3) post-accident sampling systems and accessibility.

LWR SEVERE ACCIDENT SOURCE TERM RESEARCH IN THE USA

11

In contrast to a specified source term for design basis accidents, severe accident source terms first arose in probabilistic risk assessments (e.g., Reactor Safety Study, WASH-1400) in examining accident sequences which involved core melt and where containments could fail. Severe accident source terms represent mechanistically determined “best estimate” releases to the environment, including estimates of failures of containment integrity. This is very different from the combination of the nonmechanistic conservative release to containment postulated by TID-14844 coupled with the assumption of very limited containment leakage used for Part 100 siting calculations for design basis accidents. The worst severe accident source terms resulting from containment failure (especially early failures, i.e., within a few hours from onset of an accident) or containment bypass can lead to consequences that are much greater than those associated with a TID-14844 release into containment and where the containment is assumed to be leaking at its maximum leak rate for its design conditions. Indeed, some of the most severe source terms arise from some containment bypass events, such as “event V” and multiple steam generator tube ruptures. Source term estimates under severe accident conditions began to be of great interest shortly after the Three Mile Island (TMI) accident. A major NRC research effort began about 1981 and has been under way since then to obtain a better understanding of fission-product transport and release mechanisms in LWR’s under severe accident conditions. This research effort has included a very large and extensive staff and contractor effort, involving a number of national laboratories as well as nuclear industry groups, and has resulted in the development and application of several new computer codes to examine core-melt phenomena and associated source-terms involved in severe accident sequences. Work by the NRC staff has also included significant review efforts by peer reviewers, foreign partners in NRC research programs, industry groups, and the general public. Current risk assessment methods, including the latest research efforts on severe accident source terms, have been reflected in the issuance of NUREG-1150 (Ref. 2) which provides an assessment for five US nuclear power plants. Finally, the occurrence of the accident at Unit number 4 of the Chernobyl reactor in the Soviet Union on April 26, 1986 and the large accidental release of fission products resulting from it has provided further impetus to understand severe accident source terms as well as to prevent such occurrences. This paper discusses the major developments that have taken place in our understanding of this “source term” technology, starting with the simplified assumptions of WASH-1400 to the present use of detailed mechanistic codes such as VICTORIA and CONTAIN. The paper also summarizes some of the areas where additional research will add further to our understanding, as well as how programs such as the Phebus-FP can contribute to this understanding. 2. SEVERE ACCIDENT SOURCE TERM RESEARCH IN THE USA Scope of Research Efforts Beginning shortly after the accident at Three Mile Island, the NRC has sponsored numerous experimental and analytical research projects on fission product release and transport. Table 1 list the major NRC research projects conducted and makes reference to major publications resulting from that work. Table 1: Major severe accident fission product research projects sponsered by NRC Project Description

Laboratory

Reference

Out-of-Pile Release Measurements Post-Accident Chemistry Containment Aerosol Behavior TRAP/MELT Code Validation Aerosol Models for VICTORIA High Temperature Experiments VANESA Code Development Core-Concrete Aerosol Experiments ACRR Source Term Experiments VICTORIA Code Development FASTGRASS Code Development ACE Support PBF Fission Product Tests & Analysis PHEBUS Support

ORNL ORNL ORNL ORNL ORNL SNL SNL SNL SNL SNL ANL ANL INEL INEL

3, 4 5, 6 7–9 10–12 13 14–16 17 18 19 20 21 22, 23 24, 25 26

12

THE PHEBUS FISSION PRODUCT PROJECT

Project Description

Laboratory

Reference

SPARC & ICEDF Code Development Source Term Reassessment Activity Coefficient Measurements LACE Support

PNL BCL BCL HEDL

27, 28 29–31 32 33

Early experiments and analytical work tended to focus on release from fuel material under high temperatures and severe accident environments. Later, aerosol deposition and transport modeling was done for behavior in the reactor coolant system and in the containment. Currently, fully integrated models are in the process of being completed for release, transport, condensation of vapors, agglomeration and settling of aerosols, and chemical reactions in the reactor coolant system (VICTORIA) and in the containment (CONTAIN with the TRENDS models). General Overview of the Phenomenology and State of Knowledge on Source Terms In-vessel source term: release from fuel and retention in RCS The Reactor Safety Study, WASH-1400 (Ref. 34) analyzed two specific reactors: Surry, a three-loop PWR with a large dry containment and Peach Bottom, a BWR with Mark I containment. For each, calculations were performed for a number of accident sequences and the results used to define a series of release categories. WASH-1400 assumed that most of the release of radionuclides from the reactor fuel occurred as it melted. Once radionuclides escaped the fuel, they were assumed to pass out of the reactor coolant system (RCS) with no attenuation. Neglect of radionuclide deposition or retention in the RCS was recognized to be unrealistic but was considered conservative. A review of the state-of-art for calculating fission product release and transport were the objectives of a major NRC initiative following the TMI-2 accident [NUREG-0772, NUREG-0773, BMI-2104]. The NRC’s Source Term Code Package (STCP), Fig 1, emerged as an integral tool for analysis of fission product transport in the RCS and containment. STCP models release from the fuel with CORSOR (Ref. 35) and fission product retention and transport in the RCS with TRAPMELT (Ref. 36). For the ex-vessel source term, the release from core-concrete is modeled by VANESA (Ref. 17). Depending on the containment type, NAUA, SPARC or ICEDF (Ref. 28) are used to model the transport and retention of fission product release

LWR SEVERE ACCIDENT SOURCE TERM RESEARCH IN THE USA

13

Fig. 1: STCP

from the RCS and from core-concrete interaction into the containment, and subsequent release of fission products to the environment consistent with the state of the containment. CORSOR is a simple temperature correlation (Arrhenius form) of the results of out-of-pile release experiments. Implicit assumptions are made about the chemical form of the 18 species treated, but they are invariant and cannot be adjusted to the changing accident conditions (different ratio of steam and hydrogen environment). Barium, for example, can exist either as an oxide or as a metal in the fuel debris. At the same temperature, a choice of one vs. the other, depending upon the oxidizing state of the environment, can mean a difference in the predicted release rate using CORSOR of about a factor of 500. Furthermore, no provision is made in CORSOR for predicting releases during melting or eutectic formation, from rubble beds or molten pools, or under extremely oxidizing conditions accompanying air ingress after vessel breach. Table 2 shows some STCP results for the fractions of initial core inventory released to the reactor vessel prior to pressure vessel failure for a PWR and BWR and for both high and low pressure sequences. Deficiencies in CORSOR were identified in NUREG-0956, and improved modelling of fission product release from fuel is being developed and implemented in the VICTORIA code (Ref. 20), which is NRC’s most current and sophisticated method for in-vessel release and transport. Particular attention has been devoted in the STCP to radionuclide retention in the RCS. This is an area where significant additional model development has taken place since WASH-1400. The expertise developed for aerosol transport in fast reactor safety programs was applied to the behavior of radionuclides in light water reactor coolant systems. This was done via the TRAPMELT code, which calculates aerosol and vapor transport within the RCS. Table 2: STCP results for fraction of initial core inventory released to vessel prior to RTV failure Radionuclide Group Surry (high pressure) TMLB’ Surry (low pressure) V Peach Bottom (high pressure) Peach Bottom (low pressure) TC2 TC1 NG I Cs Te Sr Ba Ru

0.98 0.98 0.98 0.46 7×10−4 0.013 10–6

1.0 1.0 1.0 0.63 1.5×10−3 0.03 3×10−6

0.87 0.87 0.87 0.62 5×10−4 0.01 10−6

0.92 0.92 0.91 0.3 6×10−4 0.01 8×10−7

14

THE PHEBUS FISSION PRODUCT PROJECT

Radionuclide Group Surry (high pressure) TMLB’ Surry (low pressure) V Peach Bottom (high pressure) Peach Bottom (low pressure) TC2 TC1 Ce La

0 10−7

0 2×10−7

0 10−7

0 3×10−8

Using TRAPMELT, it was found that there could be significant retention of released radionuclides in the RCS for many, but not all important sequences. For some accidents analyzed, TRAPMELT calculated that less than 20% of the radioactive material released from the degraded reactor core emerged from the RCS. A convenient way to describe the overall effect of retention in the RCS is to indicate the fraction of materials released from the fuel which is released from the vessel. STCP results for SURRY and Peach Bottom are shown in Table 3. A comparison of these values indicates that retention in the RCS is primarily a function of the RCS pressure. Low pressure sequences are characterized by rapid blowdown of the RCS with little gravitational settling, the dominant mechanism for aerosol deposition. On the other hand, for high pressure accident sequences, fission products released from the fuel are retained in the RCS with high efficiency (except for noble gases). While the PWR results show a fairly regular trend toward increasing release fraction with decreasing RCS pressure, trends among the BWR data are less clear. Reduction in RCS retention for volatile materials in BWR accident sequences illustrates the effect of revaporization because of fission product heating of the structures where fission products had originally deposited. Table 3: STCP results for fraction of initial core inventory released from vessel into cotainment Radionuclide Group Surry (high pressure) TMLB’ Surry (low pressure) V Peach Bottom (high pressure) TC2

Peach Bottom (low pressure) TC1

NG I Cs Te Sr Ba Ru Ce La

0.98 0.81 0.81 0.13 0.8 0.8 0.8 0 0.76

1 0.22 0.21 0.62 0.26 0.26 0.26 0 0.3

N/A N/A N/A N/A N/A N/A N/A N/A N/A

1 0.9 0.8 0.15 0.62 0.6 0.6 0 0.78

Since the fractional releases tabulated in Tables 2 and 3 are correlated in a phenomenological sense, it is more reasonable to present the results in terms of the fraction of initial core inventory released from the vessel into the containment at, or before, vessel failure. This is shown in Table 4. (Individual values in Table 4 may not precisely equal the product of Tables 2 and 3, since these represent mean values of distributions.) The estimated fractional releases depend strongly on the volatility of the fission products, as might be expected. Volatile fission products Iodine and Cesium have similar releases. The difference between semivolatile fission products Sr and Ba are not great. Low volatile fission products Ce and La show similar behavior. For bypass sequences due to the multiple steam generator tube ruptures, STCP predicts very little retention in the RCS or in the secondary side of the steam generator system. Because these predicted releases are high, bypass sequences at Surry and Sequoyah dominated the risk. The radionuclide release for the Surry bypass sequence is shown in Fig. 2. The figure shows that the uncertainty ranges for the source terms for these sequences are large. Table 4: STCP results for fraction of initial core inventory released from the vessel into containment (puff release) Radionuclide Group Surry (high pressure) TMLB’ Surry (low pressure) V Peach Bottom (high pressure) Peach Bottom (low pressure) TC2 TC1 NG I Cs Te Sr Ba Ru

0.98 0.22 0.21 0.28 2×10−4 3×10–3 3×10–7

N/A N/A N/A N/A N/A N/A N/A

0.96 0.92 0.79 0.2 4×10−4 7×10−3 7×10−7

0.9 0.75 0.74 0.04 5×10−4 9×10−3 7×10−7

LWR SEVERE ACCIDENT SOURCE TERM RESEARCH IN THE USA

15

Radionuclide Group Surry (high pressure) TMLB’ Surry (low pressure) V Peach Bottom (high pressure) Peach Bottom (low pressure) TC2 TC1 Ce La

0 44×10−8

N/A N/A

0 6×10−8

0 7×10–8

In the STCP analyses, CORSOR assumes iodine and cesium to be in the form of CsI and CsOH, and Te to be in elemental form. These species are transported from the core as vapor. In the RCS, fission products may condense as aerosols or react with surfaces. However, TRAPMELT does not account for chemical reactions of CsI. Several processes are known to alter the chemical form of iodine (e.g. reactions of CsI with borates, metal surfaces). TRAPMELT is basically an aerosol code that treats inert particles. In a rudimentary way, however, TRAPMELT treats some chemical reactions using empirical deposition velocities, but deposition is irreversible. True chemical reactions are not modeled, and revaporization cannot occur. Chemical reactions between aerosols and vapors are not modeled either in TRAPMELT, nor is reentrainment calculated. These processes though are being modeled in VICTORIA. In-vessel source term: revaporization from reactor coolant system Deposition of radioactive material in the RCS has focused on the stability of the deposited radioactive material, particularly whether continued decay heating induces substantial revaporization. As long as the reactor core has not penetrated the reactor vessel, TRAPMELT considers revaporization of the condensed radionuclides, and assumes iodine and cesium to be in especially volatile forms. It probably over-predicts revaporization of deposited materials because the code has no ability to predict evolution of deposited materials to less volatile chemical forms (i.e. does not take into account the change in the chemical form of deposited CsOH). It underpredicts the revaporization of Ba, Ru and Te simply because TRAPMELT contains no chemistry to allow these species to become vapors. The issue has become of particular concern as models of core degradation have evolved to predict that, for some sequences, natural circulation of gases through the reactor core may also heat structures in the RCS to substantially higher temperatures than had been previously predicted. Radioactive material deposited on surfaces within the PWR RCS and BWR reactor vessel can also be reevolved after vessel failure because of self-heating. In NUREG-1150, two cases were considered for the PWRs: one vs. two holes in the RCS (i.e., opening in the vessel due to melt penetration of vessel bottom and/or failure of RCS piping at certain location due to high temperature natural circulation). The latter case offers the opportunity of a “chimney effect” and a greatly different environment. The “chimney effect” provides a natural circulation of air through the RCS following failure of the RPV. The circulation is driven because gases in the RCS are heated by deposited radionuclides and retained fuel. The effect is of some significance since current analyses of core degradation indicate substantial fractions of the core could be retained within the RPV after vessel failure. Evidence from TMI-2 suggests that as much as half of the core material may have stayed within the original confines after the rest of the core had melted and drained into the lower plenum. This remaining fuel in the core region could be exposed to air once the plenum has been breached. The analysis considered the balance of buoyancy forces and pressure drop to determine heat transfer from radionuclides and core debris to the gas, and the chemical thermodynamics of revaporization. The results are most strongly affected by the thermo-chemistry of the deposited radionuclides and the geometry of breaches in the RCS. Releases for three elemental groups: iodine, cesium, and tellurium were considered. The results show that the fractional release is greatest for iodine and least for tellurium. Fig. 3 illustrates the distribution for the release of iodine for the PWR case with two holes in the RCS. The range for this case, which produces the greatest release, is from 0 to 70 percent, with a median release of 20 percent. Cesium release fractions were comparable to the iodine values, but slightly less. The median release of tellurium was 0 percent for all cases, but the upper bound varied from 20 to 60 percent. This skewed distribution is indicative of a general belief that there will be little or no revaporization of tellurium, but it recognizes that substantial revaporization cannot be ruled out. Recognizing that TRAPMELT is inadequate to address revaporization from the RCS, the VICTORIA code is incorporating such a model. Ex-vessel source term In low pressure accident scenarios where the reactor vessel fails, high-temperature core debris may fall into the reactor cavity where it interacts with structural concrete. At high temperatures (approximately 1,300–1,500°C), concrete decomposes, and the ablation products commonly include water vapor and carbon dioxide as well as the refractory oxides CaO and SiO2. The liquefied oxidic components of the concrete mix with the uranium oxide fuel and metallic oxides of the debris. Typically, the core debris is initially all or partially molten; gases released at the debris-concrete interface bubble through the debris pool reducing some low-vapor-pressure oxides like La2O3 to high-vapor-pressure forms like LaO. These more volatile forms then vaporize into the bubble volume thus releasing fission product species that were not released in the vessel. Aerosols are

16

THE PHEBUS FISSION PRODUCT PROJECT

Fig. 2: Release fraction for containment bypass at Surry

Fig. 3: Revaporization release fraction for iodine, PWR case with two openings.

formed when the bubbles exit the upper surface and fragment. If an overlying water pool exists, a considerable amount of the aerosols may be scrubbed and kept out of the containment atmosphere. STCP uses CORCON-MOD2 (Ref. 37) for modelling core-concrete interactions and VANESA for radionuclide releases driven by bubbling of reaction gases into the melt. VANESA calculates the releases by vaporization of fission products and other melt constituents from the melt into the gas bubbles. Among the factors that influence the magnitude of the ex-vessel releases are the composition and temperatures of the core debris. Concrete composition also has a major impact on the amount of aerosols entrained into the containment atmosphere. Limestone concrete produces larger gas flows and is more oxidizing compared to basaltic. Among the five plants analyzed in NUREG-1150, only Surry has basaltic concrete. CORCON-MOD3 is the latest computer code for predicting core-concrete interactions. It combines CORCON-MOD2 and VANESA together into a single code. For core-concrete interactions, the code predicts heat transfer to the containment, noncondensible and combustible gas generation, and radionuclide release and aerosol generation. CORCON-MOD3 is near completion and it is expected to be adequate for analysis of ex-vessel source terms.

LWR SEVERE ACCIDENT SOURCE TERM RESEARCH IN THE USA

17

Release from containment Ultimately the amount of fission products released to the environment depends on the containment’s ability to withstand the various challenges which result from the evolution of a severe accident and the accompanying thermal and mechanical loads. The ability then of the containment to maintain its integrity is determined mainly by two factors: (i) the magnitude of these thermal and mechanical loads, and (ii) the response of the containment structure to those loads. From a risk perspective, containment is considered to have failed to perform its function when the leak rate of radionuclides to the environment becomes substantial. Failure can occur as the result of a structural failure of the containment, tearing of the containment liner, or a high rate of leakage through a penetration. In some accidents, loss of the containment function is independent of these loads. For example, in interfacing system LOCAs, the containment is effectively bypassed. In these sequences, check valves isolating low-pressure piping fail and the piping connected to the RCS fails outside containment. Radionuclides can escape to secondary buildings through the RCS piping without passing through the containment. For most severe accidents, if the containment function is maintained, the radiological consequences will be small. If the containment does fail, the timing of failure can be very important. The longer the containment remains intact relative to the time of core melting and fission product release from the RCS, the more time is available for radioactive material to be removed from the containment atmosphere by engineered safety features and natural deposition processes. A delay in containment failure also provides time for protective actions to be taken. Thus, in evaluating containment performance, it is convenient to designate no failure, late failure, bypass, and early failure of containment as separate categories characterizing different degrees of severity. Plants which have the option to vent the containment are represented by a separate category. In NUREG-1150, containment performance was analyzed with respect to the timing of containment failure and the magnitude of leakage to the environment. However, radionuclide release to the environment is also affected by the performance of engineered safety features. Engineered safety features typically employed in PWRs are sprays, fan coolers, and ice condensers; and in BWR’s, filters and suppression pools. Flooding of reactor cavities or pedestals may also be employed. Suppression pools are effective in the removal of radionuclides in the form of aerosols or soluble vapors. Some of the most important radionuclides, such as iodine, cesium, and perhaps tellurium, are largely released during the in-vessel release period, and directed to the suppression pool where they are subjected to scrubbing, even if containment failure has already occurred. For the Peach Bottom plant, decontamination factors ranging from 1.2 to 4,000 with a median value of 80 were calculated in the study. Depending on the timing and location of containment failure, the suppression pool may also be effective in scrubbing core-concrete releases after vessel failure. Although decontamination factors for the suppression pool are large, iodine captured in the pool will not necessarily remain there. The re-evolution of iodine was important in accident scenarios in which the containment has failed and the suppression pool is boiling. In a containment with an ice condenser, borated ice beds remove fission products from the air by processes similar to the BWR pressure suppression pools. The decontamination factor is very sensitive to the volume fraction of steam in the flowing gas, which in turn depends on whether the air-return fans are operational. With the air-return fans on, decontamination factors range from 1.2 to 20, with a median value of 3. Containment sprays are also effective in reducing airborne concentrations of fission product aerosols and vapors. In the Surry (sub-atmosphere) and Zion (large dry) designs, approximately 20 percent of core meltdown sequences were predicted to eventually result in delayed containment failure or basemat meltthrough. The effect of sprays, in those scenarios in which they are operational for an extended time, is to reduce the concentration of particulate radionuclides airborne in the containment to negligible levels in comparison to the noble gases. For shorter periods of operation sprays still have a substantial mitigative effect on the releases. The likelihood and amount of water accumulation below the reactor vessel is determined by the configuration of the reactor cavity or pedestal regions. For the Surry plant, if borated spray is not operating, the cavity will be dry at vessel failure. For Peach Bottom, there is a maximum of approximately 2 feet of water available on the pedestal and drywell floor because of the configuration of the downcomer. If a coolable debris bed is formed in the cavity or pedestal and makeup water is continuously supplied, core-concrete release fission products would be avoided. Even if molten core-concrete interactions occur, an overlying pool of water can reduce the release of radioactive material to the containment by scrubbing. Other more dynamic processes such as steam explosions, which under some circumstances can take place when molten core debris comes into contact with water will cause debris fragmentation which results in additional aerosol formation. Depending on the aerosol size, it may or may not transport far from the source of generation. An example of source terms (fractions of the core inventory of groups of radionuclides released to the environment) from NUREG-1150 for the Surry plant is shown in Table 5. Groups of release fractions are shown vs. the mean exceedance frequency. The results in this table may be contrasted with those of WASH-1400, which indicated that 70 percent of the core inventory of iodine and cesium were predicted to be released with a probability of about 10−5 per reactor-year (for release

18

THE PHEBUS FISSION PRODUCT PROJECT

category PWR-2). NUREG-1150, in contrast, indicates that releases of this magnitude would have a probability of occurrence almost two orders of magnitude lower. Some of this difference is attributed to improved estimates of containment performance since WASH-1400. Table 5: Exceedance frequencies for release fractions for Surry: all internal initators (mean values) Exceedance Freq. (per reactor year)

Release Fractions

NG

I

Cs

Te

Sr

Ba

Ru

La

Ce

10−5

7.5×10−3

3×10−4

10−8

<10−8

<10−8

<10−8

<10−8

0.015 0.2 0.35

2×10−3

<10−8 8×10−4 0.015 0.04

10−6 10−7 10–8

1.0 1.0 1.0

0.3 0.75 1.0

0.3 0.75 1.0

0.15 0.4 0.65

0.015 0.1 0.4

0.03 0.08

<10−8 10–3 0.1 0.2

Source Term Uncertainties and Present Research Efforts Since the TMI-2 accident, a great deal of progress has been made in understanding severe accident phenomena and reducing uncertainties in risk assessments. NUREG-1150 has identified specific source term issues as contributors to uncertainty in risk estimates. This information is useful in providing guidance for future research and will be discussed briefly. In-vessel source term: release from fuel and retention in RCS For fission product release, key questions that have been raised are: (a) How significant is fission product release during the late stages of core melt vs. the early phase? and (b) Are releases of refractory nuclides enhanced by vaporization of UO2 at high temperature? The present theoretical model for the late-stage rubble bed (significant relocation and melting of ceramics) assumes that release is governed by gas phase mass transport. For the molten pool, the main mechanism for fission product release is governed by diffusion and surface convection. Both of these theoretical models need experimental data for validation. The predominant sources of uncertainty in these theoretical model are: - geometry of the core debris, - magnitude of gas fluxes through the debris, - thermo-chemical properties of the high-temperature vapor species that vaporize from the debris. The VICTORIA code is incorporating such models. The uncertainties are handled in a parametric fashion reflecting current understanding of the evolution of accident scenarios. Out-of-pile testing at Battelle for small fuel samples at high temperatures has shown strong evidence of congruent vaporization (matrix stripping). Similar behavior was observed in experiments conducted in the UK and at ORNL. Tests with small fuel samples tend to release relatively large amounts of the less-volatile elements (Ru, Zr, Ce, and U). Congruent vaporization is generally well understood and occurs when vaporization is rapid, which is the case for small fuel samples. Because of the large size of a reactor core, vaporization will occur at near equilibrium partial pressure conditions and should be slow. The total vaporization of U species depends on the gas flow rate and the total amount of U. The less-volatile fission products released will be small even if congruent vaporization were to occur. Currently, this phenomenon is not being modeled in any computer codes. Integral experiments to study fission product release from late phase core melt progression are not being planned in the U.S. Hence, PHEBUS-FP could provide some of the needed experimental data, especially if some of the tests were to be performed under more severe accident conditions. In-vessel source term: revaporization from reactor coolant system For late-phase revaporization of fission product from the RCS, key questions are (i) After the reactor vessel is breached and air ingress occurs, what are the release rates from fuel remaining in the vessel?, and (ii) What chemical forms are important during the transport and retention of aerosols and vapors? As mentioned earlier, if the vessel is penetrated, air from the containment atmosphere will circulate over retained fuel. The air will react exothermically with the cladding remaining on the fuel, producing high temperatures in this fuel. Once the cladding has been oxidized, vapors of radionuclides not usually considered highly volatile—notably ruthenium and

LWR SEVERE ACCIDENT SOURCE TERM RESEARCH IN THE USA

19

molybdenum will be produced due to the strongly oxidized conditions. Tests performed at Chalk River Laboratory in Canada with uncladded and cladded fuel, and at ORNL with fuel fragments under highly oxidized conditions, showed that a large fraction of the Ru, Mo, and Te was released. The predicted release rates are dependent on a highly oxidized condition for the fuel. However, it is necessary to conduct tests for radionuclide releases for typical LWR fuel (as opposed to the thinner cladding used in the CANDU fuel). The results could then be incorporated into the current VICTORIA fuel release model. The chemical form assumed by iodine affects late-phase revaporization. If cesium iodide or other iodides decompose to produce HI or I2, there will be little or no retained iodine in the RCS to revaporize after containment failure. If CsI is stable, substantial amounts will be retained temporarily in the RCS and will be able to revaporize creating an iodine source term after containment failure. Tests at Winfrith and SNL have shown that CsI will react with boron oxides vaporized into the RCS atmosphere to yield cesium borate, I, and HI. Tests at SNL have shown thermal instability of CsI in the RCS environment. What has not been clarified in these tests is whether or not I or HI produced by reactions of CsI can subsequently react to form other iodides such as NiI2(g,c). Analyses done in the UK suggest that high vapor fractions of iodine (CsOH and CsI) at the time of RCS failure could yield aerosols in containment that do not settle rapidly and are affected little by containment sprays. Currently, there are no suitable experimental data to validate revaporization models. The PHEBUS-FP tests could provide some. It is likely that continued examination of the chemical fate of I or HI produced by reaction of CsI could alter the perceptions of risk by altering the predicted amount of suspended radioactivity in containment at the time of containment failure. This could be accomplished by sensitivity analysis using VICTORIA. It is also important to utilize risk perspectives regarding the uncertainties in late stage core-melt and late phase revaporization. Generally, risk is increased by the release of fission products into containment early and for early containment failure or bypass. Hence, additional fission products released later in an accident phase will denote lesser releases at an earlier stage. Modelling sensitivity can be made in risk assessments that can test these uncertainties and their implications. Ex-vessel source term Key questions regarding the ex-vessel source term are (i) What effect does hydrogen combustion have on aerosols suspended in the containment atmosphere?, (ii) For previously deposited fission products, what are the mechanisms and rates of revolatilization of iodine from water pools? Aerosol materials containing CsI must be dehydrated and vaporized before chemical interactions of cesium iodide can be expected. Tests conducted at ORNL, as part of the EPRI-ACE program, found that vaporized CsI was unstable in hydrogen flames with the iodine redistributing as iodide, I2, and iodate. An excess of metal cations (Cs) reduced the extent of I2 formation. Oxidation to I2 was consistent with thermal decomposition of CsI, but iodate production was related to the nonequilibrium OH and O radical concentrations found in hydrogen/air flames. Data are believed to be adequate to address this question. However, the presence of other aerosol species in the containment atmosphere such as aerosols produced by coreconcrete interactions could affect the stability of CsI during hydrogen combustion events. For instance, silica could trap cesium to form cesium silicate so that it cannot recombine with iodine. Other basic species could react with iodine produced in the combustion to reform iodides:

Currently, the TRENDS model (Ref. 38) treats this in a conservative fashion, i.e., it assumes that I2 will be formed in a hydrogen burn. No specific model has been developed to take this phenomenon into account in a mechanistic fashion. The experimental data on the effects of hydrogen burns on fission products will be used to develop a model for inclusion in the CONTAIN code. With respect to the potential release of iodine from suppression pools and reactor cavity water, the EPRI-ACE program and the ORNL iodine chemistry research has provided data that addressed this issue. At ORNL, the research included iodine partition coefficient tests, hydrolysis chemical kinetics tests, radiolysis chemical kinetics tests, hydrogen burn/iodine chemistry tests, and TRENDS models development. These efforts were further enhanced by the EPRI/ACE program, which included hygroscopic aerosol/iodine chemistry tests, and hydrogen burn/iodine chemistry tests. NRC is in the process of incorporating the TRENDS models into the CONTAIN code. The validation of the CONTAIN code would utilize integral experimental data from the PHEBUS-FP program. In conclusion, although additional research to reduce uncertainties in source term phenomena with respect to physics and chemistry can be done, it is again important to consider these from a risk perspective. NUREG-1150 indicates that uncertainties in overall estimation of risk are largely driven by uncertainties in containment performance, primarily those associated with estimation of containment loads, estimation of containment performance at load levels beyond the design basis, and estimation of the probability and location of containment bypass.

20

THE PHEBUS FISSION PRODUCT PROJECT

3. POSSIBLE REGULATORY APPLICATIONS/IMPLICATIONS Development of Updated Source Term Design basis accident source terms have been used in the USA for licensing purposes in three distinct ways, namely: (1) for siting evaluations as required by 10 CFR Part 100, (2) to define the radiological environment conditions for certain plant systems, and (3) to assess the effectiveness of plant mitigation systems. The NRC is presently preparing a technical update of the source term contained in Regulatory Guides 1.3 and 1.4 making use of current severe accident research insights. This effort is expected to be reflected in changes in the timing of the release, the composition and magnitude of fission product releases into the containment, and the chemical form of the iodine fission products. A draft of an updated report replacing TID-14844 is expected to be issued for comment by January 1992. Rather than an instantaneous release into the containment, as presently given, the revised formulation is expected to be stated as a series of fission product releases into containment, each one associated with a particular stage of an accident or group of accidents. Hence, the revision is expected to begin with the release of coolant activity, followed by release of activity in the fuel gap, the release of fission products associated with gross fuel degradation prior to reactor vessel failure, and finally, release of fission products from core-concrete interactions. Additional nuclides other than the noble gases and iodine are expected to be given consideration. For example, preliminary indications are that the fraction of core inventory of Cesium released into the containment is generally comparable to that of Iodine. In addition, some Tellurium and smaller fractions of the remaining nuclides are also expected. A recent study on iodine chemical form and behavior entering the containment from the RCS, and the subsequent revolatilization of iodine from water pools in containment has been completed at ORNL. ORNL examined a group of severe accident sequences used in NUREG-1150. These accident scenarios were for both high and low pressure sequences that are risk significant. For the RCS, the analysis considered the chemical kinetics of 20 reactions of iodine with water, hydrogen and cesium, and determined the temperature and time when chemical equilibrium was established. Once chemical equilibrium was established, an analysis determined the iodine chemical forms present. In most calculations, iodine was released from the RCS into the containment as Cesium Iodide (CsI) with very small amounts of I or HI. The ORNL results indicate that the iodine entering containment is at least 95% CsI, 5% as I and HI with not less than 1% as either I or HI. This is in contrast to the iodine chemical form specified by the TID source term, which is predominantly (91%) elemental. The ORNL iodine research and the EPRI-ACE program has provided data that address revolatilization of iodine from water pools. A comprehensive model to estimate the revolatilization of iodine from water pools was developed by ORNL. Once iodine enters containment, it dissolves in water pools or plates out on wet surfaces as I−. Subsequently, the iodine behavior within the containment depends upon time and pH of the water solutions. If the pH is maintained at a value of seven or greater, then the amount of iodine in solution which converts to I2 and organic iodine later in the accident sequence will be very low. If pH is not controlled, radiation levels in water pools are sufficient to convert much of the dissolved iodine to elemental iodine for release into the containment atmosphere. Regulatory Implications At the present time, the NRC is pursuing several regulatory initiatives to incorporate insights from updated severe accident source terms. A revision of the NRC’s reactor site criteria (10 CFR 100) is being carried out in parallel with an interim revision of 10 CFR 50. The reactor site criteria will be revised to remove source term and dose calculations and to add exclusion area size and population density requirements based upon those from Regulatory Guide 4.7 directly into Part 100. At the same time, Appendix A to Part 100, containing site seismic criteria, is also being revised to reflect the latest understanding. Source term and dose criteria will continue to be important for plant design; consequently, an interim revision of 10 CFR 50 will be carried out in parallel and will contain the present source term (i.e., that from TID-14844 and Regulatory Guides 1.3/1. 4). These proposed rule changes are expected to be issued for comment by about December 1991. Updated source term insights arising from the technical update of TID-14844 are expected to be made available for voluntary use by existing licensees. A final revision of 10 CFR 50 to incorporate updated source term and severe accident insights will then be undertaken, with a proposed rule for comment expected to be issued by September 1992. Although regulatory positions arising from updated source term insights remain to be developed, some preliminary implications can be seen at this time. It is clear that updated source term insights indicate the need for consideration of nuclides (e.g., cesium) in addition to iodine and the noble gases. In addition, revised insights on iodine chemistry calls into question the need for high efficiency charcoal adsorbers (assuming that the pH is controlled, post accident). These can, in turn, impact such important plant systems as fission product cleanup systems, control room habitability, and allowable containment leak rate. Finally, and most importantly, the above discussion and all recent risk studies have shown the importance of

LWR SEVERE ACCIDENT SOURCE TERM RESEARCH IN THE USA

21

maintaining containment integrity under severe accident conditions in order to assure low risk. This strongly suggests that the appearance of a severe accident source term within containment should be more closely linked with the temperatures, pressures and containment loads and challenges associated with such releases, rather than an arbitrary linkage with a single sequence such as a large break loss-of coolant-accident. REFERENCES 1 2 3 4 5 6 7 8 9

10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28

() J.J.DiNunno et al., “Calculation of Distance Factors for Power and Test Reactor Sites,” U.S. Atomic Energy Commission, TID-14844, March 1962. () USNRC, “Severe Accident Risks: An Assessment for Five U.S. Nuclear Power Plants,” NUREG-1150, December 1990. () M.F.Osborne, J.L.Collins, and R.A.Lorenz, “Experimental Studies of Fission Product Release from Commercial Light Water Reactor Fuel Under Accident Conditions,” Nucl. Technology, 78, 157, (1987). () M.F.Osborne, R.A.Lorenz, and J.L.Collins, “Atmospheric Effects on Fission Product Behavior at Severe LWR Accident Conditions,” ANS International Topical meeting, Portland, Oregon, July 1991. () E.C.Beahm, C.F.Weber, T.S.Kress, and R.J.Anderman, “Calculations of Iodine Source Terms in Support of NUREG-0956,” ORNL/NRC/LTR-86/17, January 1987. () E.C.Beahm, C.F.Weber, T.S.Kress, W.E.Shockley, and S.R.Daish, “Chemistry and Mass Transport of Iodine in Containment,” Proceedings of the 3rd Chemical Congress of North America Meeting, Toronto, Canada, June 2–3, 1988. () R.E.Adams, M.L.Tobias, and J.C.Petrykowski, “Aerosol Behavior in a Steam-Air Environment,” Proceedings of the CSNI Specialists’ Meeting on Nuclear Aerosols in Reactor Safety, Karlsruhe, Federal Republic of Germany, September 4–6, 1984. () R.E.Adams, A.W.Longest, and M.L.Tobias, “Influence of Moisture on the Behavior of Aerosols,” Proceedings of the 14th Water Reactor Safety Information Meeting, Gaithersburg, Maryland, October 27–31, 1986, NUREG/CP-0081, 1986. () M.L.Tobias, “Comparison of Computer Code Calculations with Experimental Results Obtained in the NSPP Series of Experiments,” Proceedings of the Workshop on Water-Cooled Reactor Aerosol Code Evaluation and Uncertainty Assessment, Brussels, Belgium, September 9– 11, 1987. () R.D.Spence and A.L.Wright, “The Importance of Fission Production/Aerosol Interactions in Reactor Accident Calculations,” Nuclear Technology 77(2), May 1987. () A.L.Wright, “Summary of TRAP-MELT 2 Results for Aerosol Transport Tests A103 and A104,” ORNL/NRC/LTR-85/22 (February 1985). () A.L.Wright and W.L.Pattison, “Summary of TRAP-MELT 2 Results for Aerosol Transport Tests A105 and A106,” ORNL/NRC/ LTR-86/9, Oak Ridge National Laboratory (December 1986). () A.L.Wright, W.L.Pattison, and J.-Y.King, “Series-2 Aerosol Resuspension Test Data Summary Report,” letter report sent to Dr. R.Y. Lee, RES, USNRC (Draft), dated March 1990. () R.M.Elrick, “Reactions Between Some Cesium-Iodine Compounds and the Reactor Materials 304 Stainless Steel, Inconel 600 and Silver, Vol. I: Cesium Hydroxide Reactions,” NUREG/CR-3197, June 1985. () R.Sallach, “Chemical Aspects of CsI Interaction in Steam with 304 Stainless Steel and Inconel-600,” NUREG/CR-4241, April 1986. () R.D.Elrick, “Boron-Carbide-Steam Reactions With Cesium Hydroxide and With Cesium Iodide at 1270 K in an Inconel System,” NUREG/CR-4963, September 1987. () D.A.Powers, “VANESA: A Mechanistic Model of Radionuclide Release and Aerosol Generation During Core Debris Interactions with Concrete,” NUREG/CR-4308, July 1986. () E.R.Copus, “Core-Concrete Interactions Using Molten Steel With Zirconium on a Basaltic Basemat: The SURC-4 Experiment,” NUREG/CR-4994, April 1989. () M.Allen et al., “Fission Product Release and Fuel Behavior of Irradiated LWR Fuel under Severe Accident Conditions: The ST-1 Experiment,” NUREG/CR-5345, to be published. () T.J.Heames et al., “VICTORIA: A Mechanistic Model of Radionuclide Behavior in the Reactor Coolant System under Severe Accident Conditions,” NUREG/CR-5545, October 1990. () J.Rest and S.A.Zawadski, “FASTGRASS: A Mechanistic Model for the Prediction of Xe, I, Cs, Te, Ba and Sr Release from Nuclear Fuel Under Normal and Severe Accident Conditions,” to be published as NUREG/CR report. () D.H.Thompson and J.K.Fink, “ACE MCCI Test L1 Test Data Report, Volume I—Thermal Hydraulics,” EPRI Report ACE-TRC14, Volume I, ANL, January 1990. () J.K.Fink, and D.H.Thompson, “ACE MCCI Test L1 Test Data Report, Volume II—Aerosol Analysis,” EPRI Report ACE-TR-C14, Volume II, ANL, January 1990. () Z.R.Martinson et al., “PBF (Power Burst Facility) Severe Fuel Damage Test 1–3 Test Results,” NUREG/CR-5354, October 1989. () D.A.Petti, Power Burst Facility (PBF) Severe Fuel Damage Test 1-4 Test Results, NUREG/CR-5163, April 1989. () D.J.Osetek, et al., “A Review of the PHEBUS-FP Test Program,” EG&G Technical Report , January 1991. () P.C.Owczarski, A.K.Postma, and R.I.Schreck, “Technical Bases and User’s Manual for the Prototype of SPRAC--A Suppression Pool Aerosol Removal Code,” Battelle Pacific Northwest Laboratories, NUREG/CR-3317, May 1985. () W.K.Winegardner, A.K.Postma, and M.W.Jankowski, “Studies of Fission Product Scrubbing within Ice Compartments,” Battelle Pacific Northwest Laboratories, NUREG/CR-3248, May 1983.

22

29

30 31 32 33 34 35 36 37 38

THE PHEBUS FISSION PRODUCT PROJECT

() J.A.Gieseke, M.Leonard, and R.DiSalvo, “Fission Product Transport in the Primary System, Important Phenomena, and Code Status,” International Seminar on Fission Product Transport Processes in Reactor Accidents, May 22–26, 1989, Dubrovnik, Yugoslavia. () R.S.Denning, P.Cybulskis, and J.A.Gieseke, “Changes in Source Term Perspectives,” Supplemental Proceedings for the ANS Thermal Reactor Safety Conference, February, 1986. () R.S.Denning and P.Cybulskis, “Source Term Analysis Methods for NUREG-1150,” 14th Water Reactor Safety Research Meeting, Gaithersburg, Maryland, October 30, 1986. () C.A.Alexander and J.S.Ogden, “Vaporization of UO2 at high temperatures and high pressures. A generic relation for volatilization,” High Temperatures-High Pressures, Vol. 21, pg 149–156, 1990. () F.J.Rahn, “Summary of the LWR Aerosol Containment Experiments (LACE) Program,” LACE TR-12, EPRI, Palo Alto, January 15, 1987. () USNRC, “Reactor Safety Study--An Assessment of Accident Risks in U.S. Commercial Nuclear Power Plants,” WASH-1400, NUREG-75/014, October 1975. () M.R.Kuhlman, D.J.Lehmicke, and R.O.Meyer, “CORSOR User’s Manual,” NUREG/CR-4173, March 1985. () H.Jordan and M.R.Kuhlman, “TRAP-MELT2 User’s Manual,” NUREG/CR-4205, May 1985. () R.K.Cole, Jr., D.P.Kelley, and M.A.Ellis, “CORCON-MOD2: A Computer Program for Analysis of Molten-Core Concrete Interactions,” NUREG/CR-3920, August 1984. () E.C.Beahm et. al., C.F.Weber, T.S.Kress, W.E.Shockley, S.R. Daish, Chemistry and Mass Transport of Iodine in Containment,” pp 251– 266 , Proceedings of the 2nd CSNI Workshop on Iodine Chemistry in Reactor Safety, Toronto, Canada, June 1989.

SURVEY OF SEVERE ACCIDENT EXPERIMENTS AND ANALYSES IN JAPAN M.AKIYAMA*, K.TAKUMI** AND K.SODA*** University of Tokyo, 7–3–1, Kongo, Bunkyo-ku, Tokyo* Nuclear Power Engineering Center, 4–3–13, Toranomon, Minato-ku, Tokyo** Japan Atomic Energy Research Institute, Tokai-mura, Ibaraki*** Japan

SUMMARY An overview on Japanese activities of LWR severe accident experiments and analyses is presented, covering various fields and topics of experimental investigation on severe accident phenomena such as fuel damage and melt progression, fission products release and transport, and component and containment integrity. The current status of analytical investigation on severe accident is also described in the fields of the level-1 and level-2 PSA studies, code development and assessment activities. The basic considerations on accident management is summarized. 1. BACKGROUND Present Status of Nuclear Power Generation in Japan Since the first operation of Tokai Unit-1 (GCR) in 1966 followed by the start of Tsuruga Unit-1 (BWR) and Mihama Unit-1 (PWR) in 1970 and of Fukushima Unit-1 (BWR) in 1971, the nuclear power generation in Japan evolved steadily to become to play a leading role in the total electric power supply. Currently, forty nuclear power plants amounting to the capacity of 32, 059 MWe are in operation, accounting for approximately 18% of the total electric power generation capacity and producing about 26% of total power generation as of 1989 in Japan. Most of these nuclear power reactors are light water reactors (PWR and BWR), comprising 21 BWRs with total capacity of 18,137 MWe and 18 PWRs with total capacity of 13,756 MWe. Five units of BWR and also five units of PWR are now being constructed, and they will start operation one by one in the near future. In parallel with BWR and PWR, various types of advanced nuclear reactors such as an advanced boiling water reactor (ABWR), an advanced pressurized water reactor (APWR), an advanced thermal reactor (ATR) and a sodium-cooled fast breeder reactor (FBR) have been developed, and other types of reactors such as a small or medium sized light water reactor and a high temperature gas-cooled reactor are under investigation. The construction of two units of ABWR, each having the capacity of 1360 MWe, has just begun, and will be completed in mid 1990s. The prototype plant of ATR named “Fugen”, a heavy-water moderated and light water cooled pressure tube type reactor of 165 MWe, is being operated and its first commercial plant is planned to be installed within ten years. The construction work of the prototype plant of FBR named “Monju” with capacity of 280 MWe has been completed, and after integrated testings it will go to critical in October, 1992. With respect to the first commercial FBR, the basic design concept has been established and various kinds of engineering research and development works are being promoted. The average capacity factor of commercial nuclear power plants in Japan in the fiscal year (FY) of 1989 stood at 70.0% for the 37 units in operation, with total installed capacity of 29,280 MWe. The maximum allowed operation cycle period of Japanese nuclear power plant is thirteen months by regulation, and the actual average operation period of all units was 347 days (11.6 months) in FY 1989. And the average inspection time for units on which periodical inspection was performed in FY 1989 was 155 days (5.2 months), an increase of 20 days (0.7 month) on the previous year. The effort based on the preventive maintenance is contributing to ensure high reliability and safety. In recent years, a mechanical failure of the component of the reactor recirculation pump in a BWR and a steam generator tube rupture of a PWR followed by the activation of the safety injection system have been experienced in Japan. These have brought a longer inspection and repair time to result in a slightly lower marks of the average load factor. However, these experiences will be reflected in the future improvement of plant operation and performance.

24

THE PHEBUS FISSION PRODUCT PROJECT

It is to be noted that there has been no uncontrolled radioactivity release into the environment in Japan, which may be attributed to the fact that ensuring safety has been the top priority in the development and utilization of nuclear energy. Assurance and Enhancement of Safety The purpose of achieving safety of a nuclear power plant is to protect the lives and health of the public from exposure to radioactivity during normal operation, abnormal occurrences and accidents(1). Sufficient measures are taken during the stages of site selection, design, construction, operation and maintenance. Appropriate protective measures are also taken to reduce the radiation dose to workers at the plant. Regulatory practices in Japan are set to conform with the basic safety principles in which the first priority is given to prevention of the occurrences of any accident which might develop into a serious situation. Rigorous practices of operation and maintenance to detect and prevent any abnormal occurrences at the plant are commonly pursued by the Japanese utilities. Government initiated activities for further enhancing safety of a nuclear power plant are seen in the Safety 21 Project of the Agency of Natural Resources and Energy (ANRE) of the Ministry of International Trade and Industry (MITI)(2), the demonstration tests at the Nuclear Power Engineering Center (NUPEC) to prove the safety of a nuclear reactor to the public, and safety research conducted by the Japan Atomic Energy Research Institute (JAERI) in accordance with the annual safety research plan set by the Nuclear Safety Commission (NSC)(3). Efforts are also made by the manufacturers and the utilities in Japan not only to enhance the safety of nuclear reactors of current generation, but also to develop a nuclear reactor of next generation with much improved and innovative safety features. The “Safety Culture” must be the basis of assuring and enhancing the safety of a nuclear reactor all over the world as the International Nuclear Safety Advisory Group (INSAG) of the International Atomic Energy Agency (IAEA) promotes in its report, “The Basic Safety Principles for Nuclear Power Plants”(4), and Japan is no exception with this regard. National Position on Severe Accident(5) In the procedure of nuclear power plant licensing in Japan, the safety examination on basic design is performed based upon the relating laws and guidelines whose requirements on safety design are prescribed within the design basis accident. Accordingly, the severe accident issues are not involved in the current licensing procedures. In recent years, however, it became widely recognized that severe accident experiments and analyses are important to understand the safety tolerance of the system in detail and also to investigate how to improve the accident management measures. These will be achieved by utilizing the theory and knowledge base accumulated in the course of performing severe accident experiments and analyses. As a background for the safety examination of a particular design of a nuclear power plant or for the examination of its operational safety, PSAs on a reference plant similar to the particular design are being performed, and the results are being taken into account as reference materials for the safety evaluation. In this sense, the severe accident experiments and analyses are becoming a useful support for the safety licensing procedure. The NSC’s position on severe accident is summarized as follows; (1) The knowledge of severe accident is one of the most important basis for the formulation of safety design criteria, siting criteria, and guideline for emergency planning. (2) Plant operator should have knowledge of severe accident and reflect it upon the plant management so as to be able to cope with properly even in cases of beyond design basis accident. (3) Industry and research organizations should perform severe accident research of which purposes are; * * * *

To identify phenomena associated with a severe accident, To develop analytical tools for source term analysis, To estimate a risk and safety margin of plant, and To evaluate measures to prevent and mitigate severe accident by design and/or accident management.

SURVEY OF SEVERE ACCIDENT EXPERIMENTS AND ANALYSES IN JAPAN

25

Fig. 1. Fuel Component Interaction Test Apparatus

Fig. 2. Structural Analysis of the TMI-2 Vessel Bottom Head

2. CURRENT STATUS OF SEVERE ACCIDENT EXPERIMENTS AND ANALYSES 2.1 Experimental Investigation on Severe Accident Phenomenology Fuel Damage and Melt Progression Experimental investigation of in-vessel melt progression in Japan relies largely on the international research collaboration involving a large scale experiment such as the Cooperative Severe Accident Research Program (CSARP) and the TMI-2 R & D Program in U.S.A., the CORA experiment in Germany, the LOFT program of OECD and the PHEBUS program in France. Analyses of such experimental data have provided us insights into how core melt progresses during a severe accident. To better understand and interpret the data of the large scale experiments, fuel damage experiment was performed at JAERI by using the Nuclear Safety Research Reactor (NSRR). It should be mentioned that NSRR is the only operating research reactor in the world that is capable of performing a test simulating reactivity initiated accident (RIA) conditions. Interaction of fuel with control rod and component materials has been studied to supplement the large scale experiment. The apparatus used for this purpose is shown in Fig. 1. It is noted that the TMI-2 debris will be examined at JAERI beginning spring of 1991. Mechanism of vessel failure due to the attack of molten core to the reactor vessel still remains with a large uncertainty. The TMI-VIP program is expected to provide useful information to reduce the uncertainty. Structural analysis of the reactor vessel was performed at JAERI to interpret the data obtained in the TMIVIP program as shown in Fig. 2.

26

THE PHEBUS FISSION PRODUCT PROJECT

Fig. 3. Schematic Diagram of ALPHA

Fig. 4. Schematic Diagram of the Test Apparatus for Reaction Kinetics Experiment

Ex-vessel melt progression after the vessel failure plays an important role in determining timing of the containment failure and source terms. Problems arising during the ex-vessel melt progression include core-concrete interaction and melt coolability in a containment. Hydrogen generation and burn in the containment are also affected by the ex-vessel melt progression. The Anticipated Load and Performance of a Containment in a Hypothetical Accident (ALPHA) program at JAERI focuses on the ex-vessel melt progression, especially core-concrete interaction and molten core coolability in a containment in which steam explosion may have an influence on the integrity of a reactor vessel and a containment(6). Schematic diagram of the ALPHA test facility is shown in Fig. 3 and the major capability of the facility is summarized in Table 1. Table 1: Major Dimensions and Capabilities of ALPHA Volume Height Diameter Design Pressure Design Temperature

50 5.7 3.9 2 250

m3 m m MPa °C

Fission Product Release and Transport Fission product release and transport have been studied at JAERI from the view point of supplementing the large scale integral experiment data such as those obtained from the CSARP program and the PHEBUS FP program. Formation of organic iodine in the radiation field is experimentally studied with the small test apparatus at JAERI. Basic reaction kinetics of iodine, cesium and tellurium with component materials such as Fe, Ni and Cr are investigated by using the apparatus illustrated in Fig. 4. Future test is being planned at JAERI to fabricate a test assembly with which fission product

SURVEY OF SEVERE ACCIDENT EXPERIMENTS AND ANALYSES IN JAPAN

27

Fig. 5. Schematic of Out-of-Pile Fission Products Release Test Facility

Fig. 6. Schematic Diagram of EPSI

Fig. 7. Radioactive Material Trapping Test Apparatus

release from a damaged fuel will be investigated at high temperature of 2800°C under various conditions ranging from oxidized to reduced environment. The schematic diagram of the fission product experiment is presented in Fig. 5. The design of the facility is in progress. As was pointed out by the PSA studies at JAERI, pool scrubbing efficiency is one of the dominant factors influencing source term and therefore, the experimental facility for pool scrubbing investigation (EPSI) shown in Fig. 6 was fabricated at JAERI to quantify the efficiency. Experiment results indicate that pool scrubbing is extremely effective to remove fission products even at elevated temperature and pressure(7). NUPEC is now promoting a test program of radioactive material trapping in the leakage path of a containment such as electric penetration assembly (EPA) and an equipment hatch. In this test, iodine trapping effect in the leakage path will be investigated under simulated severe accident conditions. Organic seal materials such as epoxy resin and silicon resin will be used in these penetrations as an insulator or a gasket. It is assumed in the tests that the leakage path grows at the organic seal due to the increased temperature and pressure beyond the design limit. The test program consists of a bench scale test for surveying controlling parameters which affect the efficiency of trapping iodine in the leak path and a large scale test for confirming and evaluating iodine trapping effect in which a test assembly simulating that used in a actual plant will be utilized. The schematic diagram of the test apparatus and the proposed test sections are shown in Fig. 7 and Fig. 8, respectively.

28

THE PHEBUS FISSION PRODUCT PROJECT

Fig. 8. Proposed Test Sections

Fig. 9. Predicted Yielding Zones for the 1/6 SNL RCCV Experiments

Component and Containment Integrity Integrity of the containment has become focus of attention especially since the Chernobyl accident for which no strong containment of the western type existed and the role of the containment was reaffirmed by the accident. The load to the containment comes from the ex-vessel phenomena such as steam explosion, hydrogen burn, over pressure and over temperature in the containment. Effect of dynamic pressure as well as static pressure exceeding the design limit to the containment integrity has been studied at JAERI by the finite element analysis code. The result indicated that the containment will maintain its integrity even if the pressure reaches 4 to 5 times of the design pressure. The predicted result by JAERI of the 1/6 scale reinforced concrete containment vessel (RCCV) experiment at the Sandia National Laboratories (SNL) is shown in Fig. 9 in which the first yielding zones were in good agreement with experimental results(8). Leak rate tests at high pressure and temperature have been carried out as a part of ALPHA program at JAERI to characterize the failure mechanism of the penetrations for instrument cables and power cables. Experiments conducted so far show no leakage resulting from high pressure, but a potential leakage may occur due to the high temperature in the containment. A proving test program on containment integrity is promoted by NUPEC(9), whose objectives are; to investigate in detail the behavior of hydrogen which may be generated in a severe accident and may threaten the integrity of a containment vessel, and to confirm that the function and the integrity of a containment vessel are maintained. In the program are included hydrogen mixing and distribution tests, hydrogen burning tests, and tests to failure of a steel containment vessel (SCV) and a prestressed concrete containment vessel (PCCV). The hydrogen mixing and distribution tests are to investigate their behaviors in the containment vessel with multiple compartments representing a typical large dry containment of a PWR. The test vessel has a volume of 1,600 m3 that is about 1/ 4 scale of an actual PWR containment vessel. A schematic drawing of the test facility is shown in Fig. 10 and the test conditions are given in Table 2. In the experiment, helium gas is used for the safety of the test and items for investigation include the effects of forced convection driven by injection of steam and helium into the containment, the effects of heat sink by the vessel wall and compartment wall. Table 2. PWR MIxing and Distribution Test Conditions Concentration of He Concentration of Steam Water Spray No. of Compartments

< 18 < 60 70 25

vol % vol % m3/h

SURVEY OF SEVERE ACCIDENT EXPERIMENTS AND ANALYSES IN JAPAN

Maximum Pressure

5.0

29

MPa

Fig. 10. Hydrogen Mixing and Distribution Test Facility

Fig. 11. Helium Concentration Distribution

Example of the test result is shown in Fig. 11 where the pretest prediction of concentration of helium in the compartments is compared with experimental data. The NUPEC tests conducted so far suggest that hydrogen will be well mixed in a containment vessel and the prediction by the computer code is in excellent agreement with the data. Hydrogen burning test is conducted at NUPEC with the objectives to investigate hydrogen burning phenomena including mitigation effect of steam, spray, and nitrogen inerting in a containment vessel, and to confirm containment integrity against hydrogen burning. The hydrogen burning tests are conducted by using a small scale cylindrical vessel with 5 m3 as shown in Fig. 12 and a large scale spherical vessel with 270 m3 as shown in Fig. 13. The summary of test conditions is given in Tables 3 and 4. In the small scale test, the effects of temperature, pressure, turbulence, spraying, distribution and concentration of gases have been investigated in detail prior to the large scale test. An example of the result is shown in Fig. 14 in terms of isoarrival time contour of burning front, and a comparison of the NUPEC data with previously performed FITS test data at SNL is presented in Fig. 15 in terms of the peak combustion pressure normalized with respect to the initial pressure. The NUPEC data are in good agreement with the FITS data which were obtained at the lower hydrogen concentration condition. New data bases have been added in the higher hydrogen concentration by the NUPEC data. Table 3. Hydrogen Burning Test Conditions for Small Scale Test Items

BWR

PWR

H2 Concentration (vol%) Steam Concentration (vol%) N2 Concentration (vol%) Oxygen Concentration (vol%) Spray Flow Rate (m3/h)

<70 <60 <97 <10 <15

20 60 atmospheric atmospheric <3

Items

BWR

PWR

H2 Concentration (vol%) Steam Concentration (vol%)

<70 <60

18 60

Table 4. Hydrogen Burning Test Conditions for Large Scale Test

30

THE PHEBUS FISSION PRODUCT PROJECT

Items

BWR

PWR

N2 Concentration (vol%) Oxygen Concentration (vol%) Spray Flow Rate (m3 /h) No. Compartments

<97 <10 <350 none

atmospheric atmospheric 45 8

Fig. 12. A Small Scale Hydrogen Burning Test Facility

Fig. 13. A Large Scale Hydrogen Burning Test Facility

Fig. 14. Iso-Arrival Time Contour of Burning Flame Front

Fig. 15. Comparisons of Normalized Peak Combustion Pressure

Failure tests of a containment due to over-pressurization of SCV and PCCV are in preparation in which the scales of these tests are 1/10 th scale for SCV with 1/5 thickness and 1/6 th scale for PCCV as illustrated in Fig. 16. Failure test of a flange is also planned.

SURVEY OF SEVERE ACCIDENT EXPERIMENTS AND ANALYSES IN JAPAN

31

Fig. 16. Shapes and Sizes of Test Containtment Vessels

Experimental Study on Accident Management Schemes Accident management has become an important issue in terms of prevention and mitigation of a severe accident. For the prevention phase of accident management, the utilities have set up operational procedures to terminate an accident early enough so that the accident never goes into a severe accident. Experimental and analytical studies have been also made for this aspect to prove and propose methods of accident management. ROSA-V program is now planned at JAERI to focus accident management during a transient and accidents. Experiments will be conducted to demonstrate the effectiveness of the method of accident management by using the 1/48 scale large test facility simulating a PWR. ALPHA program pays an attention to the ex-vessel phase of accident management scheme such as terminating further core degradation by adding water on top of molten core material and mitigating consequences of severe accidents by scrubbing and/or filtering. Experiments are in progress by using the EPSI facility. 2.3 Analytical Investigation on Severe Accident Recent Level-1 and Level-2 PSAs for LWR Plants Probabilistic Safety Assessment (PSA) is recognized as the convincing tool to support the deterministic method to assess the balance of design and assist regulatory activities of nuclear power plants. From this point, the preparation and application of PSA methodologies are under way with collaboration among the government organizations and industries. Among the governmental organizations, JAERI has been developing a methodology of PSA while the Japan Institute of Nuclear Safety (JINS) of NUPEC has been conducting level-1 and level-2 PSAs for typical Japanese BWRs and PWRs since 1987 for the level-1 PSA. The evaluations of 1,100 MWe-class BWR-5 with MARK II containment and 1,100 MWe-class four loop PWR with a large dry steel containment were completed in 1989. The initiating events selected for the level-1 PSA at JINS were limited to the internal events such as a loss of coolant accident (LOCA) and abnormal operational transients during high power operation. In this study, small event trees and large fault trees were constructed in which the SETS code series were used. The data base of component failure rates was mainly composed of the IREP and the LER data while the data on the emergency diesel generator failure rate and recovery rate of failed off-site power were based on Japanese experiences. The result of the JINS PSA showed that the total mean core damage frequency (CDF) for the BWR is about 2.0×10−7/ Reactor Year (RY). The upper 5% value is 4.9×10−7/RY and the lower 5% value is 8.9×10−9/RY. Corresponding error factor is 7.4. The CDFs are presented in Fig. 17 for all initiating events considered. Among the initiating events, LOCAs have contribution of 83% to the total CDF and the secondary side break, steam generator tube rupture (SGTR), and other events have 7%, 5%, and 5%, respectively. The reevaluation for the 1,100 MWe-class BWR and PWR is ongoing using the revised data base which includes more domestic data on the component failure rates based on the operational experiences. Two year program has started of surveying the possibility of core damage occurrence during the maintenance activities at the plant shutdown state. Level-2 PSA at JINS follows the level-1 PSA mentioned above. Containment event trees were developed, considering physical phenomena influencing on fission product release timing and the recovery of failed safety systems. Dominant accident sequences were analyzed using the Source Term Code Package (STCP) and point-estimated values of fission product release frequency and source term were obtained for each release category. The result shows that the dominant accident sequences concerning fission product (FP) release frequency are not necessarily the same as those for the core damage frequency as seen in Fig. 18. For example, in case of a PWR, the sequences initiating from SGTR (group “G”) occupy about 36% as regard to the release frequency, whereas it was only 5% in the core

32

THE PHEBUS FISSION PRODUCT PROJECT

Fig. 17. Core Damage Frequencies (CDF) Classified by Initiating Events

Fig. 18. FP Release Frequency in Comparison with Core Damage Frequency

damage frequency. In BWR case, the TQUX sequence was the most important regarding core damage frequency, however, its FP release frequency is very low. From the view point of the containment failure mode the delayed containment failure due to over-pressurization is the most important as shown in Fig. 19. In the future, the results will be reevaluated, based on the revised level-1 PSAs and using the MELCOR code. Accident Management Strategies In the accident management, various measures will be involved such as operational procedures, special equipments, communications and so on. Many of these have to be prepared primarily by the owners of nuclear power plants in coordination with the basic safety considerations of the regulatory body. From this point of view, the Japanese electric utilities have been making a lot of effort, for instance to provide emergency procedures, while MITI and NSC are making studies on the basic considerations for further measures in accident management. As to the operational procedures for accidents, the electric utilities have already partly proceeded to the procedure to cope with the unexpected events which has not been described in the event-based operational procedures during accidents. They have decided to develop so-called symptom-based procedure, an improved version of the emergency procedures as mentioned above. The operator actions systematically analyzed are categorized into three basic functions, namely, reactor shutdown, core cooling and containment vessel integrity. The procedure is divided into the following two categories; accident progression from initiating events to core damage (category 1) and after core damage (category 2). They should be supported by the information obtained from level-1 and level-2 PSAs respectively. Operators are expected to use the symptom-based procedure when the event is beyond or not under control of the eventbased procedures. The addition of symptom-based procedure brings flexibility to the operators, and the preparation of authorized recovery procedure against the error following the event-based procedure surely contributes to the reduction of work load.

SURVEY OF SEVERE ACCIDENT EXPERIMENTS AND ANALYSES IN JAPAN

33

Fig. 19. Relative Occurrence of Containment Failure Mode

Fig. 20. Simulator Rods Temperature in CORA-2 Experiment

Comprehensive studies on the measures for keeping integrity of containment vessel, particularly on venting for BWR and igniters for ice-condenser type containment of PWR, are being made in the light of the results of level-1 and level-2 PSAs. Because the results of probabilities of damage of core and containment vessel are small, the venting and igniters do not seem to bring a sensitive effect on improving safety. However, the survey on the reliability of igniters are being continued. Analytical investigation of accident management was performed at JAERI with emphasis on effectiveness of intentional depressurization and of reflooding of damaged core. Findings from such analyses will be further investigated experimentally to confirm and quantify the effectiveness of accident management scheme. For this purpose, experiments are made with ALPHA, ROSA-V and EPSI facilities as described in the previous chapters. Code Development and Assessment Code Development has been primarily pursued at JAERI and verification and assessment of the codes have been extensively done. The THALES code package has been developed by JAERI for level-2 PSA studies of typical nuclear reactors, BWR and PWR. In order to benchmark the THALES code modeling, the detailed mechanistic codes have been developed for experimental data analysis and benchmark calculation. Such codes include MUFLAR, HORN and REMOVAL. MUFLAR is a twodimensional core wide analysis code which was used for analyzing core damage progression. HORN is the code which is capable of predicting chemical forms of fission products along the release path of fission products. REMOVAL is the aerosol analysis code which has been validated against the LACE experiments. Assessment of the integrated code such as SCDAP/RELAP5 and MELCOR has been carried out by applying to reactor situation such as the TMI-2 accident. Large scale tests such as PBF/SFD, CORA and PHEBUS/SFD were also used for code assessment. An example of the CORA experiment analysis is shown in Fig. 20. For accident management, analysis was performed by using RELAP/SCDAP for intentional depressurization procedure with or without pump seal break in collaboration with USNRC. Participation in the international standard problem exercise organized by the Committee of Safety of Nuclear Installations (CSNI) of OECD has become one of the important activities for the code assessment such as the TMI-2, CORA, BETA and PHEBUS experiments.

34

THE PHEBUS FISSION PRODUCT PROJECT

3. CONCLUDING REMARKS It should be taken deep in mind that the future of nuclear power generation does depend largely whether it can be safe enough or not. Ensuring safety by means of advanced design, engineering development, human factor research, regulatory considerations etc. are to be continuously promoted all over the world. With respect to this point, one of the most important things may be the unlimited effort toward physical understandings of system behaviors and processes involved, which certainly leads to the sufficient capabilities for safe control of systems and for the emergency control. This is believed to be one of the primary concerns of both experts and the public. There is a growing need for investigation of severe accident phenomena, because it is inevitable for the reliable evaluation of safety margin, for the reasonable management of accident if happened, and so on. Although there has been significant progress to understand severe accident phenomena, the physics seems to be much complicated and certain aspects of severe accident phenomena still remain with large uncertainties, for which we should recognize needs of additional works. The international cooperation on severe accident experiments and analyses is truly indispensable, since the severe accident is the common issue among the countries and its investigation can be promoted effectively by sharing information and resources. In this sense, the PHEBUS/FP program is playing a key role with its distinguished facilities and associated research environment, and the Japanese programs on severe accident experiments and analyses at JAERI and NUPEC are complementary to PHEBUS/FP and other programs. Finally, further promotion of computing capability to simulate accident progression and consequences is highly recommended. REFERENCES 1 2 3 4 5 6 7 8 9

() Uchida, H., The Principal Nuclear Safety Policies in Japan, The Second NEA Seminar on Interface Questions in Nuclear Health and Safety, Paris, September 12–13, 1990. () ANRE, Improvement of Safety Assurance Measures for Nuclear Generation —Safety 21, MITI, Tokyo (1986). () Special Committee on Safety Researches on Nuclear Facilities, The Annual Plan on Safety Research for Nuclear Installations (FY 1991–1995), NSC, September 1990. () INSAG, Basic Safety Principles for Nuclear Power Plants, IAEA-INSAG-3, March 1988. () Kondo, S., Overview of Severe Accident Research in Japan, Japan-France Information Exchange Meeting on Reactor Regulation, Paris, November 1990. () Soda, K. (ed), Proc. of the First LWR Severe Accident Research, November 21–22, 1990, Tokaimura, Japan, JAERI-memo 02–428 (Jan. 1991). () Hashimoto, K. et al., High Pressure Pool Scrubbing Experiments for a PWR Severe Accident, ANS Thermal Reactor Safety Meeting, July 21–25, 1991. Portland, U.S.A. (to be presented). () Kimura, H., Structural Analysis of the RCCV Under Extreme Static and Dynamic Loading, Proc. the Second International Conference on Containment Design and Operation, October 14–17, 1990, Toronto, Canada. () Nonaka, A., Proving Test on the Reliability for Reactor Containment Vessel, Proc. the Second International Conference on Containment Design and Operation, October 14–17, 1990, Toronto, Canada.

Discussion following the presentations of Session I Summary of the chairman Mr. S.Finzi

Summarizing the introductory remarks of Messrs. Livolant (CEA/IPSN) and Contzen (CEC/JRC), and the presentation and discussion of the situations in Europe, the USA and Japan regarding source term research, two important messages can be obtained: 1) It is important to situate the source term issue in the general context of NPP safety, in particular as far as the containment performance is concerned, in order to optimize the research programme. The US-NRC in addition applies criteria which assure the integrity of the containment for beyond design basis accident loads which fall in the predetermined probabilistic framework. 2) It is desirable to reach consensus at a Community level and possibly worldwide, on the- relevance of research results for NPP safety assessments in order to provide for adequate regulatory measures and decision making approaches. The overall confirmation through integral experiments in the field of source term requires rather expensive tools, the operation of which calls for highly specialized teams of scientists and for careful planning. As Mr. Contzen pointed out, it is difficult to mobilize such large-scale efforts over a long period and serious considerations should be devoted to the ways and means of maintaining continuously the convection in the decision makers’ minds that research in this field is vital. The establishment of a wide cooperation in research, and particularly in ambitious projects like Phebus-FP is the most efficient way to achieve these goals. To assure a substantial gain in information on FP transport and behaviour, the fuel degradation in the Phebus-FP should be pushed as far as feasible to enhance FP release. May be some experiments could also be tailored to get some additional information on the oxidation aspects.

SESSION II STATE OF THE ART DEDUCED FROM PREVIOUS LARGE EXPERIMENTS

Core degradation and fission product release R.W.Wright , NRS Washington and S.J.L.Hagen , KfK Karlsruhe Fission product transport W.Schöck , KfK Karlsruhe and J.O.Liljenzin , Chalmers Tekniska Högskola, Göteborg Fission product chemistry in severe reactor accidents: review of relevant integral experiments A.L.Nichols , AEA Technology Winfrith and C.Hueber , CEA/IPSN Cadarache Phebus-CSD Phebus severe fuel damage programme: main experimental results and instrumentation behaviour C.Gonnier , G.Repetto , CEA/IPSN Cadarache and G.Geoffroy , CEA Saclay Review of B9+benchmark results B.Adroguer , CEA/IPSN Cadarache and P.Villalibre , CSN Madrid Summary of discussion A.Meyer-Heine , CEA/IPSN Cadarache

CORE DEGRADATION AND FISSION PRODUCT RELEASE Robert W.Wright U.S. Nuclear Regulatory Commission and Siegfried J.L.Hagen Kernforschungszentrum Karlsruhe

SUMMARY Experiments on core degradation and melt progression in severe LWR accidents have provided reasonable understanding of the principal processes involved in the early phase of melt progression that extends through core degradation and metallic material melting and relocation. A general but not a quantitative understanding of late phase melt progression that involves ceramic material melting and relocation has also been obtained, primarily from the TMI-2 core examination. A summary is given of the current state of knowledge on core degradation and melt progression obtained from these integral experiments and of the principal remaining significant uncertainties. A summary is also given of the principal results on in-vessel fission product release obtained from these experiments. 1. OVERVIEW This paper describes the current state of knowledge on core degradation and in-vessel core-melt progression in LWR core uncovery accidents and gives a summary of the results on in-vessel fission product release from integral in-reactor tests. Melt progression describes the state of an LWR reactor core from core uncovery up to reactor vessel meltthrough in unrecovered accidents, or through temperature stabilization in accidents recovered by core reflooding. Melt progression provides the initial conditions for assessing the core-melt threat to containment integrity, in particular the threat of early containment failure, which, along with the containment bypass sequences, provides the principal contribution to severe accident risk (1). Significant parameters involved here are the melt mass, composition, temperature (superheat), and rate of release from the vessel. Melt progression provides the in-vessel hydrogen generation and the conditions that govern the in-vessel release of fission products and aerosols and their transport and retention in the primary system. Melt progression also provides the core conditions for assessing accident management strategies. Sensitivity studies have shown that uncertainties regarding melt progression provide major uncertainties in assessing severe accident consequences and risk. Much has been learned about the processes involved in core degradation, in the early phase of melt progression that extends through metallic (but not ceramic) material melting and relocation, and in in-vessel fission product and aerosol release. This information has come from many integral tests in the PBF, ACRR, NRU, NSRR, and TREAT test reactors, from the LOFT FP-2 test, from the Phebus CSD tests, from tests in the CORA ex-reactor fuel damage test facility, and from separate-effects experiments on significant phenomena. Most of the available information on the late phase of core-melt progression that involves ceramic material melting and relocation has come from the post-accident examination of the TMI-2 core. Despite the core reflooding that successfully terminated the TMI-2 accident, the general late-phase melt progression phenomenology of that accident appears to be applicable to unrecovered as well as to recovered PWR accidents. If there are any BWR accidents which involve metallic core blockage like that at TMI-2, then this general phenomenology should also be applicable to such accidents. The sources of the current information base on melt progression from integral experiments and an outline of the information obtained from these experiments is given in Table 1. The results of these integral tests and the TMI-2 core examination have provided a very consistent picture of melt progression(2, 3). This picture involves the development of a debris-supporting metallic blockage across the lower core and above the water level during coolant boil down, and is called the blocked-core accident pathway. The TMI-2 core examination has shown that a pool of mostly ceramic melt grows from decay heat in the particulate, mostly ceramic debris bed that is supported by the metallic core blockage. The growing pool melts through the supporting metallic blockage and secondary ceramic crusts that surround the ceramic melt pool, or out the side of the core as happened at TMI-2 with a reflooded core, and the melt then drains into the vessel lower plenum.

38

THE PHEBUS FISSION PRODUCT PROJECT

The end-state configuration of the TMI-2 core, shown in Figure 1, illustrates the blocked core accident pathway(3). The central region of the core contains refrozen ceramic melt from the undrained portion of the ceramic melt pool. Below this pool is a metallic crust that had previously blocked the core, and above the pool is a mostly ceramic crust. Below the metallic crust are undamaged sections of fuel assemblies that were cooled throughout the accident by water in the bottom of the core. At the side of the core is the drainage pathway through which 20% of the core mass drained into the lower plenum water upon pool meltthrough out the side of the core. Refrozen ceramic melt and solid debris are in the vessel lower head. There was no meltthrough at the metallic downward protuberance near the core axis. Above the pool and upper crust are a mostly ceramic particulate debris bed and a large void produced by subsidence of the bed from debris densification in melting and from drainage of the melt pool. The TMI-2 configuration illustrates essentially all of the detailed melt progression phenomenology thought to apply to blocked core accident sequences in PWRs and also to any BWR accidents that follow blocked core sequences. According to current knowledge, however, it is possible that meltthrough in unrecovered accidents would occur near the core axis causing full drainage of the ceramic melt pool. A major finding from all the integral tests and also from the TMI-2 core examination is that the unoxidized zircaloy and the control rod materials and their eutectics melt before or during the rapid temperature transient from steam oxidation of the core zircaloy and at temperatures ranging from as low as 1200K for the eutectics up to 2250K for zircaloy(4). A diagram of the threshold temperatures for liquid phase formation for the relevant materials in LWR accidents, both from melting and from eutectic materials interaction, is given in Figure 2(5). TABLE 1 Sources of Current Integral Experimental Information DATA SOURCE

KEY INFORMATION

PBF Severe Fuel Damage Tests: SFD-ST, 1–1, 1–3, 1–4 ACRR Damaged Fuel Tests: DF −1, −2, −3, −4 NRU Full-length Tests: FLHT 1, 2, 4, 5

Integral information base on core degradation and melt progression Phenomenological information on core degradation and melt progression Data on length effects, lack of cut off to hydrogen generation, and high burnup fuel swelling Basic information base on material-interaction effects and metallic melt relocation Information for BWR & PWR geometries, including reflood effects, using electrically-heated, simulated fuel-rod bundles up to 59 rods Phenomenological information on core degradation processes Unique data on metallic melt relocation and lack of hydrogen cutoff with a large flow-bypass area Significant results on reflood effects on melt progression Unique results with fission-product decay heating Dry UO2 debris-bed melting characteristics Dynamics of pool growth in ceramic debris beds Major source of significant information on late-phase melt progression Results applicable to basic phenomenology for both recovered & unrecovered accidents

CORA Ex-reactor Tests: 9 PWR Tests 3 BWR Tests Related Experiments Phebus SFD Tests LOFT FP-2 Large-Bundle (101-rod) Test

ACRR Late Phase (Ceramic Melt) TMI-2 Core Examination Tests

Videotapes of the CORA experiments have shown that metallic melt relocation is essentially a noncoherent, noncoplanar rivulet flow process that is quite different from the coplanar modeling of film flow along the rods that is used in current codes (6). This explains why the developing partial blockages in the PBF and other integral multi-rod tests did not cut off the steam flow and hydrogen generation. Metallic melt relocation leaves behind free-standing UO2 fuel pellets and ZrO2 oxidized cladding shards that have melting points (including eutectics) in the range for 2800K to 3100K. During late-phase melt progression in unrecovered accidents and in very severe recovered accidents such as TMI-2, a ceramic melt pool forms in the mostly ceramic debris bed with further core heatup. Thus the metallic and the ceramic debris with melting points that differ by 600K or more become separated in space, and, as the TMI-2 core examination show, they behave quite differently in continued melt progression. Older simplified codes treat the core melt as a single fictitious “corium” fluid with a unique (high metallic) composition, a relatively low melting point (usually 2550K), and with an assumed fraction of the core mass released as “corium” melt upon an assumed nonmechanistic core “slumping.” BWR core geometry tests in ACRR (DF-4) and CORA have shown that eutectic interactions of the control-blade B4C powder and their stainless steel sheaths at about 1500 K liquify the blade materials(4). This melt undergoes further eutectic interactions with the zircaloy channel box walls that fail the walls and open up the compartmentalized BWR core geometry. The control-blade materials relocates downward and possibly even out of the core in a BWR accident with potential recriticality

CORE DEGRADATION AND FISSION PRODUCT RELEASE

39

Fig. 1. TMI-2 End State Configuration

Fig. 2. LWR Severe Accident Relevant Liquied Phase Formation Temperatures

problems for core reflooding. In these tests, however, which were all performed for BWR wet core conditions (that is with relatively high steaming rates and relatively cold temperatures in the lower sections of the test bundle), a metallic partial blockage formed at the bottom of the test bundles as in PWR core geometry tests. This question of metallic melt drainage or core blockage, particularly for BWRs, is a major branch point for in-vessel core melt progression, and it has a large effect upon the characteristics of the melt released from the core into the lower plenum. In the core blockage case, a large mass of mostly ceramic melt at about 3000K drains rapidly into the water filled lower plenum, as happened at TMI-2. In the drainage case, layers of quenched melt are formed under the lower plenum water in the order of their time of melting and drainage from the core, with the low-melting metals at the bottom and the ceramics at the top. These

40

THE PHEBUS FISSION PRODUCT PROJECT

differences also have a major effect on the vessel failure process and on the characteristics of the melt released into the containment upon vessel failure. 2. EXPERIMENTS AND INTERPRETATION ON CORE DEGRADATION AND MELT PROGRESSION A summary of the conditions for the integral tests that have been performed on core degradation and melt progression (including the TMI-2 accident), mostly taken from Hobbins, et. al., is given in Table 2(4). These tests have covered a wide range of conditions. They involve both open PWR core geometries and the compartmentalized BWR geometries that TABLE 2. SUMMARY OF CONDITIONS FOR INTEGRAL TESTS AND THE TMI-2 ACCIDENT Test/ (Accident) PBF SFD-ST SFD 1-1 SFD 1-3 SFD 1-4 ACRR DF-1 DF-2 DF-3 DF-4 PHEBUS SFD B9 B9R C3 AIC CORA 9 PWR Tests 3 BWR Tests NRU FLHT-1 FLHT-2 FLHT-4 FLHT-5 LOFT FP-2

Fuel Rods

Length (m)

Irradiation (Gwd/tU)

Control Matls.

Heating

System Pres. (MPa)

32 32 28 28

0.9 0.9 1.0 1.0

Trace Trace 36 36

None None None Ag-In-Cd

Fission Fission Fission Fission

6.9 6.8 6.8/4.7 6.95

9 9 8 14

0.5 0.5 0.5 0.5

Trace Trace Trace Trace

None None Ag-In-Cd B4C

Fission Fission Fission Fission

0.28 1.72 0.62 0.69 0.5, 3.5

21 21 21 21

0.8 0.8 0.8 0.8

Trace Trace Trace Trace

None None None None

Fission Fission Fission Fission

24, 52 18, 48

1.0 1.0

None None

Ag-In-Cd B4C

Electric Electric

0.2, 1.0 0.2

12 12 11 11

4.0 4.0 4.0 4.0

Trace Trace 1-30, 10-Trace 1-30, 10-Trace

None None None None

Fission Fission Fission Fission

1.38 1.38 1.38 1.38

100

1.7

0.45

Ag-In-Cd +H3BO3

Decay

1.1

36, 816

4.0

3

Ag-In-Cd +H3BO3

Decay

5–15

TMI-2

have zircaloy channel boxes and B4C control blades. They range in scale from 8 rods to 100 rods in the test fuel bundles and up to the 36,816 rod TMI-2 core, and from 0.5 meter length to full 4.0 meter core length. The fuel irradiation range includes fresh fuel, trace irradiated fuel, low burnup fuel, and high burnup (30 Gwd/tU) fuel. Fission heating, decay heating, and electrical heating of the test fuel rods have been used. Most of the tests have had boil off steam flow rates, with some tests at higher or lower flow. Both inconel and zircaloy grid spacers have been used. This table does not include the NSRR power excursion tests or the ACRR late-phase melt progression experiments that start from the late-phase debris bed geometry rather than from the initial intact fuel rod geometry. Also not included in the table are small in-reactor separate-effects tests on fission product and aerosol release, four STEP tests in TREAT and two somewhat similar ST tests in ACRR that had a reducing hydrogen environment. These tests all used 4-rod bundles of high burnup BR-3 fuel. The PBF SFD 1–3 and 1–4 tests also used high burnup BR-3 fuel. The NRU FLHT-4 and -5 tests included a single high burnup rod of commercial PWR fuel. In all of the tests for prototypic coolant boildown conditions, steam oxidation of the uncovered cladding with attendant hydrogen generation became significant at about 1500K and increased the heating rate above the initial 0.5 K/sec prototypic

CORE DEGRADATION AND FISSION PRODUCT RELEASE

41

of decay heat. Initially the local oxidation is rate limited by oxygen diffusion through the growing ZrO2 sheath which gives parabolic kinetics. This process is strongly temperature dependent. At about 1700K the rate of local oxidation heating increases rapidly to tens of K/sec, limited only by conversion of the entire steam flow into hydrogen (steam starvation) and by relocation of unoxidized molten zircaloy downward to cooler regions where it freezes and forms a partial metallic blockage. A burn front moves down the bundle during this process and leaves the hot upper part of the bundle in a reducing hydrogen environment. This has significance for fission-product release and transport and for core reflooding. This reducing environment in the upper part of the core is a transient effect that does not persist except for the general hydrogen build up in the primary system. In tests that have either stainless-steel clad-Ag-In-Cd PWR control rods in zircaloy guide tubes, stainless-steel clad BWR B4C control blade in the gaps between zircaloy BWR channel boxes, or inconel grid spacers, the control materials start to liquify and to attack and liquify adjacent zircaloy well below the melting points of the individual materials and well before the start of the rapid oxidation temperature transient. Intact core geometry in reactor accidents is lost from these eutectic interactions at temperatures well below the melting points of the individual materials. Figure 2 shows a chart of the threshold temperatures for liquid phase formation for the relevant materials in LWR accidents, both from melting and from materials interaction eutectics(5). For PWRs, Fe-Zr and Ni-Zr eutectics occur at about 1200K, but in integral core-geometry tests the first eutectic melts (and loss of core geometry) occur at about 1520K. For BWRs, B4C-Fe eutectics occur at about 1420K, and rapid liquifaction in integral tests occurs at about 1520K. In contrast, the melting point of as-received zircaloy 4 is 2030K and that of oxygen-stabilized alpha zirconium is 2250K. It has also been found in these integral tests that the molten metallic zircaloy dissolves a significant fraction of the solid UO2 fuel. This has potential significance for fission product release. Core degradation experiments in the CORA ex-reactor fuel-damage test facility, the videotapes of the processes involved, and the post test sectioning and examination of the test and bundles have provided the most detailed observations available on core degradation processes(6). Corollary work at KfK has also provided/basic information on the fundamental materials interactions involved in the core-degradation process and on their rate limitations(5). In the CORA experiments, electrically heated UO2/Zry-4 fuel rod simulators with absorber rods and spacers are subjected to temperatures up to 2700K in a steam environment. Most of the PWR tests have stainless-steel-clad Ag-In-Cd absorber rods within zircaloy guide tubes, and the BWR tests have B4C absorber rods in stainless-steel-clad blades in the gap between Zry channel box walls. A schematic drawing of the CORA experimental configuration is shown in Figure 3. The CORA tests are terminated by power reduction with continuing gas flow (slow cooldown) or by flooding with water (fast cooldown). Most of the information from the CORA tests is obtained by post-test sectioning and metallurgical examination of the test fuel bundles. Shown in Figure 4 are post-test horizontal and axial sections and a bundle photograph for the CORA-5 PWR test which included a PWR stainless-steel-clad Ag-In-Cd control rod inside a zircaloy guide tube. CORA-5 reached a maximum temperature of about 2300K and had a slow cooldown. The cross sections show the tungsten heater rods in the simulated fuel rods (a third of fuel rods do not contain heater rods but are radiation-heated by the adjacent rods), the control rod, the gray relocated metallic melt, and the blacker epoxy filler for sectioning. The general state of the damage, the missing control rod from the central and upper bundle, and the massive partial blockage from the relocated metallic melt are to be noted. At the end of 1990, two tests with alumina pellets and ten tests with UO2 had been performed, 7 with PWR and 3 with BWR fuel bundles. Two of the PWR tests and one BWR test were terminated by water flooding. In general, all the tests have demonstrated the competitive processes of cladding oxidation and low temperature melt formation caused by the interactions of zircaloy with the structural materials. After failure of the cladding in PWR bundles, the Ag-In-Cd and the inconel of the grid spacers liquefy cladding zircaloy at far below its melting temperature. The interaction between boron carbide and stainless steel causes liquefaction of the absorber blade in BWR bundles. The resulting melt penetrates the guide tube (PWR) or the zircaloy channel box wall (BWR) and initiates the liquefaction of the zircaloy cladding of the fuel rods. The liquid metallic zircaloy dissolves some of the UO2 even though it is only a constituent of the melt which is below the normal melting temperature of zircaloy. The major fraction of the metallic melt in the CORA tests relocated downward inside the bundle and froze at locations where the bundle temperature was below the melt freezing temperature to form a large partial blockage. The presence of a spacer grid favors blockage formation, but is not a necessary condition for blockage. In the water flooding tests, a large increase in the hydrogen generation occurred along with a large local transient temperature increase in the upper part of the bundle. This shows the influence of the strong zircaloy oxidation reaction from the reflood steam. A major branch point in the melt progression sequence, particularly for BWRs, involves whether a metallic core blockage develops, as occurred at TMI-2, or whether the metallic melt and later the ceramic melt drain from the core (and BWR core plate) when formed. The mechanism of vessel failure and the characteristics of the melt released into the containment upon vessel failure are strongly dependent upon which of these paths is followed. Accordingly, determination of the accident conditions, if any, for which core blockage does not occur is important. In U.S. BWR accidents in which automatic depressurization is actuated, blowdown lowers the water level below the core and core plate so that core heatup occurs in a

42

THE PHEBUS FISSION PRODUCT PROJECT

Fig. 3. CORA Experiment Configuration (PWR)

dry core at a very low steam flow rate. It has been hypothesized that in this case the metallic and later the ceramic melt drain from the core as formed onto the core plate and later into the lower plenum. Other early (metallic-melt) phase phenomena with potentially significant uncertainties are: the effects of the eutectic interactions of control-rod materials with zircaloy, particularly for BWRs, and of inconel grid spaces with zircaloy; the threshold mechanism (not just the melting point) for zircaloy melt relocation in fuel-rod geometry; oxidation and hydrogen generation from relocating and relocated metallic melts; and the effects of high burnup fuel. Research is currently underway in these areas. A most important current uncertainty regarding in-vessel melt progression is the failure threshold and the failure location of the debris-supporting metallic crust (and also the secondary ceramic crust) under attack by the growing molten ceramic pool in blocked core accident sequences. These determine the mass of ceramic melt released from the core into the vessel lower plenum and also the composition and the temperature (superheat) of the melt. Resolution of this uncertainty involves experiments and analytical modeling of the dynamics of debris bed melting and of crust relocation and failure. There are other late (ceramic-melt) phase phenomena that have potentially significant uncertainties. There is a question of whether the blocked core scenario may develop from the growth of the ceramic crust that surrounds the growing melt pool in the particulate mostly ceramic debris-bed in the core region regardless of whether a metallic blockage has previously formed. There are uncertainties regarding the natural circulation thermal hydraulics in the growing ceramic melt pool at prototypic Rayleigh numbers and with turbulent flow, and also uncertainties regarding the time constant for flow start up. There is uncertainty as to whether or not declad high burnup fuel naturally fragments into debris-bed geometry, whether the transition occurs by gravity collapse of an unstable array of free-standing declad fuel rods, or whether core reflooding as occurred at TMI-2 is necessary to produce true debris-bed geometry. There is also uncertainty regarding the rate of oxidation of the remaining metal in the debris bed and its supporting metallic crust (but not in the melt pool itself), although this rate cannot be large. A summary of the current state of phenomenological understanding of melt progression is given in Table 3. The table summarizes those phenomena that are reasonably well understood as well as those for which we have a more general understanding.

CORE DEGRADATION AND FISSION PRODUCT RELEASE

43

Fig. 4. Horizontal and Vertical Sections of the CORA 5 (PWR) Test Bundle

3. SUMMARY OF FISSION PRODUCT RELEASE IN INTEGRAL TESTS Information on in-vessel fission product release and aerosol generation has been obtained from the integral tests in PBF with both high-burnup fuel and with trace-irradiated fuel, from the full length FLHT-4 and 5 tests in NRU that included a high burnup commercial PWR fuel rod, from the LOFT FP-2 integral test with low-burnup fuel, from post-accident examination of the TMI-2 core which had low-burnup fuel, from in-reactor fission product and aerosol release experiments with high-burnup fuel in TREAT (the STEP tests) and ACRR (the ST tests), from ex-reactor single-rod integral experiments at Oak Ridge and at Grenoble, and from laboratory separate effects experiments at Whiteshell, Battelle, and other laboratories. This discussion will be limited to a summary of the principal results from the integral in-reactor tests that have been presented previously, primarily as given by Hobbins et. al.(4). A summary of the conditions for these integral tests, including the fuel burnup, is given in Table 2. High-burnup fuel from the BR-3 reactor that will be used in the Phebus FP tests has also been used in the PBF-SFD 1–3 and 1–4 tests, the four TREAT STEP tests, and the two ACRR ST tests in a reducing hydrogen environment.

44

THE PHEBUS FISSION PRODUCT PROJECT

Fig. 5. PBF SFD1–4 Noble Gas Fractional Release Rate and Clad Temperature

For these tests, the nominal burnup of the BR-3 fuel was 30 Gwd/tU. The fuel at TMI-2 had a burnup of 3 Gwd/tU and that in the LOFT FP-2 test had (0.45 Gwd/tU. A principal result of the PBF high-burnup fuel tests was that the release rates of the noble and volatile fission products was highly dependent on the changing fuel morphology during core degradation and melt progression, and that these rates were not a unique function of fuel temperature, as sometimes represented(4). This result is illustrated in Figure 5 which shows the fractional release rate of the noble gases and the calculated cladding temperature near the core center as a function of time for the PBF 1–4 test. The fractional release rate of the volatiles was nearly constant at about 0 1 % /sec for a period of about 30 minutes, although the mid-core temperature varied from 2200K to 3000K and back to 2200K during this period, and the rate then fell slowly as the fuel cooled further. This test had four PWR stainless-clad Ag-In-Cd control rods in zircaloy guide tubes in the 32 rod (total) test fuel bundle. The PBF results show that accurate prediction of the fission product release rate depends on knowledge of the state of degradation of the core and of the fuel itself. Potentially significant effects here include fuel cracking, fuel dissolution in molten zircaloy, and whether the environment is oxidizing or reducing. In the PBF tests and in the TMI-2 core examination, tellurium was found to be retained preferentially in the unoxidized zircaloy, and not released into the gas stream. This retained tellurium is then a potential source for later ex-vessel release by zirconium oxidation in melt-concrete and melt-coolant interactions. The TMI-2 core examination showed that from 3% to 10% of the cesium and iodine were retained in the molten ceramic melt in the core and the vessel lower head. Large fractions of the medium and low volatility fission products were also retained in the ceramic melt and in metallic inclusions in the melt. The cerium and strontium were retained in the ceramic melt as oxides, and a large fraction of the antimony, ruthenium, and tellurium were retained in nickel-based metallic inclusions in the ceramic melt. In the upper particulate mostly ceramic debris bed in TMI-2 that had not melted, the retention was about 20% for the volatile cesium and iodine, about 50% for the less volatile metals antimony and ruthenium, and nearly 100% for the low volatility oxides of strontium and cerium. The LOFT FP-2 test was performed with a large 100-rod test fuel bundle which was irradiated in the reactor to 0.45 Gwd/ tU burnup. The bundle also had PWR Ag-In-Cd control rods(7). The test was unique (except for TMI-2) in that the test fuel bundle was heated by fission-product decay. The test was terminated by reflooding with borated water. The principal addition to the PBF fission-product results from the LOFT FP-2 test was that while 3% of the volatile fission product inventory was released during the rapid oxidation transient to about 2200K when reflood was initiated, about 12% of the inventory was released during and after reflood. The reflood steam produced rapid local oxidation of unoxidized zircaloy in the upper part of the core. During the reflood transient, local regions reached UO2 melting (3100K), and this local heating during reflooding produced most of the hydrogen generation and most of the fission product release in the test. Two ST separate-effects experiments were performed in ACRR on fission product release from high burnup fuel in the local reducing hydrogen environment that results from steam starvation during the rapid oxidation transient(8). In these tests, large solid-state swelling of the fuel occurred that closed the cooling channels between the rods. These tests used a 4-rod fuel bundle that contained 15 cm sections of fresh fuel as a nuclear preheater followed by 15 cm sections of 30,000 MWd/tU BR-3 test fuel. The high-burnup sections were maintained for 20 minutes at about 2500K, which is well below the melting point of UO2 but above that of zircaloy. Similar large swelling of high burnup fuel without melting has been observed in the PIE of the steam-starved section of the NRU FLHT-2 test. This phenomena is qualitatively understood in terms of reduction of the UO2

CORE DEGRADATION AND FISSION PRODUCT RELEASE

45

to metallic uranium at the grain boundaries, liquefaction and fluidization of the UO2 grains, and fission gas pressurization to produce the fuel swelling. The range of this effect in accident parameter space and its significance for fission product release and melt progression itself are not currently understood. No strong effects of the reducing hydrogen environment upon the fission-product release rates were observed. In the ACRR DF experiments, a dense tin aerosol was observed when the zircaloy cladding melted. Table 3 Core Degradation and Melt Progression: Status of Current Understanding Reasonably Well Understood Phenomena in Early (Metallic Melt) Phase Clad ballooning Intact-core-geometry oxidation heating and hydrogen generation UO2 liquefaction (dissolution) by molten Zircaloy Eutectic material interactions and rates among UO2, ZRO2, Zry, and control materials and their oxides Opening up of the compartmentalized BWR core early in a BWR accident by the eutectic interaction of control-blade material with Zry channel box walls Molten Zry relocation is a noncoherent, noncoplanar, rivulent-flow process that does not block steam flow and hydrogen generation. It is not a film flow process General Understanding of Late (Ceramic Melt) Phase Based primarily on TMI-2 core examination Results also generally applicable to PWR unrecovered accidents Ceramic melt pool growth and meltthrough from block core Limited melt mass released from core Hydrogen Generation and Strong Heating of Uncovered Core from Zircaloy Oxidation by Reflood Steam (LOFT FP-2 and CORA)

4. ACKNOWLEDGEMENTS The authors are indebted to Drs. Richard R.Hobbins and David A.Petti for discussions and for information in the paper on the results on in-vessel fission product release in integral in-reactor tests that we gratefully acknowledge. REFERENCES 1 2 3 4 5 6

7 8

. NUREG-1150, “Severe Accident Risks: An assessment for Five U.S. Nuclear Power Plants,” (December, 1990). . J.M.Broughton, P.Kuan, D.A.Petti, and E.L.Tolman, “A Scenario of the Three Mile Island Unit 2 Accident,” Nucl. Tech. 87, p. 34 (August 1989). . R.W.Wright, “Melt Progression Modeling Implications of the TMI-2 Accident,” Proc. ICHMT Int. Seminar on ‘Fission Product Transport Processes in Reactor Accidents,’ Dubrovnik, Yugoslavia (May 22–28, 1989). . R.R.Hobbins, D.A.Petti, D.J.Osetek, and D.L.Hagrman, “Review of Experimental Results on LWR Core Melt Progression,” Nucl. Tech. (to be published 1991). . P.Hofmann, S.Hagen, G.Schanz, and A.Skokan, “Reactor Core Materials Interactions at Very High Temperatures,” Nucl. Tech., 87, p. 146 (August, 1989). . S.Hagen, P.Hofmann, G.Schanz, and L.Sepold, “Results of the CORA Experiments on Severe Fuel Damage With and Without Absorber Material,” Proc. 26th National Heat Transfer Conf., Philadelphia, August 6–9, 1989, AICHE Symposium Series 269. Vol. 85 (1989). . M.L.Carboneau, V.T.Berta, and M.S.Modro, “Experiment Analysis and Summary Report for the OECD LOFT Project Fission Product Experiment LP-FP-2," OECD LOFT-T-3806, OECD (June, 1989). . M.D.Allen, H.W.Stockman, K.O.Reil, J.W.Fisk, “Fission Product Release and Fuel Behavior of Irradiated Light Water Reactor Fuel Under Severe Accident Conditions: The ST-1 Experiment, NUREG/CR-5345, SAND 89–0308, (to be published 1991).

FISSION PRODUCT TRANSPORT J.O.LILJENZIN Institut för Kärnkemi, Chalmers Tekniska Högskola Göteborg, Sweden W. SCHÖCK Laboratorium für Aerosolphysik und Filtertechnik Kernforschungszentrum Karlsruhe, Karlsruhe, Germany

SUMMARY Fission product transport and retention in the reactor building system plays a central role in the assessment of source terms from severe accidents. Marviken-V, DEMON A and LACE were large scale experimental programs investigating aerosol behavior in the reactor cooling system and in the containment. Various thermal hydraulic conditions and different fission product simulants were used in the experiments. Large efforts were made to interpret experimental results and to compare them to computer code calculations. In this paper the experiments are described, main results are presented and discussed. The remaining open problems are stated and recommendations for the Phebus FP program are given. 1. INTRODUCTION In severe accident analysis for nuclear power plants fission product transport and behavior play a central role in the assessment of the radiological source term. This statement is illustrated by the evolution of risk figures which have been generated in the past 15 years of risk assessment studies. Major reductions have always been achieved when better FP transport and behavior models became available. With a few exceptions, as noble gases and some iodine compounds, the physical state of all important fission products airborne in the reactor cooling and containment system is condensed, which means that they are airborne in particulate form as an aerosol. Additionally this aerosol contains large amounts of non-radioactive fuel and other core or structural material released from different sources and at different times of the accident. Large efforts have been undertaken to develop and improve computer codes for calculating the transport and depletion behavior of aerosols from their origin to the containment boundaries. Also numerous experiments have been performed, ranging from investigations of elementary details of aerosol physics and chemistry to simulations of overall aerosol behavior in large scale integral tests. The computational tools and the experimental data bases have certainly been improved enormously. Comparing calculations with experiments, many questions have been answered, other new problems have been identified. In this paper the three large scale aerosol transport and behavior experiments Marviken-V, DEMONA and LACE, which were performed within the last ten years, will be described, the common results and the remaining open problems will be discussed. 2. PURPOSE OF LARGE SCALE EXPERIMENTS The main issue in the Marviken-V, DEMONA and LACE experiments [1] was aerosol behavior. At the time of these experiments the state of the art of this sub-discipline of general aerosol physics had been reviewed and documented repeatedly [2, 3, 4]. It was concluded that the relevant processes of nuclear aerosol behavior were well understood from an aerosol physical point of view. Also the mathematical description with numerical methods was successful and has been improved more and more in the course of the code development. The aerosol codes were believed to be adequate tools to calculate the time dependent behavior of particulate radioactivity under the conditions of relevant accident scenarios. It has to be noted, however, that most codes made, and still make, among others the following simplifying assumptions, more for economic than for physical or numeric reasons:

FISSION PRODUCT TRANSPORT

47

- Homogeneous spatial mixing of the aerosol is assumed, at least within one control volume. The boundaries of control volumes in risk assessment codes which have multi-compartment options should coincide with physical walls of real building compartments. - Shape factors and composition of particles are mostly modelled size independent. This assumes that internal mixing processes due to coagulation are fast compared to removal time constants. - Condensation on particles, which is one of the dominant processes in LWR nuclear aerosol behavior, requires thermodynamic input from a containment code. This input, temperatures, (partial) pressures and steam condensation rates, is mostly transferred off line from the thermodynamics code to the aerosol code assuming that no feedback of aerosol behavior on thermodynamics takes place. The validation of these assumptions goes beyond the capabilities of small and intermediate scale experiments and can only be done with large scale investigations. This situation, which was the starting point 8 years ago, is still true to a significant extent. Although much work has been done improving codes and developing integrated methods, a need can still be seen for even large scale investigations on isolated topics as well as on integral sequences. Some experiments have been directly triggered by past experience, looking into the details of specific phenomena, others are extending earlier investigations. Phebus FP [5] will be an experiment of so far unrivaled complexity, comprising most of the fission product transport phenomena at once and on top of it using real fission products from burnt fuel. The probability that new effects will be encountered is large. It is a great challenge to the Phebus FP team to be prepared to obtain as much information from the experiments as is feasible today, to be able to answer the questions that the experimental results will raise. 3. DESCRIPTION OF THE EXPERIMENTAL PROGRAMS In this chapter very brief descriptions of the three large scale experimental programs Marviken-V, DEMONA and LACE will be given. These programs have been presented separately and together in the past, we can omit many of the details here. The objectives, experimental design and results are given for the experiments individually, general issues will be discussed in the next chapter. 3.1. Marviken-V Aerosol Transport Tests The Marviken-V-ATT experiments were the fifth in a series of nuclear safety related projects using the abandoned nuclear power station at Marviken. The Marviken-V program was initiated in 1982 as an internationally financed project to provide aerosol transport and deposition data in a simulated full-scale water-cooled reactor primary circuit. The first test was conducted in May 1983 and the experiments were completed in March 1985. 3.1.1. Objectives Retention of radio-nuclides in the reactor coolant system (RCS) is a key factor in reducing source terms for severe accident situations since the behavior of fission products in the containment is determined by the physicochemical properties of aerosols leaving the RCS. The objectives of the Marviken-V tests were twofold: - To create a data base on the behavior of vaporized fissiumand corium- simulant materials in typical large scale reactor coolant system vessels. - To provide a large scale demonstration of the deposition of vaporized fissium- and corium-simulants in the reactor coolant system. The two objectives have quite different requirements on the type, completeness and accuracy of the data obtained. The first objective may be more important in the long run, but it is difficult to assess its final value.

48

THE PHEBUS FISSION PRODUCT PROJECT

3.1.2. Facility and instrumentation The major components of the Marviken-V test facility, as incorporated in the former Marviken nuclear power station, are shown in Fig. 1. The reactor test vessel had a volume of 140 m3. A number of measurement and sampling stations were located throughout the facility. The measurements provided data on thermal hydraulics, aerosol source strength, gas-borne aerosols, floor and wall deposition, and mass balance. Since the Marviken tests were conducted at much higher aerosol concentrations and temperatures and more vigorous thermal hydraulic conditions, the instrumentation had to be less sophisticated than that used in the DEMONA and LACE tests. 3.1.3. Test matrix Five tests were conducted in the Marviken-V program, see Table 1. Non-radioactive materials were used as fission product simulants (fissium) and reactor core structure simulants (corium). Tests 1, 2a, 2b and 7 studied the transport of fission products that might be released during a fuel damage process. Tests 1, 2a and 2b used only the portion of the facility downstream of the reactor vessel. Sequences with simultaneous fuel damage and structural aerosol release were studied in Test 4, which simulated the general geometry of a PWR primary circuit. 3.1.4. Results The Marviken-V experiments have provided the first data on the transport and deposition of high-concentration aerosols in a large-scale RCS. The aerosol mass concentrations ranged from 35 to 135 g/m3 and covered temperatures from 25°C to over 1200°C using superheated steam, condensing steam, and water. The results have been published in a series of reports [6–18]. The extent and quality of the data have also been evaluated [19]. Table 1: Marviken-V test matrix Test No Geometry 1 2a 2b 4 7

Features

Aerosol

Pressurizer Pipe Pressurizer Pipe Relief tank

High temperature Relief tank dry Low temperature Condensation in piping Water in relief tank Pressurizer Pipe Relief tank Medium temperature Condensation in piping Water in relief tank Reactor vessel PWR internals Pipe Pressurizer Relief High aerosol concentration Water in relief tank tank Reactor vessel PWR internals Pipe Pressurizer Relief Low aerosol concentration Water in relief tank tank

Fissium:

Fissium Fissium Fissium Fissium Corium Fissium

Corium:

Table 2: Marviken-V, aerosol concentration and retention Test No Concentration

[g/m3]

Retention [%] Reactor vessel Piping to pressurizer Pressurizer Piping to relief tank Relief tank Scrubber Final filter

1

2a

2b

4

7

35

62

51

132

52

na na 32 4 10 41 0.3

na na 14 1 85

na na 45 5 49

30 8 25 11 26

11 3 6 20 59

0.1

0.04

0.1

0.2

FISSION PRODUCT TRANSPORT

49

Fig. 1: Scheme of the Marviken V ATT facility, list and location of the aerosol instrumentation

As can be seen from Table 2, the retention in the RCS before the water scrubber or water filled relief tank ranged from 15% to 74%. Moreover, the data show the efficient removal of aerosols during passage through the water filled relief tank in Tests 2a, 2b, 4 and 7. 3.1.5. Lessons learned While the Marviken-V-ATT project met its overall objectives to produce large-scale data and demonstrate aerosol transport, a number of specific technical issues were not resolved to the desired extent. Schedule, budget, facility and measurement restrictions limited the fulfillment of all the objectives. It is important to realize that all the tests have been carried out under different conditions. The originally envisaged need for repetition of experiments to verify reproducibility of the data base was not fulfilled. Thus we have a fair amount of data, but with little information on their reproducibility. In the case of particle size distributions rather large uncertainties remain. However, all raw data from the cyclones are presented in the test reports, and future investigations of cyclone and sampling line behavior may resolve this issue.

50

THE PHEBUS FISSION PRODUCT PROJECT

Fig 2: Phase diagram and measured isobars in the CsOH-H2O system

Another type of uncertainty also exists. That is “how did flow pattern affect the aerosol transport and deposition?”. This and other issues have been partly addressed in the Marviken Intermediate Program, but measuring flow patterns is difficult even with a very large effort. Moreover, it is not clear if the computer codes could use these data if they were available. From these observations we should learn that duplication of experiments is very important as it greatly increases the confidence in the results. Furthermore, repetition offers another chance to correct failing instrumentation or modify measurement techniques. The ideal situation would be to perform post-test calculations before repetition, as this could point out to the experimentalists where more detailed, more accurate or new data were needed. One of the surprises in the Marviken-V program was the observation that the presence of large amounts of CsI and CsOH converted the aerosol particles to liquid droplets even in superheated steam. In fact, small scale experiments have shown later that there is no solid phase at all in the CsOH-water system above room temperature at steam partial pressures above 123 mbar [20, 21, 22]. This is illustrated in Fig. 2 which shows the solid-liquid phase boundary as function of temperature and water mole fraction in the CsOH-H2O-system. Additionally, measured isobars are shown for some constant steam partial pressures. The 123 mbar isobar shows that at partial pressures higher than 123 mbar there is a transition from dissolved CsOH to molten CsOH but no formation of a solid phase. Hence, there will always be some liquid present in the aerosol “particles” during LWR accidents.

3.2. DEMONA DEMONA (Demonstration Experiment for Modelling of Nuclear Aerosol behavior) was a large scale experimental program aimed at demonstrating the natural aerosol removal from the containment atmosphere under core meltdown conditions in a pressurized-water reactor. It was the final step in a series of smaller scale experiments performed in support of the NAUA code development. The experiments were performed with international participation from 1983 to 1986. 3.2.1. Objectives The objectives of DEMONA [23] were to demonstrate the validity of the calculations with the NAUA aerosol behavior code [24] and to verify the thermodynamics codes COCMEL [25] and FTPLOC [26] with which input data for the NAUA code are calculated.

FISSION PRODUCT TRANSPORT

51

3.2.2. Facility and instrumentation The experiments were done in a 640 m3 concrete model containment facility, which could be operated in the correct temperature and pressure range. Non-radioactive metal and metal oxide aerosols were produced using three commercial plasma torches as heat sources [27]. Concentrations in the range of 10 g/m3 could be achieved. The aerosol measurements were done using a variety of different instruments ranging from sampling to on-line optical techniques [28, 29, 30]. Fig. 3 shows a section of the facility and the locations of the main aerosol measuring instruments inside the containment. Additional sampling and monitoring instruments were used outside the containment, including one continuous and one discontinuous dilution and drying system. One of the outstanding features of DEMONA was that one third of the total budget was used for aerosol measurement purposes. 3.2.3. Test matrix Table 3 gives an overview of all DEMONA experiments. The DEMONA test matrix consisted of nine experiments, a reference experiment B3 and six variations of aerosol material, thermodynamic conditions and containment geometry. Two special tests were conducted at the beginning of the program. A1 was a thermodynamic test of the containment without aerosol. B2 was a dry aerosol experiment without steam to show the difference in aerosol behavior between non-condensing and condensing atmospheric conditions. 3.2.4. Results Detailed test reports are available for every individual test [31]. General results of the experiments have been published in annual reports [32] which also contain a complete list of publications of the program. Some reviews have been published later, e.g. [33, 34, 1]. Individual experiments of the test matrix were aimed at different features in aerosol behavior. The main results were the following: Natural aerosol removal was demonstrated to occur. With the NAUA code the reduction of airborne mass over four orders of magnitude in 8 hours was predicted correctly. The reproduction of experiment B3, thus the reliability of the experimental procedures, was shown with B4. In comparison to experiment B2 the dominating influence of processes related to steam condensation was confirmed. The difference in aerosol behavior between dry and condensing conditions was shown also in one single experiment (B6), any accidental differences between experiments were ruled out. It was shown that different insoluble aerosol species did not behave principally different, the influence parameters are known. Furthermore, the properties of mixed aerosols with high concentration Table 3: DEMONA test matrix No

Aerosol material max. conc. Description [g/m3]

A1 B2 B3 B4 B5 B6 A7 A8 A9

SnO2 SnO2 SnO2 SnO2 Ag +MgO Fe2O3 Fe2O3 +SnO2 Fe2O3 +SnO2

12 8 10 12 2 3 5 7

Source term ratio Th./Exp.

thermodynamic test of the model containment dry experiment without condensation base test base test, reproduced test with delayed onset of condensation test with low aerosol concentration and delayed onset of condensation base test, repetition with iron oxide aerosol test with mixed aerosol multi-compartment test with mixed aerosol and delayed onset of condensation

3.6 2.6 2.8 2.7 3.0 1.7 2.3 2.4

followed the coagulation concept in NAUA. Coagulation time constants are small compared to removal time constants. Spatial inhomogeneities did occur, especially in experiment A9. They were too small, however, to influence the overall removal behavior to such an extent that a real multi-compartment treatment would have been necessary. On the other hand,

52

THE PHEBUS FISSION PRODUCT PROJECT

Fig 3: Schematic axial section of the DEMONA containment model showing aerosol instrumentation. Not shown are 3 rainout samplers and instruments outside the containment.

thermalhydraulics appeared to be more sensitive to spatial and temporal fluctuations than aerosols, which reflects the different time constants involved. 3.2.5 Lessons learned Although much efforts were undertaken, the aerosol generation mass balance of material comsumed by the generators to aerosol arriving airborne in the containment could not be closed completely. The question where the lacking mass disappeared remains open, since the retention in the aerosol feed line was not investigated in DEMONA. Some of the more sophisticated aerosol measurements failed more often than expected. The harsh environment in the containment was a problem for the optical particle size spectrometer and the inertial spectrometers. The liquid airborne water calorimeters worked well after initial problems had been overcome. The calorimeters and the optical spectrometer were also used later in some of the LACE tests. Since only insoluble, although different, aerosol materials were used, no significant differences in aerosol behavior due to aerosol material were observed.

FISSION PRODUCT TRANSPORT

53

A clear response of thermal hydraulic data to local (cold spots) and temporal (steam injection) events was noticed. Most sensitive to such events was the bulk volume condensation rate which is the dominant influence parameter for aerosol depletion rates. All significant differences in aerosol behavior between experiments were caused by thermal hydraulics, i.e. transport phenomena. This was seen in experiment A9, the multi compartment experiment, which showed a clear time dependent correspondence between changing convection patterns and the immediate response in aerosol concentration. 3.3. LACE The LACE (LWR Aerosol Containment Experiments) program was organized by EPRI and was sponsored by an international consortium. The accident situations considered were those for which high consequences were calculated in risk assessment studies. 3.3.1. Objectives The LACE program aimed at a closing of the gap perceived to exist between the results from the Marviken-V and DEMONA projects and to expand the data-base for some specific high consequence accident scenarios. A specific goal of the LACE program was to determine the ability and accuracy of the various computer codes used to analyze and predict aerosol behavior. As a result, a computer code comparison effort with pre and post test calculations was an integral part of the overall LACE program. 3.3.2. Facility and instrumentation The LACE experiments were performed using the Containment Systems Test Facility at the Hanford Engineering Development Laboratory. The arrangement of aerosol generator, pipes and instrumentation varied with the experiments. Fig. 4 shows the setup used for the containment test LA4. As the LACE program was carried out after most of the Marviken-V and DEMONA tests were finished, its measuring program could benefit from previous experience and instrumentation development [35]. The thermodynamic measurements comprised gas, wall and sump temperatures, pressure, temperature and flow rate of the gas, steam and aerosol fed into the tank and released from the tank, temperature and composition of the atmosphere, local flow rates in the tank, heat transfer coefficients and wall condensation rates. The aerosol measurements were made by many different instruments. Cascade cyclones, impactors, filter clusters and deposition trays were used for aerosol sampling. Additional instrumentation was used outside the tank to measure the solid fraction of the aerosol. The liquid (water) fraction of the aerosol was measured with two calorimeters, giving the mass concentration of airborne water, and with an optical particle counter, giving the droplet size distribution. Three photometers and an optical particle size spectrometer were used inside the tank to monitor the spatial and size distribution of the aerosol. 3.3.3. Test matrix The LACE program consisted of three main tasks (cf. Table 4): large scale tests to investigate aerosol retention behavior for selected high-consequence containment aerosol and thermal-hydraulic sequences, a computer code validation program, and a support program to provide the additional help needed to perform or interpret the large-scale experiments. 3.3.4. Results The LACE program considerably improved our ability to assess aerosol behavior under transient conditions and showed that even relatively simple leak paths may remove a large fraction of suspended material by turbulent deposition, see Table 5. Detailed results have been published as a series of LACE reports, see references [36–44]. Some observations on specific phenomena will be further discussed in the following chapters.

54

THE PHEBUS FISSION PRODUCT PROJECT

Fig 4: Schematic of the LACE test facility with aerosol instrumentation Table 4: LACE program test matrix No

Aerosol material

Max conc [g/m3]

Description

CB1 CB2 CB3 LA1 LA2 LA3 LA4 LA5 LA6

NaOH NaOH+Al(OH)3 Al(OH)3 CsOH+MnO CsOH+MnO CsOH+MnO CsOH+MnO none CsOH+MnO

10 3 6 4 4 7–30 5 na 2

Containment bypass pretest, soluble aerosol Containment bypass pretest, mixed aerosol Containment bypass pretest, insoluble aerosol Containment bypass (piping and auxiliary building) Failure to isolate containment Containment bypass, pipe flow only, three tests: LA3A, LA3B,LA3C Late containment failure with overlapping aerosol injection periods Rapid depressurization with spiked pool Rapid depressurization with aerosol injection

Table 5: Aerosol retention in the LACE tests Test No

CB1

CB2

CB3

LA1

LA2

LA3

Retention [%] 63 mm pipe 200 mm pipe 300 mm pipe Containment vessel Auxiliary building, Aux. building ventilation Upper leak Lower leak

3 na 55 na 40 2 na na

5 na 48 na 45 2 na na

1 na 14 na 66 19 na na

>98 na na na <1 <1 na na

na 2 na 62 na na 19 17

>70 na na na na na na na

3.3.5. Lessons learned The importance of water soluble and/or hygroscopic materials on aerosol behavior was confirmed.

FISSION PRODUCT TRANSPORT

55

The tests showed that passage of aerosol material through a pipe changes the aerosol characteristics in such a way that subsequent depletion processes are enhanced. It also demonstrated the large influence of the chemical/physical nature of the aerosol material, i. e. liquid (water soluble), liquid/solid (partly water soluble) or solid (insoluble in water), on the behavior of an aerosol before and after deposition. Whereas 100% of a solid aerosol was vented from the containment building to the environment, less than 2% of a liquid or liquid/solid aerosol escaped to the environment. Finally, it was shown that water can leach the soluble components from fission product deposits, leading to relocation. 4. COMMON RESULTS OF THE EXPERIMENTS It must be remembered, when comparing the general results, that each program used a different set of conditions, and hence the various phenomena assumed different importance in each test. In some cases a major phenomenon observed in one experiment could thus barely be noted or not at all observed in another test. The fact that a phenomenon was only observed in one or two experiments does not necessarily imply that its existence was contradicted by the other experiments. 4.1. Basic phenomena Before the Marviken-V experiments, thermophoretic and diffusiophoretic deposition mechanisms were judged to be important in the RCS. However, the test results showed that gravitational settling and inertial impaction were in fact the dominating deposition mechanisms. In the containment enhanced settling and diffusiophoresis are the dominating processes of aerosol removal in condensing atmospheres. The dominating influence of steam condensation can be seen easily by comparing the removal rate of DEMONA experiment B2, the dry experiment, to the removal rate of any other experiment. In experiment B6 the two different removal rates have also been observed in one single experiment. Gravitational settling is further enhanced by the effect of water-soluble aerosol material [45]. In Marviken-V the use of CsOH, CsI and Te as inactive fissium simulants led to the unexpected formation of droplets instead of solid aerosol particles, even from superheated steam. In hindsight, this was of course the predictable behavior of many water-soluble materials. The liquid nature of aerosol particles and deposits had several effects on the deposition and on the deposited material, e.g. run-off of deposits from some vertical walls, adsorption of water in some samples that came in contact with ambient air after the experiment, clogging of filters for aerosol sampling or passage of liquid through aerosol filters. 4.2. Time dependent aerosol composition In Marviken-V different types of aerosol particles were found in some deposits, e.g. white and black aerosol particles in nearby deposits. Chemical analysis of several samples also indicated that composition varied with particle size. Similar observations were also made in some of the LACE tests. This can most easily be understood if the aerosol particles had had a composition and size which changed with time. Particle composition as a function of size could result from e.g. the temperature and humidity history [46]. Modeling of particles with a size dependent composition as a homogeneous aerosol could be insufficient in some cases. In DEMONA experiments B2 through B5 tin oxide was used as the aerosol material. In experiment B6 the aerosol was composed of a mixture of silver and magnesium oxide particles. In experiments A7 iron oxide was used, and experiments A8 and A9 were done with mixtures of iron and tin oxides. In experiment A8 the components were not generated at the same time. This variety of aerosols constitutes a good basis for examining the effect of particle composition on the overall aerosol behavior in the containment. The composition of the mixed aerosol particles in experiments B6, A8 and A9 was found to be independent of the particle size. In A8 where the generation of the species was not simultaneous it was also found that the composition followed the mass balance of generated aerosol with a time constant that was short compared to the overall removal time constants. The behavior of the mixed aerosol in LACE test LA4 will be discussed in the next chapter.

56

THE PHEBUS FISSION PRODUCT PROJECT

4.3. Transport phenomena One particular objective of DEMONA was to check the applicability of the single well mixed volume concept which is underlying in both the NAUA and COCMEL codes. Spatial inhomogeneities have been observed in some DEMONA experiments, but only during the aerosol generation interval. As a final test case experiment A9 was scheduled as an ‘experiment with complex geometry’. Large spatial differences were seen during aerosol injection and during the stagnant dry phase. As soon as the steam injection was resumed, however, spatial differences disappeared quickly. This finding would remain unexplained without the results of the multi- compartment FIPLOC calculation, which show that even with the partially obstructed interconnections between compartments large convection loops developed and linked all compartments together. The flow patterns that developed explain the relatively good mixing of aerosol throughout the containment. Experiment A9 showed that aerosol behavior calculations with the single volume version of NAUA may be applied also to certain multicompartment situations. More importantly, the experiment showed that this finding can be explained on the basis of a more detailed thermal hydraulic investigation. Many issues related to the transport and deposition of fission products and structural materials in LWR primary systems have been illustrated by the experience obtained from the Marviken-V-ATT project. We will concentrate on one of these which has some bearing on the proposed Phebus FP experiments. The complex flow pattern in the vessels, especially near the feed point, presumably led to trapping and recirculation of aerosol particles back to the injection point, cf. Fig. 5. This could have led to excessive particle growth. However, none of the tested aerosol codes could make full use of the in-vessel flow-patterns estimated in the Marviken Intermediate Programs. It is thus important to consider the possibility of such phenomena also in the Phebus FP tests and to provide the instrumentation necessary to uniquely define the flow-field in all vessels and larger pipes. The LACE program confirmed the observation about liquid deposits made in the Marviken-V program. It also showed that condensing steam can leach soluble material from deposits or wash away deposits by condensing water. 4.4. Chemical reactions The DEMONA experiments were performed with chemically unreactive materials and no effects of chemical reactions were observed. A similar conclusion was reached from some of the LACE tests. On the contrary, Marviken-V showed that chemical reactions do occur and do influence the aerosol behavior. This was confirmed in a few of the LACE test where reaction products, e.g. between Mn and CsOH, were observed. Furthermore, the LACE CB tests confirmed the Marviken-V observation about formation of a liquid aerosol from soluble material. In the test with a large fraction of insoluble corium simulants (Ag and Mn) solid deposits were found. However, microscopic examination of the collected particles showed that these sometimes had signs of chemical reactions between the constituents. The formation of minor amounts of carbon dioxide in the graphite mixing chamber also seems to have converted some of the CsOH to CsHCO3, thus changing its physico-chemical behavior. 5. COMPARISON TO CODE CALCULATIONS Comparison of experimental results with code calculations is one of the most important issues with large scale experiments. In the Marviken-V, DEMONA and LACE programs most code efforts have been directed towards aerosol behavior and thermal hydraulics, coupling between the two was done less frequently, very few calculations have been made on aerosol formations and chemistry issues. In this chapter we will give examples of successful comparisons (aerosol in the containment) and of cases where initial problems have been resolved (mixed aerosols). In DEMONA the result of all pre-test and post-test calculations was an overprediction of the aerosol mass concentration. The peak concentration was overestimated by up to a factor of two, thereafter the calculated and measured values showed a more or less pronounced divergence. The overshooting of the calculation in the peak concentration region could be explained by turbulent depletion effects which are not modelled in NAUA and which take place as long as the aerosol generators are operating. The diverging slopes of the calculated and measured mass concentration curves in the long term regime are mainly due to a mismatch of calculated and real condensation rates. In some cases the calculated condensation rates were far too low to explain the droplet concentrations measured with the droplet calorimeters. NAUA calculations using condensation rates corrected with the calorimeter data show a much better agreement with the experimental data than the calculations with direct

FISSION PRODUCT TRANSPORT

57

Fig 5: Measured recirculating flow patterns in the Marviken V reactor vessel

input from COCMEL. The single volume approach, although it worked well for the aerosol calculations with NAUA, appears not to be sufficient for calculating the sensitive condensation rates with a thermodynamic code. Condensation on particles obviously is more space and time dependent than the mechanics of aerosol behavior. The problem was recognized, improvements of thermalhydraulics modelling is ongoing. For source term assessment, the quantity which really matters is the accumulated leaked mass. In DEMONA it could be evaluated from the measured mass concentration by time integration and compared to NAUA results. In the last column of Table 3 the ratios of calculated to measured values of the accumulated leaked mass values for all DEMONA experiments are shown. It is seen that all calculated values are conservative, but by no more than a factor of three in the worst case. The only exception is experiment B2, the dry experiment, here the slower removal rate caused a larger time integral although the difference between calculated and measured mass concentration values were smaller than in the other experiments. In an international code comparison organized by the Commission of the European Communities eleven participants from seven countries have performed post-test calculations for experiment B3 [47]. The main outcome was that nine codes calculated practically identical results, the deviations among the calculations were a factor of two for the accumulated leaked mass (Fig. 6) which is smaller than the difference to the experimental values. This illuminates the uniform degree of development in international aerosol codes. Since mixed aerosols will be present in all of the Phebus FP experiments, we will give two examples of mixed aerosol behavior which, on the first attempt, led to opposite types of misinterpretation. The first example is one where a mixed aerosol experiment helped to discover an unnoticed experimental artefact which had caused large problems in code comparison. In the first DEMONA experiments the observed long term aerosol mass concentrations were much higher than the calculated ones. Not earlier than in experiment B6 it was discovered that a considerable amount of rust particles was delivered by the steam generator, thus constituting a continuous aerosol source. These particles dominated the long term airborne concentration after the original aerosol was depleted. As long as only total mass concentration measurements were made this source was not noticed. When in experiment B6 (mixed aerosol of Ag and SnO2) the aerosol samples were chemically analyzed the additional fraction of iron oxide was found. The removal of the Ag and SnO2 fractions was found to be in agreement with the calculations.

58

THE PHEBUS FISSION PRODUCT PROJECT

Fig 6: Accumulated leaked mass from DEMON A test B3, comparison of experimental values with 11 code calculations

Fig 7: Measured mass concentration of aerosol components in LACE LA4

The second mixed aerosol example is one where a mixed aerosol experiment had been misinterpreted as long as a detailed code calculation was missing. In LACE test LA4 the airborne mass concentrations of the aerosol constituents CsOH and MnO diverged after some hours as shown in Fig. 7, MnO showing higher concentrations than CsOH. The existence of a ’persistent fraction’ of small MnO particles was immediately conjectured. In fact there is nothing special at all in this behavior, the curves are in perfect agreement with NAUA calculations and can be explained by plain aerosol mechanics [48]. The MnO fraction is not ‘persisting’ more than normally, MnO aerosol by itself would behave exactly the same. The difference with this mixed aerosol is that the CsOH is removed faster than in cases when it were alone. This is the effect of simultaneous coagulation and depletion of an aerosol consisting of a fine (MnO) and coarse (CsOH) fraction. 6. UNRESOLVED PROBLEMS Deficiencies in nuclear aerosol research in most cases originate from mismatches between experiment and model computation, which means that they become obvious only at the time of comparison, i.e. after the experiment. So the first and most important deficiency in almost all of the large scale experimental programs so far was a lack of funding for repetition of experiments after they had been evaluated and the missing issues identified. The two examples at the end of the preceding

FISSION PRODUCT TRANSPORT

59

chapter are of that kind. A repetition of the experiment to crosscheck the explanations given after the first attempt would have been extremely valuable. Coupling of thermal hydraulics to aerosol behavior is still to be improved considerably. DEMONA A9 showed that a potential exists for explaining observations that contradicted the expectations by looking into the details of flow patterns in structured buildings. The importance of thermal hydraulics coupling in the RCS needs no explanation. Some aerosol mechanical properties may be changed by processes other than mechanical and such processes have to be incorporated into the models. As an example, the behavior of mixed aerosols of both soluble and insoluble components in steam atmosphere is not calculated appropriately up to date. Further, a growing interest in chemical changes was triggered by observations in some experiments. Many chemical reactions have been investigated but no use is made of the results in the aerosol codes. First attempts to recognize the influence of chemistry on aerosol behavior are made off line as in the coupling of IMPAIR [49] and NAUA, or on line as in CHMAAP [20], but much more work needs to be done. In a few, but very important, cases an unability was noticed to bring the requirements of the experimenter and the modeller together. Elegant theoretical formulations may not be suitable for experimental implementation, and a more crude but successful method may be favored. An example is the widely used concept of boundary layers for some mechanisms of particle deposition (thermophoretic, diffusiophoretic). Although theoretically sound this concept is useless in large scale experiments because boundary layers of millimeter thickness can never be measured. On the other hand, a formulation using total fluxes through the boundary layer without needing to know its thickness is successful because these fluxes (steam or heat) can be measured rather accurately. Similar considerations apply to supersaturation as the driving force of condensation, nobody will ever be able to measure supersaturation in a nuclear aerosol experiment. Here, however, the viable and convincing concept is still missing. 7. RECOMMENDATIONS TO PHEBUS FP From the preceding chapters the following general situation should have become evident. Fission product aerosol behavior is well understood and can be reasonably predicted in the containment, in the RCS there are many problems which have not been completely resolved. The reason for this is very simple and unavoidable at once, it is related to the basic properties of aerosol mechanics. In a sufficiently concentrated, i.e. coagulating, aerosol in a closed vessel the long term aerosol concentration and size distribution does not depend on its initial state. It has been stated that long term aerosol behavior is ‘forgiving’, which is literally true because some of the errors or deficiencies in experiments and in codes will not show up at the end. In the RCS, however, there never is time for self regulating effects to take place. Aerosol behavior and depletion depends on particle properties and their changes at every moment. It is therefore mandatory to know, i.e. to measure, particle concentration, size, shape, density and composition at every time. Many of the problems that we have in interpreting RCS experiments could be resolved if more of these data would have been measured. The Phebus FP test matrix represents this situation correctly becaus it places the emphasis on ‘medium residence time scenarios’, where the major effects can be expected. Additionally, the use of real burnt fuel introduces a new dimension of complexity, namely the relation between mass and radioactivity. Aerosol depletion is governed by mechanical properties, nonradioactive components are abundant and interactions with radioactive properties are very scarce. The time dependent composition of aerosols and the distribution of radioactive species in it have to be measured in order to understand transport of radioactivity in the experiments. More specifically: - Gas composition is very important for the development and behavior of particles. CO2 in the atmosphere e.g. may convert some of the liquid droplets into dry particles. - Relative humidity and liquid airborne water content have to be measured because they dominate aerosol sizes, and consequently depletion, by formation of droplets. - Particle size measurements should be done frequently and at many locations. When done with impactors only part of the information is obtained when droplets are sampled. The water fraction, which may dominate the removal rates, is lost. - Deposits can be assumed to be hygroscopic and chemically reactive. Hence, they must be kept in the same atmosphere and at the same temperature up to the time of analysis. - Run-off, leach-out and wash-down phenomena have to be expected, sampling has to take this into consideration. - The complex gamma spectra can only be resolved after radiochemical separation [50]. Further, some important fission products are pure beta emitters (e.g. 90Sr/90Y) and cannot be measured by gamma spectroscopy.

60

THE PHEBUS FISSION PRODUCT PROJECT

Comparing Phebus FP to Marviken-V, DEMONA and LACE the authors believe that the Phebus tests are more complex than the earlier experiments, but that the aerosol instrumentation is not increased to a similar degree over that used in the earlier experiments. Many of the problems expected in Phebus FP are certainly properly recognized, but fission product transport and depletion is dominated by the aerosols. Hence, the time dependent properties (gas phase and particles) have to be measured to such an extent that it is possible afterwards to evaluate and understand the results and to compare them correctly to results from calculations. This leads us to recommend an increased effort to measure all relevant parameters, for inactive as well as for radioactive substances. The extra investment may be very worth while. REFERENCES [1] [2] [3] [4] [5] [6] [7]

J.O.Liljenzin, J.Collén, W.Schöck, F.Rahn, Report from the Marviken-V / DEMONA / LACE workshop, CSNI Workshop on aerosol behavior and thermal-hydraulics in the containment, 26–28 November 1990, Fontenay-aux-Roses, France OECD Nuclear Energy Agency Ed., Nuclear Aerosols in Reactor Safety, Paris, OECD 1979; CSNI NEA, SOAR No.1 OECD Nuclear Energy Agency Ed., Nuclear Aerosols in Reactor Safety, Paris, OECD 1985; CSNI NEA, Supplementary Report W.O.Schikarski, W.Schöck, Eds., Proceedings of the CSNI Specialists Meeting on Nuclear Aerosols in Reactor Safety, Karlsruhe, September 4–6, 1984, KfK 3800, CSNI 95 (1985) P.Delchambre, P.von der Hardt, The Phebus FP project, EUR 12195 EN, 1989 E.Jansson and D.Mecham, Report from working group 3: Definition of accident sequences and test conditions for the Marviken Aerosol Transport Tests, MX5–8, May 1982. The Marviken Large-Scale Aerosol Transport Tests, Report Series, Studsvik Energiteknik AB

Test facility, MXE-101, Aug. 1985. Measurement system, MXE-102, Sep. 1985. Data Acquisition and Computer Processing, MXE-103, Aug. 1985. Data Accuracy, MXE-104, Nov. 1985. Evaluation methods, MXE-105, Sep. 1985. Sample Handling and Chemical Analysis, MXE-106, Nov. 1985. Test 1 Results, MXE-201, Oct. 1985. Test 2a Results, MXE-202a, Oct. 1985. Test 2b Results, MXE-202b, Oct. 1985. Test 4 Results, MXE-204, Nov. 1985. Test 7 Results, MXE-207, Nov. 1985. Summary report, MXE-301, Nov. 1985. Report Abstracts, MXE-401, Dec. 1985. Conclusions, MXE-402, Dec. 1985. [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22]

A.L., Nichols, D.A.Williams, Report on the Second Marviken-V Analysis Specialists’ Meeting, 15–17 July 1986 at the Argonne National Laboratory, AEEW-M 2486, AEE Winfrith, Aug. 1986 J.Collén, H.Akesson, Marviken intermediate program (MX-IP) for the 3rd and 4th quarter of 1985. MXIP-1001, Studsvik Energiteknik AB, June, 1985. Corium shakedown test, MXIP-200S, Studsvik Energiteknik AB, August 1986. A.Magnusson, Chemical effects—results from various investigations, MXIP-8, Studsvik Energiteknik AB, February 1986. A.Hautojärvi, L.Ström, Particle size evaluation and analysis, MXIP-10, Studsvik Energiteknik AB, February, 1986. L.Ström, Resuspension of dust deposits in pipes, MXIP-11, Studsvik Energiteknik AB, February 1986. The Marviken Intermediate Programs—Support for the Evaluation of the MX-V-ATT Experiments, Summary report MXIP-301, Studsvik Energiteknik AB, July 1986. D.Mecham and H.Unneberg, Minutes from the third TAC meeting, 1983–10–25–1983–10–26, MX5–54 (1983). A.T.D.Butland, P.A.Williams and N.A.Johns, Report on the Marviken-V Analysis Specialists Meeting, 26–28 June 1985, AEE Winfrith, AEEW-M 2263 (1985). K.Johansson, Travel Report from Marviken V Analysis Meeting, Winfrith June 26–28, 1985 (In Swedish) Studsvik, NR-85/68 (1985). H.Häggblom, Marviken V Analysis Meeting in Winfrith 26–28 June, 1985 Studsvik, NR-85/71 (1985). H.Häggblom, Conclusions from the Marviken Aerosol Transport Tests and their Analyses, NP 88/54, Studsvik Nuclear Technical Note, Aug. 1988. J.O.Liljenzin (Ed.), The Influence of Chemistry on Core Melt Accidents, Final Report of the NKA Project AKTI-150, NORD 1990: 56, ISBN 87 7303 452 5, Malmö, Sept. 1990. J.O.Liljenzin, Release and Deposition of Fission Products and Actinides, RAMA-III 89–05, ISBN 91–7010–119-1, Stockholm, 1990. L.G.Johansson, S.Kazikowski, An Experimental Study of Equilibrium Water Vapour Pressures in the CsOH-H2O System, The 3rd Chemical Congress of North America, Nuclear Reactor Severe Accident Symposium, Held in Toronto, Canada, June 5–10, 1988.

FISSION PRODUCT TRANSPORT

[23] [24] [25] [26]

[27] [28]

[29] [30] [31]

61

W.O.Schikarski Ed., DEMON A Forschungsprogramm zur Demonstration nuklearen Aerosolverhaltens—Grundlagen, Ziele, Auslegung, KfK 3636, EIR 502 (1983) H.Bunz, M.Koyro, W.Schöck, NAUA Mod5 und NAUA Mod5-M, Zwei Computerprogramme zur Berechnung des Aerosolverhaltens im Containmentsystem eines LWR nach einem Kernschmelzunfall, KfK 4278, (1987) K.Hassmann, M.Reimann, Analysis of hypothetical core melt down accidents, ENS/ANS Topical Meeting on Reactor Safety Aspects of Fuel Behavior, Sun Valley ID, August 1981 G.Weber, H.Jahn, Calculations on DEMONA experiments with the integrated multi-compartment code FIPLOC, Source Term Evaluation for Accident Conditions, Proceedings of an International Symposium, Columbus OH, Oct. 28–Nov. 1, 1985, IAEA Vienna 1986, pp 549–558 H.Ruhmann, M.Peehs, A high efficiency aerosol generator for DEMONA, Aerosols— Formation and reactivity, Proceedings of the Second International Aerosol Conference, Berlin September 22–26, 1986, Pergamon, Oxford 1986, pp 1070–1073 G.Friedrich, A.Fromentin, O.Mercier, R.Taubenberger, W.Schöck, Aerosol measurement system for the DEMONA experiment, Proceedings of the CSNI Specialists Meeting on Nuclear Aerosols in Reactor Safety, Karlsruhe, September 4–6,1984, KfK 3800, CSNI 95 (1985), pp. 329–338 A.Fromentin et al., DEMONA: Etude du comportement d’aerosols dans un confinement, Journal of Aerosol Science, 19 (1988), pp 65–72 W.Schöck, V.Vrangos, Calorimetric Measurement of Water Fog Concentration, J. Aerosol Sci. 17 (1986), pp. 525–529 DEMONA-Versuche, Technischer Fachbericht, Battelle-Institut Frankfurt

Versuch A1, BIeV-R65.523–30–1, Apr. 1985 Versuch B2, BIeV-R65.523–30–2, Nov. 1985 Versuch B3, BIeV-R65.523–30–3, Apr. 1985 Versuch B4, BIeV-R65.523–30–4, Mar. 1985 Versuch B5, BIeV-R65.523–30–5, Jun. 1986 Versuch B6, BIeV-R65.523–30–6, Aug. 1986 Versuch A7, BIeV-R65.523–30–7, Sep. 1986 Versuch A8, BIeV-R65.523–30–8, Sep. 1986 Versuch A9, BIeV-R65.523–30–9, Sep. 1986 [32]

W.Schöck et al., DEMONA Jahresberichte

Jahresbericht 1984, KfK 3942, (1985) Jahresbericht 1985, KfK 4182, (1986) Jahresbericht 1986, KfK 4285, (1987) [33] [34] [35] [36] [37] [38] [39] [40] [41] [42] [43] [44] [45] [46] [47] [48] [49]

W.Schöck et al., Results of the DEMONA Program, Transactions of the 4th International ENS/ANS Conference, Geneve, 1.-6. 6. 1986, ENS Berne (1986) Vol.3, pp. 219–224 W.Schöck, H.Bunz, R.E.Adams, M.L.Tobias, F.Rahn, Large scale experiments on aerosol behavior in light water reactor containments, Nuclear Technology Vol. 81 (1988), pp. 139–157 T.J.Albiol, J.D.McCormack, L.D.Muhlestein, The LWR Aerosol Containment Experiments (LACE), Instrumentation and Measurement Techniques used in the LACE Program, LACE TR-006, June 1987. G.R.Bloom, R.K.Milliard, J.D.McCormack, L.D., Muhlestein, The LWR Aerosol Containment Experiments (LACE), Aerosol Behavior under LWR Containment Bypass Conditions, Results of Tests CB-1, CB-2 and CB-3, LACE TR-001, Nov. 1986. R.K.Milliard, L.D.Muhlestein, T.J.Albiol, The LWR Aerosol Containment Experiments (LACE), Final Report of Experimental Results of LACE Test LA2—Failure to Isolate Containment, LACE TR-007, June 1987. D.R.Dickinson, R.K.Milliard, L.D.Muhlestein, D.C.Mecham, G.Carraro, The LWR Aerosol Containment Experiments (LACE), Aerosol Behavior in LWR Containment Bypass Piping Results of LACE Test LA3, LACE TR-011, July 1987. F.J.Rahn, The LWR Aerosol Containment Experiments (LACE), Project Summary Report, LACE TR-012, Nov. 1988. J.D.McCormack, R.K.Milliard, J.M.Salgado, The LWR Aerosol Containment Experiments (LACE), Final Report of Experimental Results of LACE Test LA4—Late Containment Failure with Overlapping Aerosol Injection Periods, LACE TR-025, Oct. 1987. B.R.Bowsher, The LWR Aerosol Containment Experiments (LACE), Minutes of the LACE Technical and Theoretical Support Group Meeting. AERE Harwell, 15–16 April 1986, LACE TR-034, May 1986. The LWR Aerosol Containment Experiments (LACE), LACE TR-036 The LWR Aerosol Containment Experiments (LACE), Water Aerosol Leakage Experiments, Commissioning Tests, LACE TR-037 The LWR Aerosol Containment Experiments (LACE), A Reassessment of the CsOH-H2O System, LACE TR-076 J.Jokiniemi, The Growth of Hygroscopic Particles during Severe Core Melt Accidents, Nucl. Technol. 83, (1988) 16. I.N.Tang, Phase transformations and growth of aerosol particles composed of mixed salts, J.Aerosol Sci. 7, (1976) 361 W.Schöck, Post-test calculations of aerosol behavior in DEMONA experiment B3 with various computer codes, EUR 11374 EN (1988) H.Bunz, W.Schöck, Co-agglomeration of multi-component aerosols, J. Aerosol Sci. 20, (1989) 1059 M.Furrer, R.C.Cripps, E.Frick, lod-Verhaltenscode IMPAIR-2, PSI-25, (1989)

62

[50]

THE PHEBUS FISSION PRODUCT PROJECT

J.O.Liljenzin, Optimum Group Selection in multi-radionuclide determinations, J.Radioanal. Nucl. Chem. 131, (1988) 51

FISSION-PRODUCT CHEMISTRY IN SEVERE REACTOR ACCIDENTS: REVIEW OF RELEVANT INTEGRAL EXPERIMENTS A L NICHOLS Chemical Physics Department, AEA Technology, Winfrith Technology Centre, UK. C HUEBER IPSN/DRS, CEA, CEN Cadarache, France. SUMMARY The attenuation of the radioactive fission-product emission from a severe reactor accident will depend on a combination of chemical, physical and thermal-hydraulic effects. Chemical species stabilised under the prevailing conditions will determine the extent of aerosol formation and any subsequent interaction, so defining the magnitude and physical forms of the eventual release into the environment. While several important integral tests have taken place in recent years, these experiments have tended to focus on the generation of mass-balance and aerosol-related data to test and validate materials-transport codes rather than study the impact of important chemical phenomena. This emphasis on thermal hydraulics, fuel behaviour and aerosol properties has occurred in many tests (eg PBF, DEMONA, Marviken-V, LACE and ACE). Nevertheless, the generation and reaction of the chemical species in all of these programmes determined the transport properties of the resulting vapours and aerosols. Chemical effects have been studied in measurements somewhat subsidiary to the main aims of the tests. This work has been reviewed in detail with respect to Marviken-V, LACE, ACE and Falcon. Specific issues remain to be addressed, and these are discussed in terms of the proposed Phebus-FP programme. 1. INTRODUCTION The fission products and other debris released from an overheated reactor core would pass through some portion of the reactor coolant system (RCS), prior to the bulk discharge of molten fuel into the containment if the pressure vessel fails. Early and intermediate stages of a wide range of postulated severe accident sequences would result in the release of high- and mediumvolatile fission products from the damaged fuel, and their transport to the containment atmosphere via the RCS. Specific types of accident could result in the bypass of the containment building so that behaviour in the RCS becomes of even greater importance in defining the eventual source term to the environment. A sound understanding is required of all the fission products released from the core, and any important chemical and physical changes associated with the RCS and the containment need to be identified and quantified to give more accurate guidance on the control, mitigation and consequences of reactor accidents. The most widely publicised estimates of the risks associated with severe accidents were presented in the WASH-1400 Reactor Safety Study (1). Although this methodology has been adopted as the basis for many subsequent studies, the releases quoted for the fission-product source terms to the environment were judged to be extremely conservative. This is clearly demonstrated by the assumption that no reactions or retention will occur in the primary circuit for all pressurised water reactor accident sequences and specific accidents in boiling water reactors because of a significant number of chemical and thermal-hydraulic uncertainties. Fission products released from the damaged fuel are assumed to be transported unchanged to the containment, irrespective of any possible reactions and attenuation processes in the RCS. It was implied that conditions within the core region and primary circuit were so poorly understood that no reliable calculations could be made. Furthermore, fission products were grouped into classes of elements with similar properties in order to simplify the modelling of their release, transport and attenuation. These groupings were assumed to apply throughout the reactor system with no change in chemical form. More recent safety analyses have included attempts to quantify attenuation in the RCS, while coarsening the fissionproduct classification (2, 3): all species were defined as unreactive aerosol particles apart from the noble gases (Kr and Xe), molecular iodine (I2) and organic iodides. Clearly such simplifications are incorrect since it is unreasonable to assume that the chemical forms of most elements will remain unaffected by changes in the oxygen potential within the RCS, or by possible

64

THE PHEBUS FISSION PRODUCT PROJECT

interactions with other fission products and miscellaneous bulk materials (eg Zircaloy cladding, 304 stainless steel and Inconel structures, control rod alloy components, boric acid and steam). Levenson and Rahn (4) and Campbell et al (5) have drawn attention to the importance of various chemical phenomena in reducing the calculated source terms. Thus, the release and transport of iodine as the molecular species (I2) was viewed as highly unlikely on the basis of chemical conditions within the RCS; the formation of less volatile caesium iodide was argued on the basis of thermodynamic considerations. A reassessment of the existing database suggested that approximately 65% of the volatile fission products (excluding Kr and Xe) and over 99% of the resulting aerosol could be retained in the primary circuit (6). After the accident at TMI-2, it was estimated that the radioactive iodine release to the environment was a factor of 105 lower than calculated on the basis of WASH-1400 (4). It was concluded that more detailed consideration needed to be given to the physical and chemical processes in the RCS and containment building, and this has led on to substantial studies involving small-scale separate-effects experiments and large-scale integral tests. Chemistry has been noted as a significant uncertainty in source term reviews by the American Nuclear Society (7) and American Physical Society (8). Both study groups advocated a sound understanding of fission-product behaviour in the RCS in order to determine the impact of various chemical phenomena. Source term uncertainties have also been reviewed for nuclear power plants in the USA to identify and assist in the resolution of a number of technical issues (9). Emphasis was placed on defining the chemical form(s) of iodine and the revaporisation of fission products; substantial deficiencies were identified with respect to reaction rates, Gibbs energies and phase diagrams. These problems were assessed in more detail by a workshop organised by the US National Research Council (10) to consider: (a) thermodynamic properties of gaseous species, and condensed oxide and metal phases, (b) heterogeneous kinetics of the release of radioactive species from the fuel, (c) fluid transport processes, (d) vapour-aerosol interactions and transport. Various review papers were presented during the course of this workshop, including efforts to assess the chemical impact of structural and neutron absorber materials under accident conditions (11). The gas composition, temperature gradient and associated fluctuations within the RCS and containment will determine the chemical species and physical forms of the fission products and other materials released from the overheated core. Temperatures, pressure and gas composition will depend on the precise accident sequence and may vary significantly over an extended period of time. A degrading core is non-isothermal, and therefore specific fission products will not all be released from the fuel at a particular time to form a well-defined mixture of constituents. Examples of the calculated variations in temperature, gas flowrate and hydrogen/steam ratio are given in Figures 1 and 2 for two different accident sequences. Changes in the hydrogen/steam ratio reflect the degree of oxidation of various metal structures within the region of the core, and define variations in the oxygen potential of the RCS. These parameters will determine the chemical species formed in the gas phase after release from the fuel, strongly influencing their transport and attenuation in the upper plenum and RCS. Defining the chemical species and understanding their behaviour in the primary circuit and containment building are challenging problems in quantifying the source terms. Some vapour species will interact and decompose as conditions change, and the generation of aerosols from bulkier materials such as the control rods and Zircaloy cladding will complicate the calculated estimates. Faced with such a task, some reasonable simplifications have to be made when undertaking experimental and theoretical studies of chemistry in severe reactor accidents, and all previous integral tests have been no exception in this respect. 2. INTEGRAL TESTS The recommendations from various international reviews have provided the impetus for a range of fission-product chemistry studies: (a) kinetic and thermodynamic data (12), (b) development of separate-effects experiments (as reviewed in ref (13), for example), (c) integral tests. Both thermodynamic and kinetic data are important inputs to the mechanistic codes under development (14, 15, 16), while separate-effects tests provided the most cost-effective method of quantifying some of the most important chemical phenomena in isolation. When a substantial body of such work has been undertaken, the perceived understanding (and resulting computer-based codes) need to be satisfactorily challenged by data from integral tests in which a number or all

FISSION-PRODUCT CHEMISTRY IN SEVERE REACTOR ACCIDENTS

65

Figure 1 Variations in Temperatures, Gas Flow Rate and Hydrogen/Steam Ratio in the Core and Reactor Coolant System: AB Accident Sequence

important processes are combined. Fission-product chemistry is reviewed below in terms of those integral tests judged to have generated relevant chemical data. 2.1 Marviken-V The Marviken-V Aerosol Transport Tests (MXV-ATT) were undertaken to study the transport and attenuation of aerosols and simulant volatile fission products within the primary system of a light water reactor (17, 18). Non-radioactive materials were used to determine the transport properties of volatile fission products (caesium iodide, caesium hydroxide and tellurium) and bulk aerosols (elemental silver and manganese (II) oxide) through a large-scale reactor system with a steam atmosphere (19, 20). The facility consisted of the following: (a) reactor test vessel (140 m3) with structural internals (200 m2 surface area),

66

THE PHEBUS FISSION PRODUCT PROJECT

Figure 2 Variations in Temperatures, Gas Flow Rate and Hydrogen/Steam Ratio in the Core and Reactor Coolant System: TMLB' Accident Sequence

(b) pressuriser (50 m3), (c) relief tank (50 m3), (d) pipework between the main vessels, (e) outlet filter, coupled to various types of aerosol generator. Measurements included flow rates, gas and surface temperatures, pressures, gas composition, aerosol mass feed rate and conversion efficiency, mass concentration, optical density, particle size distributions, deposition rates and chemical composition. While emphasis was placed on achieving mass balance and identifying the form and location of the elements and compounds that passed through the facility, efforts were also made to determine various chemical effects such as vapour-pipework/coupon and vapour-aerosol interactions (21, 22).

FISSION-PRODUCT CHEMISTRY IN SEVERE REACTOR ACCIDENTS

67

Figure 3 Aerosol Depth-Profiling: Micrograph-Variation of EDS Data with Electron Beam Accelerating Voltage

Representative data for the aerosol deposits from test 4 are given in Figure 3 (scanning electron microscopy/x-ray energy dispersive spectroscopy (SEM/EDS)). These studies illustrate two important features: (a) two distinct forms of aerosol were stabilised, one consisting of relatively large spherical silver particles (~ 1 µm volume equivalent diameter) and the other of smaller crystalline manganese oxide particles; (b) as the mixed aerosol was transported through the facility, manganese oxide particles deposited on the surface of the silver aerosols. Analyses for caesium and tellurium species in the aerosol particles were more uncertain. Tellurium concentrated at the surface of both the manganese- and silver-based aerosols, but a higher correlation was observed with silver to support the formation of silver telluride (Ag2Te). Caesium was predominantly associated with the manganese component, implying the stabilisation of a mixed caesium-manganese-oxygen compound such as caesium manganate (Cs2MnO4). A correlation was also observed between the caesium and tellurium components of the aerosol that could be attributed to the existence of caesium telluride (Cs2Te) and/or caesium tellurite (Cs2TeO3). It is possible that the elemental tellurium oxidised on exposure to air after the test (which would be enhanced by the finely-divided form of the tellurium deposit), followed by further reaction with caesium species to form caesium tellurite. Chemical effects had a significant impact on fission-product and aerosol transport within the pipework and pressuriser, particularly vapour-aerosol-steam interactions, chemisorption onto structures, competing redox reactions, and aerosol nucleation and growth processes. Thus, caesium hydroxide reacted with carbon dioxide, tellurium and manganese oxide, tellurium formed stable compounds with silver and manganese, and oxidation of the metal surfaces affected the diffusion of caesium and tellurium to the reactive substrate. Somewhat surprisingly, silver aerosols were initially formed in preference to refractory MnO to produce a chemically heterogeneous mixture of particles, with some evidence of both heterogeneous and homogeneous nucleation of manganese (II) oxide. Analytical studies of the stainless steel and Zircaloy deposition coupons included in test 4 demonstrated that the corrosion products affected the penetration of some of the vapour species. At relatively low temperatures (~ 350°C) there was some evidence of physisorption at the coupon surface; at higher temperatures (~ 750°C) caesium and silver diffused into the Zircaloy and stainless steel coupons, while the equivalent attack by tellurium was inhibited by either the metal-oxide surface or preferential interaction between the tellurium vapour and other components of the test such as the silver-based aerosol. The multi-component aerosols generated in tests 4 and 7 were analysed in detail to reveal significant inhomogeneities within and between individual particles. Although there was no even and systematic accumulation of the manganese-based species on the primary silver particles of test 4, the transport and deposition of this debris occurred in a manner consistent with a simple co-agglomeration process. An important observation during this test was the condensation of caesium-and tellurium-based vapours on the silver-manganese (II) oxide aerosol (Figures 4 and 5). Preferential reactions of the simulant fission-product vapours with one aerosol component would significantly alter the transport and deposition behaviour of the radioactive species. While thermodynamic data can be used with confidence at high temperatures close to the overheated

68

THE PHEBUS FISSION PRODUCT PROJECT

Figure 4 AES Surface Analysis, Aerosol Deposit from Sedimentation Tray in Marviken-V Internals Test 4

Figure 5 Interpretation of AES and SIMS Depth-Profiling Data: Aerosol Particle from Marviken-V Test 4

reactor core, diffusional effects can be expected to play a more important role for a limited number of systems at lower temperatures (particularly in the containment building), requiring definition in terms of their rate of reaction. These interactions can be identified to some degree in large-scale experiments with simulants, but a more precise definition of their combined effect is best achieved by analyses of the bulk surfaces and aerosols formed in tests with representative plant materials and high burn-up fuel in the appropriate conditions envisaged for specific types of accident.

FISSION-PRODUCT CHEMISTRY IN SEVERE REACTOR ACCIDENTS

69

2.2 LWR Aerosol Containment Experiments (LACE) The LACE programme consisted of six large-scale tests at Hanford Engineering Development Laboratory, using the Containment Systems Test Facility (CSTF). A two-component aerosol was generated in all of the tests, consisting of caesium hydroxide as a water-soluble species and manganese (II) oxide as an insoluble species. The aerosols were produced separately by the reactions of caesium and manganese vapours with superheated steam in a nitrogen carrier gas, before passing the mixture through pipework representative of the primary circuit to a large stainless steel containment vessel (852 m3) at 1 bar pressure and filled initially with air at room temperature (23–27). Although the main objective of these studies was to generate mass balance and aerosol transport data so that the appropriate codes could be tested, specific analytical techniques were used to determine the elemental and chemical composition of deposits collected on the internal surfaces and filters (28–32). Individual aerosol samples consisted of inhomogeneous deposits with significant variations in the elemental composition. The relative concentrations of caesium and manganese were measured as a function of the electron accelerating voltage of the SEM/EDS system (30, 31); these data revealed the stabilisation of heterogeneous particles with a caesium-enriched outer coating and a manganese-based kernel. Such entities could arise as a consequence of the initial nucleation of manganese oxide aerosol particles followed by the condensation of caesium hydroxide, or the co-agglomeration of CsOH and MnO particles followed by some relocation of the hygroscopic caesium hydroxide on the surface of the agglomerates. Preferential sedimentation of caesium hydroxide aerosol was observed in test LA4 (overlapping aerosol injection—CsOH injected first, followed by a mixture of CsOH and MnO, and ending with MnO generation only), and this effect was attributed to the more significant condensation of steam onto hygroscopic CsOH. Specific chemical species were identified by means of x-ray diffraction and infrared spectroscopy. Manganese (II) oxide was the predominant manganese species, with hausmannite (Mn3O4) detected as a minor component. Caesium was stabilised as caesium bicarbonate (CsHCO3) formed after the tests from the reaction of caesium hydroxide with carbon dioxide in the atmosphere. Despite the relatively inert nature of these aerosols, a small fraction of the debris formed an unknown species which has been provisionally identified with the Cs-Mn-O-H system (33). There is little doubt that steam condensation will have an important impact on LWR severe accidents, as exemplified by CsOH aerosol behaviour in the LACE tests. Mixed heterogeneous aerosols were formed in these studies, and the two components retained their own individual chemical properties to some extent. This behaviour was particularly noticeable in test L4, resulting in partial separation of the two species as CsOH aerosol was preferentially removed from the atmosphere of the main containment vessel. 2.3 Advanced Containment Experiments (ACE), Phase B: Iodine Chemistry Ritzman has summarised the international R and D work on iodine behaviour in the containment during reactor accidents as part of the ACE programme (34). Many experiments have been and are being performed because of the radiobiological importance of the radionuclides of iodine. ACE Phase B was devoted to the role of iodine chemistry in severe reactor accidents, focussing on the chemical species and interaction processes that can occur in LWR containments. (a) Laboratory Experiments - effect of pH, temperature, radiation and epoxy paint on iodine behaviour under conditions relevant to reactor accidents (WRNE), - oxidation of caesium iodide in premixed hydrogen flames (ORNL), - adsorption of iodine on aerosols (ORNL). (b) Intermediate-scale Experiments - effect of steam on the oxidation of caesium iodide aerosols during hydrogen combustion (CTF-WRNE), - Radioiodine Test Facility (RTF-WRNE). (c) Large-scale Experiments - large-scale iodine experiment in the Containment Systems Test Facility (CSTF-BPNL).

70

THE PHEBUS FISSION PRODUCT PROJECT

Figure 6 Radioiodine Test Facility (RTF)

2.3.1 RTF The Radioiodine Test Facility (RTF) consists of a cylindrical carbon steel vessel in which 35 l aqueous solution can be irradiated by a 60Co gamma-ray source in the presence of 350 l gas volume and painted stainless steel liners. This vessel is contained in a lead canister and connected to circuits that permit gas/liquid recirculation and sampling, injection of the initial species in an aqueous phase, and operation of a spray system (Figure 6). The temperature is maintained by heaters, the aqueous loops are insulated, and the gas lines are trace-heated to avoid any steam condensation on cold surfaces. Measured parameters include: (a) aqueous phase: pH, dissolved O2, total iodine concentration, chemical analysis and iodine speciation, (b) gaseous phase: total iodine concentration and iodine speciation using selective filters. Nine tests were specified with different iodine species injected at the beginning of each experiment in the aqueous phase (initial concentrations of CH3I (10−5 M l−1), I2(5.10–6 M l−1) and CsI (10−5 M l−1)). All of the tests were performed with painted internal surfaces and lasted six days, ending with the addition of hydrazine as a spray solution (35). The following experimental conditions were adopted: test series 1: no radiation, pH 9, test series 2: radiation, pH 9, test series 3: radiation, pH 5.5 increased to 9 after 4 days, in which the total gas-borne iodine concentration was a key parameter and was measured as a function of time. A number of complementary studies are also planned with unpainted metal surfaces. The aqueous CH3I test resulted in a maximum gaseous iodine concentration after 3.5 h, followed by a first-order rate of decrease with time; this behaviour can be explained in terms of the high volatility of CH3I and the hydrolysis of this species in aqueous solution. The initial gaseous iodine concentration was much lower for CsI, and the observed behaviour implied the formation of organic iodide from an ill-defined interaction with the painted surfaces. Molecular iodine (I2) injection resulted in intermediate concentrations of gaseous iodine species, although the gas-phase iodine sampling system did not work properly during this part of the test. Adsorption onto painted surfaces was not very important in the absence of radiation, which is consistent with low levels of I2 (unstable at pH 9). The changes in the gas-borne iodine concentrations in test series 2 and 3 were similar, except that the gas-phase iodine concentration was higher and the dry paint surfaces acted as a more efficient iodine sink in test series 3. Specific difficulties were encountered, including uncontrollable changes to the pH caused by the epoxy-painted surfaces and radiolysis in the aqueous , , H2O2 and other species). CH3I behaviour was comparable to the equivalent test without phase (ie generation of radiation, although the gas-borne concentration peaked earlier and decreased with a constant slope due to the more efficient

FISSION-PRODUCT CHEMISTRY IN SEVERE REACTOR ACCIDENTS

71

radiolytic decomposition of CH3I compared with hydrolysis. The long-term gas-borne iodine concentration level in a radiation field did not depend on the initial species, and organic iodides were stabilised as the main volatile species. 2.3.2 CSTF An iodine test was conducted in the 852 m3 vessel at Hanford Engineering Development Laboratory, which had been coated with an appropriate paint identified with the internal structures of LWR containments (Keeler and Long). Various penetrations were used to introduce the following: -

steam, soluble aerosol; reaction of caesium vapour with steam to form CsOH, insoluble aerosol; vaporisation of manganese in plasma torches to generate MnO, iodine generation; HI from a gas bottle, and I2 by heating iodine crystals.

Instruments and sampling systems included filters, coupons, sedimentation trays, condensate collectors, Maypacks, cascade impactors, thermocouples, pressure sensors and water-level indicators (36). The objective of the test was to study the interaction of gas-phase iodine with the aerosols, painted walls and steam. A condensing steam-air mixture was used slightly above 100°C at a pressure of ~ 250 kPa, and the operational procedures listed in Table 1 were adopted. The gas-borne concentrations of iodine and caesium had similar removal rates up to 400 min for test B1 and 200 min for test B2; iodine-based aerosols dominated the transport behaviour, indicating that HI and I2 reacted rapidly with the CsOH and MnO particles. Beyond this initial phase of each sub-test the removal rate decreased, and molecular iodine and organic iodides predominated, particularly after I2 injection. Increasing the pH of the pool did not result in any significant changes in the iodine chemical species or aerosol concentrations. Iodine retention on the painted surfaces of the vessel and coupons exhibited considerable scatter; the highest retention was measured for surfaces in a non-condensing environment, the lowest for condensing conditions, and intermediate values for surfaces submerged in the pool. Table 1: Experimental Procedure During CSTF Iodine-Aerosol Test time (min)

Procedure

Sump pH

0 30 35 40 1440 1440 1470 1475 1480 4275 4275 6000

start CsOH and MnO aerosol injection end CsOH and MnO aerosol injection start HI flow stop HI flow end of test B1 start CsOH and MnO aerosol injection end CsOH and MnO aerosol injection start I2 addition end I2 addition end of test B2 increase pH of pool end of test B3

5.2

5.6

8.4

2.4 Falcon Experiments have been conducted in the Falcon facility at Winfrith Technology Centre to study the interaction of fissionproduct vapours released from simulant and trace-irradiated fuel samples with aerosols generated from such materials as AgIn-Cd control rod alloy and boric acid. Falcon consists of two negative-pressure glove boxes, a system of thermal gradient tubes, a stainless steel containment vessel, and a series of filters designed for use with gamma-ray spectrometers and aerosol analysers. Small segments of fuel are clad in Zircaloy and heated to ~ 1800°C, so that the behaviour of the resulting vapouraerosol release can be studied along a pathway simulating the upper plenum, hot-leg structures and containment (37). Fissionproduct vapours and aerosols are generated and transported from high-temperature reducing conditions representing the primary circuit in the initial stages of an accident to a low-temperature predominantly oxidising atmosphere simulating the

72

THE PHEBUS FISSION PRODUCT PROJECT

containment. Various analytical techniques are used to provide information on the chemical species and physical characteristics of the release, and so provide data to assist in code development and validation. Twenty integral tests have recently been performed of increasing complexity. Elemental and chemical data were measured at the bulk and microscopic levels for the vapours and aerosols. Chemistry effects were shown to play an important role in the transport and deposition of the volatile fission products. Aerosol particles varied considerably in composition, size and morphology, significant interaction occurred between the fission-product vapours and bulk-material aerosols to modify their transport, and volatile iodine species were formed within the containment to be retained preferentially on painted surfaces and in the water sump (38). These data are suitable for comparison with mechanistic-code calculations in which a combination of phenomena can be studied, rather than focussing on a specific chemical process or species. Separate-effects experiments were conducted as part of the Falcon programme to assist in the interpretation of the data from the main test matrix. Work was undertaken on the following: (a) aerosol nucleation, (b) Ag-In-Cd control rod aerosol, (c) boric acid aerosol generated from aqueous solutions, (d) mixed aerosols of water-insoluble and water-soluble species, (e) kinetics of vapour-aerosol interactions. The aerosol formation, growth and depletion processes in LWR accidents are intimately related to a number of important chemical effects identified with nucleation, vapour condensation/interaction and steam condensation onto hygroscopic particles. Although the nucleation studies were only preliminary, the observations implied that a suitable theoretical model could be developed to define the particle size distribution, concentration and morphology of any condensed gas-borne species (39). High-temperature steam is an extremely reactive medium, and could result in the generation of a number of volatile hydroxides and oxyhydroxides from the fuel and other bulk components. Laboratory-scale experiments were conducted to investigate this effect on short sections of Ag-In-Cd control rod clad in stainless steel (40). These studies focussed on the initial stages of control rod failure and release: significant concentrations of elemental cadmium aerosol were generated in an inert atmosphere, whereas the addition of steam resulted in the release of indium (as a volatile hydroxide vapour (41)) and cadmium. Such chemical effects are important in assessing subsequent vapour-aerosol interactions, and a significant body of thermochemical data is currently being assembled to address this issue (42, 43). Similar studies have been undertaken to determine the role of boric acid (H3BO3) in the condensed and vapour phases (44, 45). While H3BO3 reacts with caesium iodide and caesium hydroxide at relatively low temperatures (~ 100°C) to form hydrogen iodide and caesium borate, extensive interaction can occur with the components of stainless steel to mitigate against such a decomposition process. Mixed aerosols of H3BO3-Cd have also been generated; the aerosol differed considerably from the equivalent debris formed in equivalent single-component experiments, resulting in a significant increase in the aerosol deposition rate (46). Thermogravimetric experiments were conducted to determine the interaction of molecular iodine (I2) with silver and cadmium deposits (47). The observed reaction kinetics depended on the temperature of the experiment with respect to the melting points of the reactants and products. Linear kinetics were observed when the temperature exceeded their melting points, and the rate-determining step was the gas-phase mass transfer of I2 vapour to the aerosol. Parabolic kinetics occurred when the aerosol existed in the solid phase, resulting in the formation of a passive deposit of AgI (or CdI2) on the surface through which the reactants had to diffuse to sustain further reaction. Rate constants were measured for both systems over the temperature ranges 20 to 500°C (Ag/I2) and 20 to 270°C (Cd/I2). Experience within the Falcon programme has shown that the main integral tests are an extremely valuable aid in assessing both primary circuit and containment codes (eg VICTORIA and CONTAIN). However, precise chemical data have been more successfully derived from the related series of separate-effects experiments. The results of these studies provide the basic data for direct use in the theoretical calculations, and the integral tests can then be used to evaluate the validity of such data in a more complex system. 2.5 Relevant Fuel Degradation Tests A number of ambitious tests have been performed at the Power Burst Facility (PBF) and Loss-of-Fluid Test (LOFT) facility, Idaho National Engineering Laboratory. Some of these studies are relevant to this review of fission-product chemistry, and are briefly considered below along with an equivalent assessment of fuel analyses from TMI-2.

FISSION-PRODUCT CHEMISTRY IN SEVERE REACTOR ACCIDENTS

73

2.5.1 Power Burst Facility (PBF) Three in-pile ‘tests were conducted with 26- or 32-rod bundles of trace- (test SFD 1–1) or highly-irradiated fuel (tests SFD 1– 3 and 1–4). Extremely limited releases of volatile fission products were observed in SFD 1–1, and significant condensation and deposition occurred in the sampling lines to reduce the radionuclide concentrations detected by the monitoring systems (48). More reliable data were obtained in SFD 1–3 (49); a tin-based aerosol was detected which agglomerated and provided nucleation sites for the vapour condensation of the perceived fission products (eg CsOH and CsI). Silver-indium-cadmium control rod samples were incorporated into SFD 1–4 (50), and at least 13 g of a dense aerosol were released from the test bundle (identified as ~ 5 g cadmium, 0.4 g silver, 0.2 g indium, 4 g tin and 3.5 g zirconium); vapour condensation and aerosol nucleation were believed to be the dominant processes immediately above the damaged fuel. Most of the chemistry in PBF appears to have been inferred from thermochemical calculations rather than being observed in a direct manner. It can be assumed that this approach reflects the difficulty of attempting to undertake definitive speciation analyses during and after such tests. 2.5.2 LOFT-FP-2 The final LOFT test is particularly relevant to the Phebus-FP programme. It is generally agreed that this experiment constitutes the only severe accident study in which measurements were made of the thermal hydraulics and fission-product release, transport and deposition in a large-scale nuclear plant with appropriately irradiated fuel (burn-up of 430 MWd/tU). A containment by-pass sequence was initiated, the central fuel bundle reached the desired temperature, and the test was safely terminated by operating the emergency core coolant system (51). Instrumentation was limited to a number of steam sample systems, gamma-ray spectrometers, various forms of deposition coupons, and an aerosol-collection filter. Operational difficulties were experienced with some of these sampling and analysis systems (eg gamma-ray detector, gas sampler and deposition coupons), which resulted in data losses. A significant number of fission products were detected at several locations within the facility, including the upper plenum, outlet line from the pressure vessel, low-pressure injection system and blowdown suppression tank (eg 103Ru, 132Te, 131I, 132I, 133I, 137Cs, 140Ba, 141Ce and 144Ce, as well as various Kr and Xe radionuclides). It was noted that the 131I concentration in the primary coolant system was over two hundred times higher than that in the blowdown system liquid. Over twenty times the radioactivity was detected above the damaged core than in the broken hot-leg, and most of the resulting aerosol particles were greater than 5 µm volume equivalent diameter (predominantly cadmium with some yttrium, caesium and barium). Ruthenium and tellurium were believed to be transported as either sub-micron particles or gaseous species. Two major aims of LOFT-FP-2 were to determine the retention of volatile fission products on RCS surfaces in the plenum and pipework, and obtain mass balance data for the fission products in the fuel, coolant system and blowdown suppression tank. While it was claimed that these and other objectives were met, there were believed to be large uncertainties in the data due to the failure or poor performance of the measurement systems. 2.5.3 TMI-2 While not classified as a well-instrumented and controlled experiment, TMI-2 has produced a number of insights into specific types of severe reactor accident. Chemistry studies have been undertaken (52, 53), and concerted efforts have been made to obtain an international concensus on the core degradation processes (54). Estimates have been made of the likely oxygen potentials in the damaged reactor pressure vessel to explain some of the fission-product behaviour. Caesium and tellurium retention were larger than expected and can be explained on the basis of various separate-effects tests, while antimony and ruthenium alloyed with stainless steel components of the core. It was also noted that a larger fraction of the cadmium inventory was released than the silver and indium constituents of the control rod alloy. A comprehensive series of post-test analysis techniques were used to good effect to study samples taken from the core. It would also have been beneficial to undertake similar studies of the debris transported and deposited elsewhere (ie RCS and containment). Such analyses would have generated additional data to ascertain the chemical phenomena that engendered some of the resulting observations.

74

THE PHEBUS FISSION PRODUCT PROJECT

Figure 7 Phebus-FP Facility: Proposed Sampling Points

3. PHEBUS-FP The main objective of the Phebus-FP project is to improve our knowledge and ability to calculate the magnitude, physicochemical form and release rates of the radioactive material emitted into the environment during a severe LWR accident. A major aim is to simulate the reactor core, primary circuit and containment building in a single facility capable of producing representative thermal-hydraulic conditions, and hence an inter-related combination of phenomena that cannot be achieved in separate-effects tests (Figure 7). Unlike other previous experiments, the Phebus-FP programme is designed to consider all aspects of fission-product transport from the fuel to the containment, and ensure that no major process has been omitted prior to pressure vessel failure (55). The chemistry of severe reactor accidents can be considered in terms of the following effects: (a) fuel behaviour and fission-product release as the core degrades in the pressure vessel; (b) fission-product interactions and transport in the primary circuit prior to failure of the pressure vessel; (c) fission-product behaviour in the containment prior to failure of the pressure vessel; (d) fuel ejection into the containment when the pressure vessel fails, resulting in molten core-concrete interactions and further fission-product release; (e) resuspension of fission products in the primary circuit and containment, induced by such phenomena as hydrogen burns, steam explosions and direct containment heating. Phebus-FP will primarily address issues identified with (a), (b) and (c). Fuel behaviour is discussed in a companion review (56), while fission-product chemistry in the primary circuit and containment constitute the main emphasis of this paper. There has been increasing awareness of the important role played by the primary circuit in altering the release and modifying the physical and chemical forms of the gas-borne fission products. Specific chemical reactions could drastically alter the physical forms of the source term, and change the resulting transport and deposition properties. Thus, a sound understanding of fission-product behaviour in the reactor coolant system is required to produce defendable consequence calculations. Any important chemical and physical changes need to be identified and quantified to give more confident assessments of reactor accidents. Furthermore, the containment building could be bypassed in certain accident sequences, and it is important to predict behaviour in the primary circuit if realistic source terms are to be derived for these low probabilityhigh consequence events. The most relevant thermochemical data are being measured (12, 43), and the resulting parameters assembled as datafiles so that the stable chemical species can be calculated with confidence for a defined set of conditions in a range of accident sequences. A combination of the oxygen potential of the system, temperature(s), partial vapour pressures and absolute pressure will determine the identities of the chemical compounds and their nucleation/condensation characteristics as they traverse the primary circuit and enter the containment. This reasonably comprehensive database should be available by mid-1992 to be coupled with mechanistic codes such as VICTORIA (14). Suitable data for heterogeneous reactions are also

FISSION-PRODUCT CHEMISTRY IN SEVERE REACTOR ACCIDENTS

75

being generated to add greater realism (and complexity) to the calculations (47). Thus, there is a clear need for a series of well-defined integral experiments in the Phebus facility to test and validate mechanistic calculations of the primary circuit. High burn-up fuel will be used in all of the main Phebus-FP tests to generate a realistic supply of mixed fission-product vapours and aerosols under representative conditions. Both Ag-In-Cd control rod alloy and boric acid will also be included in most of the tests to interact with other components of the core and generate representative aerosols. The proposed measurements of temperature, pressure, fluid flow rate and condensate flow rate are reasonably comprehensive, assuming there are no serious malfunctions (particularly when recording the high temperatures to be expected in the core). Unfortunately, sampling and analysis facilities are extremely limited for the above-core and high-temperature regions of the primary circuit. The first significant sampling point is located at C (Figure 7). At this position, the vapours will have already passed through 10 m of Inconel-lined pipework constituting 1 m2 of surface area; bulk-materials aerosols with a surface area of approximately 100 m2 will also have been transported to this sampling point. Under such conditions (typical Reynolds’ number of 2000), most reactive vapours should have undergone either vapour-surface or vapour-aerosol interactions. The relatively abrupt temperature drop above the damaged fuel bundle favours rapid aerosol formation, and cannot be judged as representative of high-flow accident scenarios that result in extensive vapour transport. Some thought should be given to further sampling or monitoring points close to the fuel bundle. Direct identification of the high-temperature vapour species emitted from the fuel does not currently form part of the measurement programme, and this information will only be obtained by interpretation of the post-test analysis data of the structural surfaces and aerosol deposits in the circuit and on filters. Important aerosol chemistry measurements are identified with the containment vessel (REPF 502). This chamber represents a key attenuation system, and fission product-aerosol emissions need to be accurately characterised at sampling points G and H (Figure 7). Time-dependent data are required using sequential sedimentation samplers, impactors and filters, and the resulting multi-component deposits merit intensive efforts to determine their chemical composition. Iodine behaviour in the Phebus containment needs to be fully understood during and some considerable period after each test. Species samplers need to be used to differentiate between radioiodine that has stabilised as inorganic aerosols, molecular iodine and organic iodides. This is probably best achieved by adopting a suitable Maypack system operated in conjunction with a number of in-situ gamma-ray spectrometers (57). However, specific technical problems must be overcome to guarantee the successful operation of such a device under high-temperature and steam-condensing conditions, and alternative procedures should also be explored (eg high-performance chromatography) in case these difficulties cannot be satisfactorily overcome. It is also clear that the post-test analysis techniques will need to be comprehensive in order to ascertain the elemental and chemical composition of the fission products deposited in the primary circuit and transported into the containment vessel. A judicious combination of automated scanning electron microscopy/x-ray energy-dispersive spectroscopy (SEM/EDS), x-ray photoelectron spectroscopy (XPS), x-ray diffraction (XRD) and secondary-ion mass spectrometry (SIMS) is recommended in conjunction with depth-profiling and etching techniques (58). 4. CONCLUSIONS The transport and deposition of gas-borne debris in a severe reactor accident will depend on a combination of chemical, physical and thermal-hydraulic effects. Both the concentrations and chemical forms of the radioactive nuclides and other materials suspended within the damaged plant will determine the extent of any aerosol formation and interaction, so defining the magnitude of any eventual emission into the environment. Moreover, knowledge of the chemical species constituting the aerosol could assist in the formulation of accident management procedures and the design of clean-up facilities. However, the precise chemical composition of the vapours and aerosols released under such conditions is difficult to determine. Apart from gamma-ray spectroscopy, attempts to undertake detailed in situ elemental and chemical analyses have been comparatively rare. Although vapour-phase chemistry is important in assessing and defining the transport of material through the primary circuit, the large thermal gradient above the damaged fuel bundle and specific sampling restraints imply that the Phebus-FP measurements will only occur after considerable aerosol formation has occurred. Under such circumstances, emphasis should be placed on measurements to determine the chemical composition of the debris deposited on surfaces and collected by filters, despite the significant problems associated with their high radiation field. The Phebus-FP tests provide a considerable opportunity to determine the chemical species released and stabilised under representative accident conditions from a realistic source of high burn-up fuel. These integral experiments should be designed to provide the necessary thermal-hydraulic and chemical data (ie species, concentrations and physical form(s)) to assess the validity of the chemistry models in both primary circuit and containment modelling codes. The aim should be to use sufficient instrumentation to determine whether any unforeseen species and phenomena occur; basic data defining the most stable vapours and aerosols and their interactions are best obtained from separate-effects experiments undertaken to complement

76

THE PHEBUS FISSION PRODUCT PROJECT

Phebus-FP. It is stressed that the composition of the fission-product emission in the RCS and subsequent behaviour in the containment are particularly important, and every effort should be made to measure these data accurately during and after each test. ACKNOWLEDGEMENTS The preparative work for this review paper was funded by the Commissariat à l’Energie Atomique and the UK Health and Safety Executive. REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14

15

16

17 18

19 20 21 22 23 24

() WASH-1400 (1975). Reactor Safety Study: An Assessment of Accident Risks in US Commercial Nuclear Power Plants, US Nuclear Regulatory Commission Report NUREG 75/014. () A Study by GRS on Behalf of the BMFP (1981). German Nuclear Power Station Risk Study—An Investigation into the Risk Resulting from Accidents in Nuclear Power Stations, Verlag TUV, Rheinland. () WCAP 9991 (1982). Sizewell B Probabilistic Safety Study. () LEVENSON, M. and RAHN, F.J. (1981). Realistic Estimates of the Consequences of Nuclear Accidents, Nucl. Technol., 53, 99. () CAMPBELL, D.O., MALINAUSKAS, A.P. and STRATTON, W.R. (1981). The Chemical Behavior of Fission Product Iodine in Light Water Reactor Accidents, Nucl. Technol., 53, 111. () GIESEKE, J.A., CYBULSKIS, P., DENNING, R.S., KUHLMAN, M.R., LEE, K.W. and CHEN, H. (1984). Radionuclide Release Under Specific LWR Accident Conditions, Battelle Memorial Institute Report BMI-2104 , Vols I-VI. () American Nuclear Society (1984). Report of the Special Committee on Source Terms. () American Physical Society (1985). Report to the American Physical Society of the Study Group on Radionuclide Release from Severe Accidents at Nuclear Power Plants, Rev. Mod. Phys. 57(3), Part II, S1. () KOUTS, H. (1987). Review of Research Uncertainties in Estimates of Source Terms from Severe Accidents in Nuclear Power Plants, Brookhaven National Laboratory Report NUREG/CR-4883, BNL-NUREG-52061. () US National Research Council (1988). Chemical Processes and Products in Severe Nuclear Reactor Accidents, Report of a Workshop held at Captiva Island, Florida, 9–12 December 1987, National Academy Press, Washington DC. () NICHOLS, A.L. (1987). Severe Nuclear Reactor Accidents: Selected Chemical Phenomena Within the Reactor Coolant System of a Pressurized Water Reactor, High Temperature Science, 24, 125. () NICHOLS, A.L. (1990). Fission Product Chemistry in Severe Nuclear Reactor Accidents, Report of a Specialist Meeting at JRCIspra, 15–17 January 1990, EUR 12989 EN. () BOWSHER, B.R. (1987). Fission Product Chemistry and Aerosol Behaviour in the Primary Circuit of a Pressurized Water Reactor Under Severe Reactor Accident Conditions, Prog. Nucl. Energy, 20, 199. () HEAMES, T.J., WILLIAMS, D.A., BIXLER, N.E., GRIMLEY, A.J., WHEATLEY, C.J., JOHNS, N.A. and CHOWN, N.M. (1990). VICTORIA: A Mechanistic Model of Radionuclide Behavior in the Reactor Coolant System Under Severe Accident Conditions, Sandia National Laboratories Report NUREG/CR-5545, SAND90–0756. () MURATA, K.K., CARROLL, D.E., WASHINGTON, K.E., GELBARD, F., VALDEZ, G.D., WILLIAMS, D.C. and BERGERON, K.D. (1989). User’s Manual for CONTAIN 1.1: A Computer Code for Severe Nuclear Reactor Accident Containment Analysis, Sandia National Laboratories Report NUREG/CR-5026, SAND87–2309. () DUFRESNE, J., CENERINO, G., DUMAS, J.M., FERMANDJIAN, J., GAUVAIN, J., L’HERITEAU, J.P. and LHIAUBET, G. (1988). Presentation du Systeme ESCADRE et Application, IAEA-SM-296/13P, p399 in Proc. Symposium on Severe Accidents in Nuclear Power Plants, Vol 2, IAEA Vienna. () COLLEN, J., UNNEBERG, H. and MECHAM, D.C. (1985). Overview of Marviken Experimental Procedures, p20–1 in Proc. ANS Topical Meeting on Fission Product Behavior and Source Term Research , EPRI Report NP-4113-SR. () LILJENZIN, J.O., COLLEN, J., SCHÖCK, W. and RAHN, F.J. (1990). Report from the Marviken-V/DEMONA/LACE Workshop, OECD/CSNI Workshop on Aerosol Behaviour and Thermal-Hydraulics in the Containment, 26–28 November 1990, Fontenay-aux-Roses. () MXE-207 (1985). The Marviken Experiments—Fifth Series Aerosol Transport Tests, Test 7 Results, Studsvik Energiteknik AB Report. () MXE-204 (1985), The Marviken Experiments—Fifth Series Aerosol Transport Tests, Test 4 Results, Studsvik Energiteknik AB Report. () MAGNUSSON, A. (1986). Chemical Effects—Results from Various Investigations, Studsvik Energiteknik AB Report MXIP-8. () ALLEN, G.C., WILD, R.K., BOWSHER, B.R., KINGSBURY, A.F., NICHOLS, A.L., POOLE, D.M. and RIVIERE, J.C. (1987). Analysis of Samples from Test 4 of the Marviken-V Aerosol Transport Tests, Winfrith Report AEEW-R 2250. () BLOOM, G.R. (1990). Aerosol Behavior Under LWR Containment Bypass Conditions, Results of LACE Test LA1, LACE TR-005. () MILLIARD, R.K., MUHLESTEIN, L.D. and ALBIOL, T.J. (1987). Final Report of Experimental Results of LACE Test LA2— Failure to Isolate Containment, LACE TR-007 .

FISSION-PRODUCT CHEMISTRY IN SEVERE REACTOR ACCIDENTS

25 26 27 28 29 30 31 32 33 34 35

36 37

38 39 40 41

42 43 44 45 46

47 48 49

50

51 52

77

() DICKINSON, D.R., MILLIARD, R.K., MUHLESTEIN, L.D., MECHAM, D.C. and CARRARO, G. (1987). Aerosol Behavior in LWR Containment Bypass Piping—Results of LACE Test LA3, LACE TR-011. () McCORMACK, J.D., HILLIARD, R.K. and SALGADO, J.M. (1987). Final Report of Experimental Results of LACE Test LA4— Late Containment Failure with Overlapping Aerosol Injecting Periods, LACE TR-025. () DICKINSON, D.R., MECHAM, D.C. and SLAUGHTERBECK, D.C. (1988). Final Report of Experimental Results of LACE Tests LA5 and LA6—Rapid Containment Depressurisation, LACE TR-026. () BOWSHER, B.R. and NICHOLS, A.L. (1987). Analysis of Samples from Test LA1 of the LACE Project, LACE TR-014. () BOWSHER, B.R. and NICHOLS, A.L. (1987). Analysis of Samples from Test LA2 of the LACE Project, LACE TR-015. () BOWSHER, B.R., KINGSBURY, A.F. and NICHOLS, A.L. (1987). Analysis of Samples from Tests LA3A, B and C of the LACE Project, LACE TR-054. () BOWSHER, B.R., KINGSBURY, A.F. and NICHOLS, A.L. (1987). Analysis of Samples from Test LA4 of the LACE Project, LACE TR-059. () BOWSHER, B.R., BROWN, G.R., DUNN, R.J.C. and NICHOLS, A.L. (1987). Analysis of Samples from Tests LA5 and 6 of the LACE Project, LACE TR-060. () LINDQVIST, O. (1985). Private Communication, Chalmers University, Sweden. () RITZMAN, R.L. (1990). Summary of International R&D Work on Iodine Behavior in Containment During Reactor Accidents, ACE-TR-B9. () MELNYK, A.J. (1990). Preliminary Results from ACE/RTF Test 1 (Quick Look #1); EVANS, G.J. (1990). Preliminary Results from ACE/RTF Test 2 (Quick Look #2); JOBE, D.J. (1991). Preliminary Results from ACE/RTF Test 3 (Quick Look #3); ACE-TRB3. () McCORMACK, J.D., DICKINSON, D.R. and ALLEMANN, R.T. (1990). Final Report of Experimental Results of the ACE Large Scale Iodine-Aerosol Test in the Containment Systems Test Facility, ACE-TR-B10. () BENNETT, P.J., BENSON, C.G., BOWSHER, B.R. and NICHOLS, A.L. (1990). The Falcon Programme: Multicomponent Aerosol Behaviour in the Primary Circuit and Containment, OECD/CSNI Workshop on Aerosol Behaviour and Thermal-Hydraulics in the Containment, 26–28 November 1990, Fontenay-aux-Roses. () BEARD, A.M., BENNETT, P.J., BENSON, C.G. and SABATHIER, F. (1990). Chemistry Aspects of the Falcon Programme, AEA Technology Report AEA TRS 5072. () BUCKLE, E.R., BEARD, A.M. and BOWSHER, B.R. (1990). Caesium Iodide Aerosol Nucleation Studies, AEA Technology Report AEA TRS 5008. () BEARD, A.M. and BENNETT, P.J. (1990). Characterisation of Control Rod Release Rates and Aerosol Behaviour, AEA Technology Report AEA TRS 5064. () BEATTIE, I.R., JONES, P.J., BOWSHER, B.R., NICHOLS, A.L., POTTER, P.E. and RAND, M.H. (1989). Metal Oxides and Hydroxides: Their Relevance to Vapour Transport in Severe Reactor Accidents, ICHMT Seminar on Fission Product Transport Processes in Reactor Accidents, 22–26 May 1989, Dubrovnik. () CORDFUNKE, E.H.P. and KONINGS, R.J.M. (1990). Thermochemical Data for Reactor Materials and Fission Products, Elsevier Science Publishers, Amsterdam. () BALL, R.G.J., BOWSHER, B.R., CORDFUNKE, E.H.P., DICKINSON, S., KONINGS, R.J.M. and RAND, M.H. (1991). Thermo chemical Data Acquisition, AEA Technology Report AEA TRS 5068. () ANDERSON, A.B., BEARD, A.M., BENNETT, P.J. and BENSON, C.G. (1991). Characterisation of Boric Acid Aerosol Behaviour and Interactions with Stainless Steel, AEA Technology Report AEA TRS 5091. () BOWSHER, B.R., DICKINSON, S., OGDEN, J.S. and YOUNG, N.A. (1989). New Estimates for the Thermodynamic Functions of Molecular Boric Acid, Thermochim. Acta, 141, 125. () BEARD, A.M., BENNETT, P.J., BENSON, C.G., BOWSHER, B.R., CHOWN, N.M., DEANE, A.M., HENSHAW, J., KETCHELL, N., KINGSBURY, A.F., NEWLAND, M.S., ROBERTS, G.J., SABATHIER, F., SMITH, P.N. and WILLIAMS, D.A. (1990). Chemistry Studies in Support of Phebus-FP: Multicomponent Aerosol Behaviour, Technical Progress Report, 1 January—30 June 1990, AEA Technology Report AEA TRS 5017. () HENSHAW, J., NEWLAND, M.S. and WOOD, S.J. (1991). Thermogravimetric Studies of. Vapour-Aerosol Interactions, AEA Technology Report AEA TRS 5074. () MARTINSON, Z.R., PETTI, D.A. and COOK, B.A. (1986). PBF Severe Fuel Damage Test 1–1 Test Results Report, EG&G Idaho Inc. Report NUREG/CR-4684, EGG-2463, Vols 1 and 2. () MARTINSON, Z.R., GASPARINI, M., HOBBINS, R.R., PETTI, D.A., ALLISON, C.M., HOHORST, J.K., HAGRMAN, D.L. and VINJAMURI, K. (1989). PBF Severe Fuel Damage Test 1–3 Test Results Report, EG&G Idaho Inc. Report NUREG/CR-5354, EGG-2565. () PETTI, D.A., MARTINSON, Z.R., HOBBINS, R.R., ALLISON, C.M., CARLSON, E.R., HAGRMAN, D.L., CHENG, T.C., HARTWELL, J.K., VINJAMURI, K. and SEIFKEN, L.J. (1989). Power Burst Facility (PBF) Severe Fuel Damage Test 1–4 Test Results Report, EG&G Idaho Inc. Report, NUREG/CR-5163, EGG-2542. () CARBONEAU, M.L., NITSCHKE, R.L., MECHAM, D.C., CORYELL, E.W. and BAGUES, J.A. (1987). OECD LOFT Fission Product Experiment LP-FP-2 Data Report, EG&G Idaho Inc. Report OECD LOFT-T-3805. () HOBBINS, R.R., CRONENBERG, A.W., LANGER, S., OWEN, D.E. and AKERS, D.W. (1987). Insights on Severe Accident Chemistry from TMI-2, p4–1 in Proc. Symposium on Chemical Phenomena Associated with Radioactivity Release During Severe Nuclear Plant Accidents, US Nuclear Regulatory Commission Report NUREG/CP-0078.

78

53 54 55 56 57

58

THE PHEBUS FISSION PRODUCT PROJECT

() HOBBINS, R.R.,. OSETEK, D.J., PETTI, D.A. and HAGRMAN, D.L. (1989). Fission Product Release as a Function of Chemistry and Fuel Morphology, ICHMT Seminar on Fission Product Transport Processes in Reactor Accidents, 22–26 May 1989, Dubrovnik. () AKERS, D.W., BART, G., BOTTOMLEY, P., BROWN, A., COX. D.S., HOFMANN, P., JENSEN, S.M., KLEYKAMP. H., MANLEY, A.J., NELMARK, L.A. and TROTABAS, M. (1990). Private Communication, EG&G Idaho Inc., USA. () SCOTT DE MARTINVILLE, E.F., DELCHAMBRE, P. and VON DER HARDT, P. (1990). The Phebus-FP Project: Status Report 1989–90, CEC Report EUR 12926 EN. () WRIGHT, R.W. and HAGEN, S.J.L. (1991). Core Degradation and Fission Product Release, Seminar of the Phebus-FP Project, 5– 7 June 1991, Cadarache. () ARMITAGE, B.H., BEARD, A.M., BOWSHER, B.R., DEANE, A.M., KINGSBURY, A.F., NICHOLS, A.L., OGDEN, J.S., PACKER, T.W., SABATHIER, F. and SIMS, H.E. (1991). Fission Product Release and Transport: Assessment of Sampling and Analysis Techniques for Falcon and Phebus-FP, AEA Technology Report AEA TRS 5089. () HAMPEL, G., POSS, G., BOWSHER, B.R. and NICHOLS, A.L. (1990). Advanced Instrumentation and Post Irradiation Examination Concepts for the Analysis of Aerosols and Vapours in Source Term Experiments, p349 in Reactor Safety Research, The CEC Contribution, W.Krischer (Editor), Elsevier Applied Science, London/New York.

PHEBUS SEVERE FUEL DAMAGE PROGRAM MAIN EXPERIMENTAL RESULTS AND INSTRUMENTATION BEHAVIOR C.GONNIER*, G.REPETTO*, G.GEOFFROY**

*CEA-ISPN-DRS CEN/CADARACHE 13108 St-Paul-Lez- **CEA-DRN-DMT-SETIC CEN/SACLAY 91191 Gif sur Durance Cedex France Yvette Cedex France SUMMARY The Phebus Severe Fuel Damage Program (SFD) consists in in-pile experiments on the behavior of PWR-type fuel rods during postualted beyond design basis accident. The six tests performed from december 1986 to June 1989 are devoted to the study of the early phase of in-vessel core melt degradation in a temperature range up to 2800 K. Different phenomena were studied during the tests: cladding oxidation, interaction between fuel pellets and zircaloy cladding, interaction due to other structure materials: Inconel, Stainless steel, Silver-Indium-Cadmium of control rods. In this framework, tests were carried out with very different thermal-hydraulic conditions: oxidizing atmosphere (steam) or not, at low or high pressure (0,5 to 3,5 MPa). Some aspects of the instrumentation behavior, and of the studies performed in order to improve the accuracy of test parameters are briefly discussed. 1. INTRODUCTION A large program is presently being conducted in France by the “Commissariat à l’Energie Atomique” (CEA) to study the behavior of PWR fuel during postulated severe accidents. The experimental program is mainly performed in the Phebus test reactor. The general objective of the SFD program is to improve our knowledge about degradation phenomena at high temperature, from 1500 to 2800 K. The results provide a technical data base for the verification of the french mechanistic ICARE code now under development. Many phenomena were observed and analysed: -

zircaloy oxidation and its corresponding aspects: hydrogen generation, and damages of cladding, interaction between zircaloy and fuel pellets (through solid-solid and liquid-solid contact), interaction between inconel spacer-grids and zircaloy cladding (with outcoming fuel dissolution), interaction between absorber materials (Ag-In-Cd alloy) and fuel rods, relocation of materials and flow-blockage formation.

The six tests of the SFD program were performed with bundles of 0,8 m active length, and 21 unirradiated fuel rods. Behavior of irradiated fuel, fission products (FP) release and transport will be studied in the forthcoming Phebus FP program planned for mid 1992. 2. PHEBUS SFD FACILITY The Phebus SFD facility allowed to reproduce thermal-hydraulic conditions relatively close to what would exist in core under beyond design basis conditions. The principle of the tests consists in supplying a bundle of 21 PWR-type fresh fuel rods with a flowrate mixture of overheated gases (steam-hydrogen or helium). The fuel power is adjusted according to the required temperature evolution. The facility mainly consists of a driver core, a SFD loop, and a pressurized water loop:

80

THE PHEBUS FISSION PRODUCT PROJECT

Figure 1 SFD test stringer

- the driver core supplies the test fuel with nuclear power. A quite flat axial profile of power (over 50 cm) is obtained with a neutronic filter of boron steel. For safety considerations the maximum lineic power is around 40 W/cm. Nevertheless, the maximum value used during SFD program is 18 W/cm, - the SFD loop allows to reproduce the thermal hydraulic conditions. The main component is the test stringer (fig. 1) located on the vertical axis of the core. It consists of three injection lines (steam, hydrogen, helium) for the gas supply, an electrical superheating device to increase the gas inlet temperature (max 1200 K) to the level required in the test scenario, a test section, and an exit line which leads the gaseous flow to a pressure regulating system (pressure range: 0,1 to 5 MPa) and then to a condenser and a storage tank. - the pressurized water loop (8 MPa, 533 K) is a buffer circuit between the driver core and the SFD loop. It mainly acts as a cooling circuit of the test stringer. The 21 rods (in a 12,6 mm pitch matrix) are maintained by two grids (inconel or zircaloy type), 560 mm apart and located on either side of the mid plane. The implementation of fusible seals in the upper plugs of the rods (melting point ~ 1120 K) allows to avoid clad balloonning due to pressure effects during the tests. They were not used when collapse of the cladding onto the fuel pellets was required. The bundle is surrounded with an octagonal zircaloy liner (0,6 mm thick; 69,5 mm between faces of the octagon), a fibrous zirconia (70 % of porosity; 94 mm ext. diameter), a dense zirconia layer (1 mm) and the external stainless steel tube (8 mm). 3. INSTRUMENTATION AND TEST PARAMETERS MONITORING Carefull attention has been devoted to provide reliable informations about temperature measurements and hydrogen generation, to provide an accurate monitoring of the test parameters: thermalhydraulic conditions, bundle power for intact and degraded geometries, and to improve our knowledge on the thermal conductivity of the shroud in porous zirconia.

PHEBUS SEVERE FUEL DAMAGE PROGRAM MAIN EXPERIMENTAL RESULTS AND INSTRUMENTATION BEHAVIOR

81

Figure 2 Temperature evolution of B9+ test

Temperature measurements Different kinds of thermocouples allowed temperature measurements of fuel centerlines, cladding, fluid and thermal shroud at different radial locations and elevations. In the “low temperature areas” (lower than 1620 K), type K thermocouples with A1203 insulator were used. Seaths were of two types: stainless steel or zircaloy, with respective diameters of 1 and 1, 5 mm. This technology has been previously used during the “Phebus LOCA program” and did not lead to any trouble during SFD program. In the “high temperature areas”, W.Re thermocouples with hafnia insulator were used. The seaths were of three types: - rhenium coated with a rhenium silicide layer - rhenium coated with an iridium layer - “duplex” clad: with a tantalum cladding and a zircaloy over-cladding. The diameter of such thermocouples is about 2 mm. The good behavior of such sensors at high temperature has been obtained by properly choosing the type of materials, the size of the components, and well-qualified manufacturing conditions. These cautions allowed to avoid early failures due to mechanical effects, due to chemical interactions with the coolant (steam) or due to electrical shunting. Nevertheless, one observed that the interaction with molten materials, and mainly with molten zircaloy, was the main phenomena which limited the good working of our sensors: - concerning chemical interaction with the coolant, the oxidation of the cladding is the limiting factor. A good behavior of the sensors was observed in steam atmosphere up to 2300 K (fig. 2), particularly for the claddings with iridium coating and ReSi2 coating. Furthermore, fuel centerline thermocouples which are more protected against steam behave better than the external cladding thermocouples. - Electrical shunting is due to the strong decrease of the electrical insulation of Hf02 as the temperature rises above 2100 to 2400 K. Studies performed on this phenomenon show that no significant effect occurs up to 2300 K. At higher temperature, up to 2700K, the discrepancy between the true temperature and the measurement ranges from 50 to 200°C, depending on the temperature distribution along the wires of the thermocouple. - The chemical failure due to interactions with molten materials occurs mainly during the two “high temperature” tests (above 2300K) due to a large amount of molten zircaloy: above 2300K, the melting of the ReSi2 protective layer leads to an attack of the rhenium cladding by molten zircaloy and to the failure of the sensors. No thermocouple survived above 2400K. The analysis of temperature measurements shows a good consistency between the measurements and experimental phenomena (melting of α-zircaloy for example). Furthermore, we observed a low scattering of the measurements.This allowed for example, to study the radial distribution of temperature in the assembly and to determine a difference of about 100 K between central and peripheral rods at around 2100K. Taking into account the accuracy given by the manufacturer of the W-Re wires and the previous remarks, the accuracy of the temperature measurements at 2200K is estimated at about ± 50K. After the failure of the thermocouples, the estimate of the hot point temperature was done by extrapolation of correlations drawn up at lower temperatures. These correlations take into account the driver core power or temperature measurements in

82

THE PHEBUS FISSION PRODUCT PROJECT

Figure 3 Hydrogen generation during the steam phase of B9+ test.

colder areas: at the bundle outlet, in the shroud. Of course, these methods require some assumptions about extrapolation of the heat losses at high temperature. Hydrogen generation Hydrogen generation due to steam-zircaloy reaction has been measured continuously at the loop outlet by means of a massspectrometer. It has been also deduced from the pressure evolution of uncondensable gases in a temporary storage tank. An example of the consistency of the two techniques is given on figure 3 which shows the hydrogen generation of the B9+ test (see section 4). The last peak corresponds to an escalation of the steam zircaloy reaction. In spite of the good consistency (around 10 %), the accuracy of this kind of measurement is probably not better than 15 to 20 %. Thermalhydraulic conditions This section is devoted to the pressure and mass flowrate measurements. The estimated value of the accuracy of the pressure measurements is around 2 %. This result has been obtained by taking into account the accuracy of the reference pressure transducers located on the exit line (~ 1 %), the error margin due to the signal processing, and the analysis of the consistency with other pressure sensors located in the test devices and on auxiliary circuits. As regard to mass flowrate measurements, the steam flowrate has been determined by measuring the weight of the feedwater tank with a precision weighing machine (accuracy: ~ 0,2 g). The flowrate of injected gases (helium or hydrogen) is measured with calibrated mass flowmeter. The accuracy in the measuring range (0,05 to 0,5 g/s) is around 1 %. The sensors of the injection lines were compared with the signals of sensors located on the outlet line: thermalhydraulic tests under isothermal conditions (533K) with steam led to a good agreement (1 to 2 %) between the weighing system and the value deduced with the turbine, and the pressure and temperature measurements in the outlet line. The agreement was not so fair during experimental tests due to the uncertainty on the gas temperature measurement in the outlet pipe (in case of large difference of temperature between the fluid and the pipe). Comparison between injected gases flowrates and the pressure increase in the uncondensable gases storage tank gives an agreement usually better than 5 %. Bundle power The total power generated in the fuel assembly is deduced from the core power measurement and the coupling factor (ratio of the core power to the bundle one). The calibration of the core chambers has been performed with thermal balances. The scattering of the thermal balances is around 4 % at low power (1 MW) and decreases to 1 % at 7 MW. The coupling factor has been experimentaly determined with a calorimetric method by measuring the bundle heat-up rate due to a power step of the driver core. A synthesis of around 15 measurements (with different test devices) leads to the following values: 209 ± 2 for a 21 fuel rod bundle

PHEBUS SEVERE FUEL DAMAGE PROGRAM MAIN EXPERIMENTAL RESULTS AND INSTRUMENTATION BEHAVIOR

83

Figure 4 Axial power profile in the bundle for different control rods elevations of the driver core

Figure 5 Axial power profile in the bundle at the end of the AIC test

252 ± 3 for a 20 fuel rod bundle and one AIC rod. These results are in fair agreement with the values deduced from neutronic calculations (taking into account the gamma heating). The coupling factor for degraded geometries has been determined by neutronic calculations. The power distribution (axial and radial) in the assembly was determined by dosimetries for intact geometries, and by calculations for intact and degraded geometries. Two examples are given on figures 4 and 5. The first one shows the effect of the driver core control rods elevations on the axial power profile. The figure 5 shows the comparison between the gammascanning performed at the end of the AIC test (see section 4) and the calculated axial profile of power at the end of the test. Due to the test conditions, this profile is representative of the axial profile of energy injected in the fuel and thereby comparable with the gamma scanning. It takes into account the control rods elevations of the driver core and the relocation of the absorber materials in the bundle, determined with the Post Test Examinations. Thermal conductivity of the shroud An important parameter for the test calculations is the thermal conductivity of the shroud in porous zirconia. An effort has been done in order to improve our knowledge about this parameter. The first tests showed a large change of thermal conductivity due to the progressive filling of the porosities with hydrogen coming from the oxidation of the zircaloy liner. This uncontroled change has been avoided in the last tests by using an injection of helium in the porous zirconia. Experimental determinations of heat flux through a 1 cm thick porous zirconia plate have been performed with a temperature of 400 K on the cold side and temperatures up to 2250 K on the hot side. The analysis of the results is still running.

84

THE PHEBUS FISSION PRODUCT PROJECT

4. DESCRIPTION OF THE TESTS The experimental program is divided into two series of tests: the first one (tests B9, B9R, B9+) is mainly devoted to the oxidation phenomena and its consequences on the fuel degradation. This series is characterized by high oxidation rates. The second series (test C3, C3+, AIC) is characterized by low oxidation rates of the cladding in order to study the interaction between the remaining zircaloy and the other materials: UO2 and Inconel for C3 and C3+ tests, AIC and stainless steel of the control rod for the AIC test. Note that the index “+” implies high temperatures: 2600 to 2800 K, and large bundle degradation. A brief description of the tests is summarized below: B9 test (December 86) The steam flowrate decrease (4 to 1 g/s at 0,5 MPa) allowed to get a low heat-up rate (0,2 K/s) up to 1820 K. Then an escalation of temperature started ~ 5 K/s). At about 2150 K a slow cooling down with helium was performed. The global geometry of the bundle has been preserved but crackings and slight deformations of the highly oxidized cladding were observed [1]. B9R test (April 88) This test was planned to be similar to the B9 one except the cooling phase which was required to be performed with a high steam flowrate. Due to an operating incident, this test was carried out in two parts: the first one is similar to the B9 test up to a maximum of 1800 K; the second one was performed in order to reach a suitable temperature of 2150 K, a high oxidation rate, and finally to carry out a “Rapid” cooling down (10 K/s). An unexpected strong escalation of the zircaloy-water reaction occured during this second part [1]. These transients led to damages larger than in the B9 test: large deformations of the cladding: 70 % relative increase of the clad perimeter; highly oxidized clads reduced to powder; break down of fuel pellets; dissolution of fuel due to local zircaloy melting. B9+ test (January 89); OECD International Standard Problem N° 28 The test pressure was maintained at 2 MPa. The scenario is divided into two phases. The oxidation phase with a 2 g/s steam flowrate allowed to reach 2150 K with an axial profile of the oxide layer increasing from few percents (lower part) to around 100 % (upper part). An escalation of temperature occured at the end of this phase in the upper part of the bundle. Then a high temperature phase with 0,5 g/s helium flowrate allowed to reach 2750 K. Dissolution of fuel by molten zircaloy, probably locally enhanced through crucible effect of the zirconia layer was observed. C3 test (October 1987) This test was performed with a very low hydrogen flowrate at high pressure (3,5 MPa). The maximum temperature was around 1900 K. Post Test Examinations (PTEs) showed a slight interaction through solid-solid contact between the U02 and the collapsed cladding, and significant interaction between the inconel grids and the clads. An hydriding of zircaloy has been observed during the heat-up phase with a temperature step of 30 K/s from ~ 850 K to 1350 K [1]. C3+ test (November 1988) After a preoxidation phase (60 µm), the test was carried out with helium at high pressure: 3,5 Mpa. The maximum temperature was around 2500 K. PTEs showed interaction between molten zircaloy and U02, resulting in a large plug at the lower end of the bundle. Interaction between zircaloy and inconel, and fuel dissolution due to this molten alloy were also observed. AIC test (June 1989) This test was performed with a 20 fuel rod bundle and a control rod (Silver—Indium— Cadmium) in the central position.

PHEBUS SEVERE FUEL DAMAGE PROGRAM MAIN EXPERIMENTAL RESULTS AND INSTRUMENTATION BEHAVIOR

85

Fig. 6 Final temperature evolution of B9 test. Increase of heat-up rate due to: 1—step of driver core power 2—escaltion of oxidation reaction

The test assembly was supplied with a 2 g/s steam flowrate at 0,7 MPa. Nuclear power was regulated to follow the specified heat-up rate (less than 1 K/s before escalation) up to the target maximum temperature of 2000 K. The escalation which occured in the upper part of the bundle, resulted in a short length hot zone and a small amount of molten materials ejected out of the guide tube. The control rod degradation occured at about 1675 K (close to the melting point of stainless steel) and the PTEs showed interactions due to the control rod materials with the guide tube and the neighbouring rods. 5. DEGRADATION PHENOMENA The aim of this section is to present general aspects of the degradation phenomena observed during the SFD tests. The three main parts of discussion will be: cladding oxidation, dissolution of U02 through interaction with zircaloy, and interactions due to structure materials. CLADDING OXIDATION The oxidation phenomenon has been studied in all of the tests except in C3. The tests conditions were such that a too early temperature escalation during the heat-up phase would be avoided. The reason of this choice was to facilitate the interpretation. Oxidation phases were carried out with high flowrates (some g/s) and/or low initial heat-up rates (from 0,15 K/ s for B9 test up to ~ 1 K/s for AIC test). Nevertheless, escalations were usually observed at the temperature of zirconia phase transition from tetragonal to cubic: ~ 1820 K. The temperature increase reached ~ 3 K/s (AIC test) and ~ 7 K/s (B9 test)— Fig. 6 -. An unexpected stronger escalation with a 20 to 30 K/s rate occured at lower temperature (1400 K) during the second part of B9R test—Fig. 7 -. The start of a such an escalation is the probable outcome of crackings of the zirconia layer and remaining zircaloy which occured during the cooling phase of the first part of the experiment. The oxidation kinetics of the whole bundle were deduced from hydrogen generation measurement. The steam consumption was usually low: 3 to 10 %, but increased largely during the escalation of B9R test: 70 %. Such an escalation resulted in local starvation as shown by the temperature evolutions (temporary local decrease of temperature)—Fig. 7 -. The analysis of the thickness and of the structure of the zirconia layer also provides informations about oxidation phenomenon. Nevertheless, this kind of analysis is difficult, and sometimes impossible to perform, mainly in C3+ and B9+ because of the “dissolution” of the oxide layer by the remaining zircaloy during the final helium phase. As far as damages are concerned, the PTEs performed after B9 test showed that the global geometry of the bundle has been preserved. Nevertheless, cladding damages were observed in the upper part. The two following slides show what could be the mecanism of the damages: a small local failure of the zirconia layer leads to a detachment of the oxide and a strong oxidation of the zircaloy sublayer (fig. 8). Open cracks and stronger damages (fig. 9) can happen if the oxidation is carried on.

86

THE PHEBUS FISSION PRODUCT PROJECT

Fig. 7 Température evolution of B9R test(2nd part).

Fig.8

Fig.9

The B9R test showed cladding deformations in the region where escalation was starting at the end of the test (fig. 10). One observes penetration of zirconia in the zircaloy sublayer, instead of a regular layer. The “flower” shape of the clad (with failures) is due to cooling down stresses. Figure 11 shows the deformations (up to 70 %) of the full oxidized clads in the area where escalation occured. These deformations are probably due to the volumic expansion of the materials (transformation of zircaloy into zirconia). This effect was enlarged by the structure of the oxide which shows macroporosities (cracks). Such porosities are the possible outcome of the oxidation process by penetration of zirconia into zircaloy sublayer. Strong cooling down turned the zirconia layer into a powder and strong flow blockage occured (fig. 12). INTERACTION BETWEEN ZIRCALOY AND FUEL PELLETS The main objective of C3 test was to study such an interaction through a solid-solid contact. As a result, one has observed nodules with high uranium content in the thickness of the clad. The amount of uranium involved in this interaction is low and does not affect in a large way the rod degradation process. Interaction between U02 and molten zircaloy was observed very locally in B9R test and more extensively spread in C3+ and B9+ tests. The C3+ test was performed with a thin oxide layer of 60 µm which probably desappeared during the helium phase, due to oxygen diffusion from the oxide to the metallic phase. PTE did not show a strong dissolution of U02: analysis of cross cuts

PHEBUS SEVERE FUEL DAMAGE PROGRAM MAIN EXPERIMENTAL RESULTS AND INSTRUMENTATION BEHAVIOR

87

Fig. 10

Fig. 11

Figure 12

show a lack of fuel of about 5 to 10 %. This result is confirmed by microprobe analysis of the frozen materials: In the melt, the mean atomic composition of the two phases (Zrα and [U-Zr] 02) is the following: Zy: 55 %, U: 7,4 %, O: 37 % This represents a mean dissolution of 8 % of the pellets. The PTEs of the B9+ bundle are under study. The present results show an axial profile of oxidation from 10 µm (lower end) to 350 µm at mid plane (elevation+400 mm) and full oxidation in the upper part. In spite of a relatively small amount of available zircaloy in the upper half-bundle, the dissolution rate (lack of fuel) reached around 10 to 14 % between elevations 480 and 580 mm from fuel bottom. The maximum temperature at these elevations was around 2600 K to 2750 K—Fig. 13 -. Local dissolutions reached higher value (up to 40 %). This could be due to strong local escalations which occured at the end of the oxidation phase. The zirconia layer behaved probably as a temporary crucible containing the molten zircaloy. The cuts

88

THE PHEBUS FISSION PRODUCT PROJECT

Figure 13: B9+ test—Cross cutting at elevation+518mm

Fig. 14 AIC test-elevation+600mm

show that the zirconia layer in the highly oxidized part of the bundle is locally very thin. This could be due to the dissolution of zirconia by molten zircaloy. INTERACTIONS DUE TO STRUCTURE MATERIALS Interaction Inconel-Zircaloy The C3 test was carried out without a zirconia layer and the interaction between the inconel spacer-grids and zircaloy resulted in an almost full dissolution of the cladding in the grid region. The damages extended significantly on 10 to 15 cm below the grid. A global evaluation shows that the zircaloy mass involved in such phenomena is about 2 to 3 times the inconel mass. C3+ test also showed the same kind of interaction during the helium phase, leading to a slight outcoming dissolution of U02. Microprobe analysis shows around 0,07 at. of U for 1 at. of Zr in the frozen inconel alloy. On the other hand, the B9 test did not show any large interaction. The computed oxide layer thickness at grid melting time is around 100 to 150 µm. It probably prevented a significant interaction. Interaction due to control rods materials PTEs of the AIC test are still underway. Nevertheless, the first results show a significant attack of the neighbouring claddings in spite of the small amount of molten alloys ejected out of the guide tube (10 to 15 cm3)—Fig. 14 -.

PHEBUS SEVERE FUEL DAMAGE PROGRAM MAIN EXPERIMENTAL RESULTS AND INSTRUMENTATION BEHAVIOR

89

Fig. 15: Axial cut of the lower end of C3+ bundle Fuel pellets (4) Relocated materials: UO2-ZrO2 power (1); U-Zr-O melt (2); Inconel-Zr alloy (3)

PLUGS FORMATION The molten alloys observed during C3, C3+ and B9+ tests, resulted in large plugs at the lower ends of the bundles: total clogging of central channels and sometimes partial clogging of peripheral ones occur ed. Figure 15 gives an exemple of the relocated alloys in C3+ test. 6. CONCLUSION The last of the six tests of the present SFD program has been performed in june 1989. The analysis of each tests and comparisons between the different results is still running. All these efforts lead to the elaboration of a major data base to improve the understanding of main processes associated with severe core damage accidents. The tests showed that: - oxidation process can lead to the failure of the clad i.e. to a double-side oxidation. Deformations of the clads can occur and lead to a larger area of contact between steam and zircaloy. The strong cooling of such highly oxidized clad turns the zirconia layer into powder and results in a strong flow blockage, - the mass of zircaloy involves in the interaction with inconel spacer-grids can reach 2 to 3 times the mass of inconel. This molten alloy lead to a slight dissolution of the U02 pellets. The protective effect of a zirconia layer has been observed, - the interaction between U02 and solid zircaloy does not affect the degradation process of the core in a large way, - the average value of the dissolution rate of U02 with molten zircaloy did not exceed 14 %. Local values can be higher: up to 40 %, - frozen materials led to large plugs at the lower ends of the bundles. Such experimental results provide a quantitative data base concerning oxidation process and interaction between materials for the ICARE code development and verification [2, 3, 4]. Unexpected aspects of some phenomena (such as cladding failures and cladding deformations due to oxidation process) occured during the tests. The informations are too limited to deduce general laws or criteria usable for codes, but they can be used as parameters for sensitivity studies (two sides oxidation or not, with intact or deformed cladding). The present SFD data base will be completed with the future FP program concerning the FP release and the degradation phenomena with irradiated fuel under more severe conditions up to the melting of U02 (3100 K).

90

THE PHEBUS FISSION PRODUCT PROJECT

REFERENCES 1 2 3 4

. C.GONNIER, S.FABREGA, E.SCOTT, G.GEOFFROY “In pile investigations at the Phebus facility on the behaviour of PWR-type fuel bundles in severe accident conditions beyond the design criteria” SFEN—NUCSAFE 88—Avignon 2, 7 October 1988. . R.CHATELARD, R.GONZALEZ, A.PORRACCHIA “Analysis of the TMI-2 core degradation using ICARE code”. 27th ASME/ AICHE/ANS National Heat Transfer Conference—Minneapolis July 1991. . BOURDON, VILLALIBRE, ADROGUER, GEOFFROY “Analysis of the severe fuel damage test Phebus B9+ using ICARE code”. 27th ASME/AICHE/ANS—Minneapolis July 1991. . C.GONNIER, G.GEOFFROY, B.ADROGUER “Phebus severe fuel damage program -Main results”. Safety of thermal reactors— ANS PORTLAND July 1991.

REVIEW OF B9+ BENCHMARK RESULTS B.ADROGUER* and P.VILLALIBRE** * Atomic Energy Commission (CEA) IPSN/DRS Cadarache, France **Consejo de Seguridad Nuclear Madrid, Spain

SUMMARY The Nuclear Safety and Protection Institute (IPSN) of CEA has proposed the PHEBUS Severe Fuel Damage B9+ test as the basis for an OECD/CSNI International Standard Problem. The objectives of the test were to study the following main phenomena occurring during a SFD accident of a PWR: the cladding oxidation, the mechanical behaviour of the cladding with a variable ZrO2 thickness layer containing molten zircaloy, the dissolution of UO2 and ZrO2 by molten Zr and relocation of the melt. These phenomena being crucial in the framework of the development and assessment of the codes, the B9+ test has been accepted with semi blind conditions as the ISP 28 exercise. Measured thermal-hydraulic conditions have been supplied to help participants to calculate thermal initial conditions as correctly as possible in order thoroughly to evaluate the bundle degradation (blind part). The calculations can be performed using two kinds of radial boundary conditions (BC): a constant temperature applied on the external cold surface of the shroud (first way) or temperature evolutions versus time at different levels of the internal surface of the shroud (second way). Results of 17 calculations were submitted by 15 different organizations from 12 countries. Eight different codes have been used and four calculations were performed using the second BC method. Two groups of calculations can be distinguished. The former performed with SCDAP Mod1, SCDAP Mod2, MELCOR and ICARE2, described the bundle degradation up to the final freezing of relocated materials. The latter, including ATHLET SA, KESS III, FRAS-SFD and MARCH3, was limited to the calculation of the thermal and oxidation behaviour of the bundle. All the first way BC calculations performed using the measured ZrO2 porous conductivity led to an overestimation of the rod temperatures. Correct thermal and oxidation behaviour of the rods were only calculated when the thermal leakage through the shroud was increased and adjusted. The capabilities of codes in calculating the main degradation phenomena have been clearly illustrated and weaknesses concerning the modelling of some degradation processes have been identified. The more severe limitations concern the UO2—ZrO2 dissolution by molten Zr, the solubility limits in the resulting U-Zr-O mixture and the cladding failure by the molten mixture. There is a lack concerning the inconel spacer-grid interactions with the rods and the dissolution of ZrO2 by molten Zr. Verification of reactor codes on small scale experiments need more code versatility. More verification and validation activities remain necessary for severe fuel damage codes. This exercise illustrates also the importance of the code user, the need for improved user guidelines with more detailed information and recommendations and the need for experts in core degradation phenomena to make more effective the use of the codes in this field. 1. INTRODUCTION The PHEBUS B9+ test has been selected for the International Standard Problem number 28. This test is one of the six tests of the PHEBUS SFD programme [1]. The acquisition of data from this programme and from complementary separate effect experiments is underway. The corresponding database which is used for the verification of the ICARE code addresses nearly all aspects of the early phase of core degradation: Zr oxidation and its corresponding aspects concerning the hydrogen generation and the cladding mechanical damages; interactions between Zr and fuel pellets, Zr inconel spacer-grids and absorber materials (Ag-In-Cd) and fuel rods; relocation of molten material and flow-blockage formation. This SFD database will be completed with the forthcoming PHEBUS FP programme by the study of the degradation of irradiated fuel and the FP release and transport.

92

THE PHEBUS FISSION PRODUCT PROJECT

Fig. 1. Bundle cross-section (1) Zr liner; (2) porous ZrO2; (3) dense ZrO2; (4) stainless steel tube

The B9+ test conducted in January 1989 was aimed at studying cladding oxidation, UO2 and ZrO2 dissolution by molten Zr and relocation in the absence of control rod materials with a nuclear heated bundle having an axially variable cladding oxidation. These phenomena being crucial for the code assessment, this test has been proposed to the OECD/CSNI for the semi blind International Standard Problem ISP 28, in which only the temperature data have been supplied to participants. Results of 17 calculations were submitted by 15 different organizations from 12 countries indicating strong interests and cooperative efforts by various institutions. Eight different codes have been used. This paper is a first review of the ISP 28. It illustrates the current capabilities of the codes to calculate some main core degradation phenomena. Code weaknesses and limitations have been identified. A large user effect has been shown. 2. THE PHEBUS SFD FACILITY AND THE B9+ EXPERIMENT The PHEBUS SFD facility allowed to reproduce thermal-hydraulic conditions relatively close to what would exist in core under beyond design basis conditions. Careful attention has been devoted to the monitoring of the bundle power and to the thermal-hydraulic conditions. The principle of the tests consists in supplying a bundle of 21 PWR-type fresh fuel rods with a flow rate mixture of overheated gases (steam-hydrogen or helium). The fuel power is adjusted according to the required temperature evolution. The facility mainly consists of a driver core supplying the test fuel with nuclear power; a SFD loop (located on the vertical axis of the core) which reproduces the thermal-hydraulic conditions; and a pressurized water loop (8 MPa, 528 K) working as an independent cooling circuit of the SFD loop. This SFD loop carries the fluid flow to the test train through three injection lines. The test train cross-section shows the bundle which consists of 21 unirradiated UO2 fuel rods in a 12.6 mm pitch matrix (Figure 1). The total fissile height is 0.8 m. Two spacer-grids in inconel 560 mm apart are located on either side of the mid plane. The ballooning of the pressurized rods with He was avoided by the low temperature melting of the fusible seal on the rod top plug (melting point—1070 K). The insulating shroud of the bundle is a multi-layer structure. The octagonal inner zircaloy liner 0.6 mm thick is surrounded by a thick porous ZrO2 layer (94 mm external diameter and 69.5 mm between faces of the inner octagonal surface), a dense ZrO2 layer 1 mm thick and an external stainless steel tube 8 mm thick. Different kinds of thermocouples enable the temperature measurements of fuel, cladding, fluid and shroud at different radial locations and elevations. The hydrogen concentration at the loop outlet is measured continuously by means of a mass spectrometer. Detailed information on the facility are given in ref.[2].

REVIEW OF B9+ BENCHMARK RESULTS

93

Fig. 2. Experimental scenario

3. DATA SPECIFICATIONS OF THE ISP 28 A first workshop was held in Cadarache in April 1990 to give the technical specifications of the test. Detailed descriptions of the B9+ test device and associated test instrumentation were given [2, 3]. The boundary conditions were specified with some recommendations. All the measured temperature evolutions within the test device were given to the participants. Parameters to be calculated and reporting format were also specified [4]. The main characteristics and specifications of the B9+ are summarized in this section. The experimental scenario has been divided into three parts (Figure 2). The first oxidizing phase performed with pure steam flow enabled to obtain a deep axial profile of ZrO2 in the cladding during a temperature increase by steps up to 1800 K. The second heat-up phase in pure helium atmosphere led to a maximum rod temperature of about 2750 K. The final phase performed with step reductions in nuclear power enabled a slow cooling down in order to keep unchanged the previous bundle geometry. The nuclear power versus time and elevation was well defined from measurements and neutronic calculations. Radial Boundary Conditions in the Shroud Two kinds of radial boundary conditions (BC) were proposed to calculate the ISP 28 [4]. First way: designed to represent as closely as possible the different layers of the shroud surrounding the 21 rods bundle. The external water cooling of the shroud maintained the steel tube surface at a constant temperature of 528 K all along the shroud for the entire test. A key parameter for the test calculation is the thermal conductivity of the porous ZrO2 layer of the shroud. An effort has been made in order to improve our knowledge about this parameter. This porous ZrO2 layer was fed with He during the oxidation phase in order to limit the variation of its thermal conductivity by H2 entering from the liner oxidation, and laboratory tests have been performed on different representative ZrO2 samples under He atmosphere and at temperatures in the range of 523 K and 2373 K. The correlation deduced from the experimental data has been recommended in the specifications [3]. Second way: in order to avoid the modelling of the actual shroud a simple surrounding structure was proposed involving only the internal octagonal liner. Temperature evolutions have been given at ten different levels [4]. These radial BCs have been defined by a best estimate ICARE2 calculation performed in modelling the actual shroud. This calculation has been ‘adjusted’ in order to find a correct thermal behaviour all along the bundle for the entire test. The tuning parameter used was the porosity of the porous zirconia. It was pointed out that these radial BCs were ‘code dependent’.

94

THE PHEBUS FISSION PRODUCT PROJECT

4. MAIN PHENOMENA OBSERVED After three power steps, a slow increase of the power during about one hour resulted in a slow heating to about 1800 K at the hot zone (60–70 cm from the bottom of the fissile length). A new power step allowed an oxidation run-away resulting from the enhancement of the oxidation kinetics when the rod temperatures reach 1850 K (ZrO2 tetragonal-to-cubic phase transition). This escalation enabled to reach the melting temperature of α-Zr(O) during about 150 s just before the steam to He switch (Figure 2). Then, the resulting quick stopping of the oxidation led to a local cooling of the rods hot zone with a temporary refreezing of the remaining non-oxidized Zr. The following heat-up phase in pure He enabled to reach a maximum temperature of about 2700 to 2750 K. Thermocouple readings became unreliable at different times due to interaction between cladding sheaths and molten materials. Post test examinations (PTE) on radial cuts enabled measurements of the cladding oxide thickness at different elevations [6]. Due to complex and non-uniform interactions in particular between molten U-Zr-O materials and the remaining cladding ZrO2 layer on its internal and external surfaces, only microprobe analysis enabled to determine a mean axial profile which goes from a few percent under 20 cm to 40% at 40 cm and complete or quite complete oxidation at the hot zone (60–70 cm). The hydrogen total mass production is 39.5 g (uncertainty of 20%). The upper spacer-grid (66 cm) has totally disappeared and the lower grid (14 cm) disappeared partially in the central zone. Few debris were located on it. No significant interaction with cladding was observed. Fuel dissolution by molten Zr was observed with a fine ‘meshing’ of 2 cm between radial cuts. The weight fraction of UO2 dissolution has been deduced from the area of the remaining pellet sections. An important spreading of dissolution was found near the hot zone (from a few % to 40% at 56 cm), so only a mean profile has been determined with an uncertainty evaluated to 1.5%. Low mean UO2 dissolutions have been found in the central zone (28–60 cm) with a maximum of 14% at 52 cm. The external rods are on average more dissolved than internal ones. In the hot zone significant local dissolution was only observed on some rods. This proves that oxidation was not quite complete on these rods at the end of the oxidation phase or that some dissolution occurred during the 150 s period of escalation during which the remaining Zr was molten. Local relocation of the non-oxidized Zr from the upper parts flowing down during the He phase between the cladding and the pellets could also explain this local UO2 dissolution in the hot zone. Zirconia dissolution could not been quantified. Azimuthal ZrO2 layer thickness differences on the same rods were observed with a minimum thickness about three times lower than the maximum thickness. These very local differences could result both from the azimuthal temperature differences and from local ZrO2 dissolution. The latter effect which seems the more important one was mainly observed under the hot zone at the upper part of the cladding failure zone. During the He phase, relocation of molten material occurred and two main plugs were obtained. The lower one around 5 cm is a metallic plug including inconel with the following atomic content: 55.4% Zr; 7.5% U; 23.6% Ni; 7% Fe; 1.5% Cr; 5% O. The upper one, localized between 16 and 28 cm, is a ceramic plug including αZr(O) and (U, Zr)O2-x phases (34.6% Zr; 6.6% U; 58.8% O). Local deposits of small amounts of mixtures rich in oxide phases have been found mainly between 28 and 40 cm and locally as far as 60 cm. A maximum bundle blockage of 50% was found in the upper plug at 26 cm, the minimum of 35 % was found at level 46 cm (the bundle blockage is defined to be equal to 37% at time zero due to undeformed rods). The cladding disappearance zone occurred mainly between 28 and 50 cm but local cladding failures were found as far as 56 cm on external rods. The liner was characterized by strong deformations above 25 cm with local contacts with external rods. A partial or total disappearance zone was observed between 25 and 75 cm. 5. OVERVIEW ON THE COMPARISONS Detailed information is given in the preliminary comparison report [7]. The main characteristics of the 17 submissions (Table 1) are summarized in this section. Eight codes have been used but only SCDAP Mod1, SCDAP- RELAP5 Mod1 and Mod2, ICARE2 V1 and V2p and MELCOR V 1.8.0 and V.1.8.EA are able to take into account the material relocation. KESS III, ATHLET SA, FRAS-SFD are under development concerning relocation processes and their calculations were mainly limited as well as the MARCH3 calculation to the prediction of the thermal and oxidation behaviour of the bundle. It must be pointed out that 16 submissions have been performed in semi blind conditions. The FRA-SP calculation has been performed in open conditions. This participant was involved in the definition and the analysis of the PTE and this submission is apart from the ISP 28 and should be considered only as a reference calculation.

REVIEW OF B9+ BENCHMARK RESULTS

95

Table 1. Characteristics of the Submissions PARTICIPANT (COMPANY)

CODE

CODE MOD IF.

CONDUCTIVITY POROUS ZrO2

CLAD. FAILURE CRITERIA Tc (K); eox µm)

MEAN ∆t (s)

FRA-SP/CEA-PEC SP-CSN/PEC SP-UPM/PEC CZECHO-NRI GER-IKE1 GER-IKE2 GER-GRS1 GER-GRS2 HUNGARY/CRIP I-PISE/ENEA ISPRA JAERI NETHER/ECN SWEDEN/Studsvik TAIWAN/AEC UK/AEA USA/NRC

ICARE2V2p SCDAP Mod1 HELCOR V180 FRA-SFD KESS III

Yes Yes No Yes Yes

Tc>2300; eox<300 Tc>2500; eox Tc>2200; eox<100 Tc>2400; eox<300 no explicit model

11.3 0.1 0.1 7.0 6.4

ATHLET SA

No

no explicit model

1.3

MARCH 3 SCDAP/R5/M2 ICARE2 V1 SCDAP/R5/M1 SCDAP/R5/M2 SCDAP/R5/SRL MELCOR V180 SCDAP/R5/M2 MELCOR V18EA

Yes No No No Yes No No Yes Yes

Miss, (tuned) Missenard λ=2×λ mes 2nd WAY λ=2×λ mes 2nd WAY λ ~ Missenard 2nd WAY λ tuned λ>2×λ mes λ>3×λ mes λ ~ 3×λ mes λ=2*λ mes λ>2*λ mes 2nd WAY MATPRO Missenard

Tc>2200; eox Tc>2500; eox<540 Tc>2300; eox<500 Tc>2700; eox Tc>2500; eox<520 Tc>2500 eox<435 Tc>2600; eox<100 Tc>2500; eox>520 Tc>2200 if eox<60 Tc>2500 if eox>60

0.015 0.015 15.9 0.07 0.03 0.14 0.94 0.043 0.80

PEC: Phebus Espana Consortium; Miss: Missenard model Tc; eox: Cladding temperature and ZrO2 thickness limits

Input data deck characteristics In general a common philosophy has been used for the nodalization of the bundle. One hydraulic channel and three different rods (central, 1st ring and 2nd ring rods) have often been used with eight to 11 axial meshes. Two participants (USA, FRA-SP) made use of three parallel and independent channels. Some problems have been met for the shroud modelling. Some codes were not able to take into account directly the multilayer shroud and some modifications (MELCOR) or modelling compromises (FRAS-SFD, MARCH 3) had to be made. A difficulty has also been found to take into account the evolution of the power profile versus time (constant in MELCOR and limited at three different profiles in SCDAP/R5). Codes usually do not allow He as a cooling gas, so an equivalent H2, air or steam cooling was used instead of He. When a pure non-condensible gas was used, a small fraction of steam was necessary to avoid a code failure or some inconsistencies (MELCOR, SCDAP/R5) and numerical instabilities occurred and led to severe code time step reductions by a factor greater than 5 or 10 (MELCOR, SCDAP/R5). This led to very low mean time steps (Table 1) and to large CPU times between 36 h (CRAY XMP 14) to 306 h (SUN 4/300) mainly consumed up to 80 to 90% to calculate the He phase. The fuel dissolution by molten Zr is only taken into account by ICARE, SCDAP, SCDAP-RELAP5 and FRAS-SFD and simultaneous ZrO2 dissolution by molten Zr is only modellized by ICARE 2 V2p. The inconel spacer-grid rod interaction was not modelled in the codes used. Only the spacer-grid melting and relocation were considered by MELCOR. A great consistency was found concerning the oxidation kinetics correlations used. Cathcart-Prater up to 1853 K and Urbanic Heidrick correlations at higher temperatures were the more often used. A diffusion model is used by FRAS-SFD. More differences between the codes concern the cladding failure criteria and relocation models when the non-oxidized Zr of the cladding melts and dissolves simultaneously the UO2 pellets and the external ZrO2 layer of the cladding. The resulting U-Zr-O mixture is contained within a ZrO2 shell, which is supposed to fail when a user-specified temperature is reached (2200 K < T user limit < 2673 K) provided that the oxide shell is not considered sufficiently thick to hold in the molten mixture. The user failure thickness value is spread between 0 and 540 µm depending on the calculation (Table 1). Thermal behaviour The calculated energy balance (Figure 3) shows that approximately 70% (during the steam phase) to 90% (during the He phase) of the nuclear power is transferred to the shroud [5]. The remaining energy is transferred to the fluid by convection and radiation absorption by steam. So the calculated rod temperatures are mainly dependent on the prediction of the radial heat flux through the shroud.

96

THE PHEBUS FISSION PRODUCT PROJECT

Fig. 3. Energy balance in the bundle

The use of the thermal conductivity of the porous ZrO2 layer of the shroud which was recommended in the experimental specifications [3] always gave calculated rod temperatures in excess of those found in the experiment. All participants who made a first way calculation had to choose a higher porous ZrO2 conductivity between two to three times greater than the experimental recommended value. The tuning of this parameter was performed by some participants in order to achieve a better fit of the measurements and obtain good initial conditions for the different degradation phenomena (USA, I-PISE, FRA-SP). In particular, FRA-SP adjusted at each level the conductivity by tuning the porosity of the porous ZrO2. A general good agreement is obtained with the different codes (Figures 4, 5 and 6). Significant discrepancies are due to incorrect calculation of the shroud thermal leakage, underestimated by SP-CSN, NETHER and SP-UPM (only in the He phase) and overestimated for JAPAN. Measured radial temperature differences are respectively about 5 to 10, 60 to 90 and 150 to 180 K between the three kinds of rods and the shroud liner at the end of the steam phase. Correct predictions were obtained with ICARE2 V2p and MELCOR V1.8.EA in using 3 hydraulic channels, and with SCDAP/R5/Mod2 and ATHLET SA in using one channel. The underestimation of these radial gradients by SCDAP/R5/ Mod1, ICARE 2 V1 and MELCOR V1.8.0 is partially due to the modelling of one mean channel which tends to reduce the temperature differences by convective and radiative exchanges with the fluid. Compensation effects by the cladding and liner emissivities explain probably the large radial temperature differences calculated in using one mean channel. Differences in the degradation of each kind of rod results from the temperature differences calculated between the rods. Participants who calculated correctly rod temperatures at the hot zone before 8000 s made in general a correct prediction of the following temperature escalation observed in the test at 60/70 cm (USA, UK, FRA-SP, I-PISE, HUNGARY). In spite of an overall correct temperature prediction GER-GRS, GER-IKE2 and TAIWAN did not predict this escalation. Due to the overprediction of the temperatures at the hot level, SP-CSN, GER-IKE, NETHER and CZECHO calculate the escalation around 7000 s, sooner than in the test. In taking into account a correct heat flux through the shroud and a correct evolution of the axial power profile I-PISE, UK and FRA-SP with SCDAP/R5 and ICARE2 made a correct calculation of the maximum temperature estimated at 2700 to 2750 K. The use of a power profile constant with MELCOR led to underestimate this value (TAIWAN, USA). Oxidation and H2 production During the short oxidation escalation at 8200 s, the remaining non-oxidized Zr of the cladding in the hot zone (about 50% of the initial value) is quickly consumed up to a quite complete oxidation. In these conditions a correct calculation of the oxidation requires a very accurate prediction of the temperatures just before the threshold phenomena of the escalation. Cladding oxidation histories at 50 and 60 cm (Figures 7 and 8) show an important spreading of the calculations mainly due to the calculated temperature differences. This spreading is larger at 50 cm in the ‘transition zone’ of the escalation. Axial oxidation profiles (Figure 9) at the end of the steam phase have been very well predicted by ICARE 2, MELCOR and SCDAP/R5 when both temperature histories at each level and temperature escalation at the hot level were correctly predicted. The total oxidation at the hot point (60 cm) is well calculated by quite all the participants except ISPRA, TAIWAN, SP-UPM, SWEDEN and JAPAN. This is not necessarily the indication of a bad thermal calculation. For instance the rapid increase of

REVIEW OF B9+ BENCHMARK RESULTS

97

Fig. 4. Fuel temperature versus time—Level 60 cm—2nd ring rods

Fig. 5. Fuel temperature versus time—Level 60 cm—2nd ring rods

the oxidation can be stopped before being complete by the steam-He switch if temperatures just before the escalation are slightly underestimated (TAIWAN). The residual steam keep by some participants using MELCOR and SCDAP/R5 to avoid a code failure in the pure He phase led to a weak increase of the oxidation in this phase. When a cladding failure is predicted with MELCOR and ICARE2, the dislocated ZrO2 layer can slump with the molten Zr but it is kept in place with SCDAP/R5. The spreading of the H2 production (Figure 10) results from the differences in the oxidation predictions. Only four calculations are within the experimental uncertainty range (32/47 g). During the temperature escalation the remaining β-Zr is transformed into α-Zr(O) and the melting temperature of the α-Zr (O) is reached during about 150 s. These transformations could decrease the oxidation of the cladding. This might explain a tendency of the calculations which do not take into account these effects and overestimate the oxidation in the upper part of the bundle and the corresponding H2 release. Part of the H2 overestimation in some calculations could also be due to the underestimation of the radial temperature gradients. UO2 and ZRO2 Dissolution by Molten Zr As the temperature rises, the non-oxidized Zr melts and dissolves the UO2 and the cladding ZrO2 layer (observed in the test from 30 to 70 cm). This phenomenon is strongly a function of the temperature but also of the oxidation profile which determines the mass of molten Zr available and its oxygen content. The UO2 dissolution at one level is limited by its solubility

98

THE PHEBUS FISSION PRODUCT PROJECT

Fig. 6. Fuel temperature versus time—Level 60 cm—2nd ring rods

Fig. 7. Oxidation versus time—Level 50 cm—1st ring rods

limit in the local U-Zr-O mixture and by the failure of the ZrO2 scale which governs the flow-down of the mixture and the end of the dissolution. Only ICARE2, SCDAP Mod1 and SCDAP/RELAP5 Mod2 gave final results of the fuel dissolution. The FRA-SFD code began a calculation of the dissolution at 75 cm but the calculation could not be continued in the He phase (Figure 11). There is no true dissolution model in MELCOR but a parametric model enables to remove and transport a user-specified fraction of the UO2 when cladding failure by molten Zr is predicted. Only ICARE2 V2p predicted a fuel dissolution in agreement with the test. In general, the fuel dissolution was significantly overpredicted at each elevation where the non-oxidized Zr melted during the He phase. Two main reasons might explain the differences between ICARE2 V2p and SCDAP calculations: the cladding failure temperature (2300 K for ICARE2 and 2500 K for SCDAP) and the solubility limit of UO2 in the U-Zr-O mixture. The first one in the ICARE2 prediction tends to limit the UO2 dissolution in the central zone due to an early cladding failure, the second one tends also to limit the UO2 dissolution due to the calculation by ICARE2 of a less important UO2 solubility in the mixture. A third reason concerns the ZrO2 dissolution. Only ICARE2 V2p takes into account this phenomenon which tends also to reduce the simultaneous UO2 dissolution. The calculated ZrO2 dissolution was weak and led to a limited reduction of the ZrO2 thickness of 10% of a totally oxidized cladding (Figure 7).

REVIEW OF B9+ BENCHMARK RESULTS

99

Fig. 8. Oxidation versus time—Level 60 cm—1st ring rods

Fig. 9. Oxidation profile at 9000 s

Relocation and Bundle Blockage Molten material relocations have been calculated by ICARE2, MELCOR, SCDAP Mod1 and SCDAP/R5. A significant spreading of the results, even with the same code, was observed on both the axial extent of the rod failure zones (Table 2) and on the resulting material relocation (Figure 12). Only the reference calculation (ICARE2 V2p) gave reasonable results on these two points. In general, SCDAP/R5 and MELCOR predicted the refreezing zones at too low elevations or under the active fuel length. This trend was also observed by I-PISE and UK in spite of a correct prediction of the cladding failure zone. As already mentioned, part of the discrepancy with the measured blockage came from the overestimation of the UO2 dissolution. The final freezing zone is calculated by melt progression models based on mass and energy equations which need some userspecified parameters as the heat transfer coefficient, the contact surface between molten material and the solid support or as the mixture properties (solidus temperature). So, the choice of these parameters might explain the prediction of too low blockage zones. Table 2. Cladding Relocation

The relocation of the spacer-grids was only calculated by MELCOR which found refreezing under the active fuel length or in the first mesh (USA) as in the test.

100

THE PHEBUS FISSION PRODUCT PROJECT

Fig. 10. Total H2 release versus time

Fig. 11. Mean UO2 dissolution profile—14000 s

The prediction of the cladding (or liner) failure zones is more often governed by two user-specified criteria. The first correspond to the debris bed formation, e.g. when the melting temperature of ZrO2 is reached or when the lower support has already disappeared (MELCOR V 1.8.EA), the second corresponds to the ZrO2 layer failure by the molten Zr. The latter is often very different between the calculations (Table 1). It includes usually two user-specified limits, a failure temperature and a ZrO2 thickness. For given temperature and oxidation conditions the failure zone increases when both the ZrO2 limit increases and when the temperature limit decreases. In using the same criteria, the extent of the failure zone depends of course on the prediction of the rod temperature and oxidation profiles. So, differences in the failure zones can result in differences in one or several of all these previous parameters. Differences between the two SCDAP/R5 calculations performed by UK and NETHER result mainly from differences in temperature and oxidation profiles. The overestimation of the temperatures by NETHER has also led to predict the melting of the ZrO2 and the formation of a particulate debris bed not observed in the test. This was also found by the SP-UPM calculation for the same reason. The two ICARE2 calculations differ significantly, because cladding failure parameters and temperature and oxidation profiles are different. The ISPRA calculation is characterized by an underprediction of the oxidation and by the choice of a ZrO2 thickness limit greater than the maximum calculated one. This leads to overestimate the failure zone which includes all the Zr melting zone. In some calculations (JAPAN, SWEDEN, TAIWAN) cladding failures were not predicted due to the underprediction of the temperatures and/or to a choice of a temperature failure limit greater than the maximum temperature reached.

REVIEW OF B9+ BENCHMARK RESULTS

101

Fig. 12. Bundle blockage profile—14000 s

Compensation effects between the parameters which affect the relocation can also lead to some acceptable results. This was the case with SCDAP Mod1 which predicts part of the failure and refreezing zones correctly in spite of a significant overestimation of the temperature and oxidation conditions. In the series of first way calculations, the liner failure and its relocation were only predicted by ICARE2. Some other participants have imposed to keep it in place by the input data. TAIWAN in using a second way calculation predicted its melting but overestimated its relocation. Few results were given concerning the mixture compositions always considered by the codes as the addition of the different relocated materials. A poor agreement was obtained with the test. 6. PRELIMINARY CONCLUSIONS The large number of participants and the use of different codes allow an overview on the ability of the codes to predict some governing phenomena of a PWR core degradation and on the ability of different users to use these codes efficiently. The semi blind nature of the ISP 28 must be considered concerning the comparisons with the experiment. The thermal-hydraulic behaviour was given to the participants. The degradation aspect of the bundle was the blind part of the exercise, so ‘usertuning’ to achieve a better prediction of the degradation was not performed. One exception concerns the FRA-SP submission. This participant was involved in the analysis of the post-test examinations. So, this submission has been considered as a reference calculation apart from the ISP 28. From the exercise the following conclusions can be suggested. The objectives of the B9+ test were reached. Different axial oxidation states, UO2 dissolution and relocation of the bundle with freezing of the liquefied mixtures on the lower part of the bundle have been characterized. In these fields the B9+ experiment constitutes a major data bank. The drawback of the test concerns the radial thermal leaks through the shroud. The laboratory measured values of the thermal conductivity of the porous ZrO2 were not suitable to be used in the actual test configuration. This unknown parameter represents a handicap for code assessment because bundle temperatures were found to be very sensitive to it. The general thermal behaviour was at least qualitatively well predicted by most of the participants. • In using the ‘first way’ for the BC all calculations showed that the prediction of the rod temperatures needed a ‘useradjusted’ ZrO2 conductivity greater than the recommended one by a factor 2 to 3. In spite of the open character of the thermal part, few participants seem to have performed a fine tuning of this conductivity in order to improve the prediction of the initial conditions of the degradation. • In using the ‘second way’ for the BC all the participants found correct rod temperatures with MELCOR, ATHLET SA and KESS III codes.

102

THE PHEBUS FISSION PRODUCT PROJECT

An oxidation run-away was measured locally at the hot point just before the end of the oxidation phase. In spite of the difficulty to calculate a rapid oxidation increase of about 50 to 100% with a melting of the non-oxidized Zr, calculations with an accurate prediction of the rod temperatures in the upper part of the bundle just before the oxidation run-away gave more often a correct prediction of the temperature and oxidation escalation (SCDAP/R5 Mod2, MELCOR, ICARE2). But there is a slight trend to overestimate the oxidation in the upper part and the H2 release. Apart from the reference calculation, the degradation processes (fuel dissolution, cladding failure, relocation of molten mixtures and final freezing) were generally not well predicted and show modelling weaknesses. Nearly all codes show the same trend in the deviations with the post test examinations. • The fuel dissolution by molten Zr is overestimated. The solubility limit, when calculated, is given by rough models of the ternary phase diagram of the U-Zr-O mixture. • The ZrO2 dissolution is not modelled except in the latest version of ICARE2. This effect which could not be quantified in the test tends to reduce the protective ZrO2 layer and to decrease the UO2 solubility. The calculated ZrO2 dissolution is weak. • Only simple criteria based on user-specified parameters have been used to predict the cladding failure by molten Zr. The diversity of conditions and parameters used shows clearly a lack of knowledge in this field and corresponding deficiencies in the modelling. The B9+ suggests that the best criterion for cladding failure is: Tc > 2300 K and eox < 450 µm. • Freezing zones of molten materials are generally calculated too low (SCDAP/R5 Mod2, MELCOR, SCDAP Mod1). • Inconel spacer-grid models including the interaction with rods do not exist. Only MELCOR described the spacer-grid melting and relocation. Various results were produced by different participants using the same code. In this ISP 28 the user effect on the results is mainly due to the choice of some degradation parameters. The main ones are the cladding failure parameters which can have tremendous effects on the relocation. The relocation models also need user-specified parameters to define the heat exchange with the solid support (ICARE2, SCDAP/R5). • User effects could be reduced if more mechanistic models verified on separate effect tests were available or if more detailed information were given in the user guidelines including the sensitivity of the user-specified parameters on the degradation phenomena. • User effect depends also to a large extent on user experience and on the time and means devoted to perform the necessary sensitivity studies. Some participants recognized that they were not familiar with the code used and with the sensitivity of some parameters on the degradation results. A good knowledge of the code, of its physical models and of the phenomena governing the degradation would help the user to improve preparation of code input data with more relevant values for some sensitive user-specified parameters. Excessive running times have been communicated by some participants, well in excess of a few hours (SCDAP/RELAP5 Mod2). Whatever computer is used, these large computing times are a real handicap for sensitivity studies. Efficient calculations have been performed with MELCOR and ICARE with more reasonable CPU time. The significant differences obtained with the experiment and between different codes illustrate an important need for code verification in the field of core degradation phenomena. This domain governs important aspects of safety studies: H2 release and remaining non-oxidized Zr, core channel blockage, fission products release greatly dependent on high temperature evolutions. • Recommended values for user-specified parameters which have very sensitive effects on the degradation processes should be specified from a larger set of small scale experiments. • Codes have been developed for reactor calculations but verification and validation activities are necessary. So, more code versatility is needed to take into account some peculiarities of small scale experiments: power profile evolution versus time, surrounding structures with imposed BC, pure non-condensible flow, view factors in particular experimental geometries,…. • Code weaknesses can only be identified in running these codes on small scale experiments which involve a large diversity of conditions. REFERENCES [1] [2]

C.GONNIER, G.GEOFFROY and B.ADROGUER. PHEBUS SFD Programme—Main results. ANS meeting, Portland, 21–25 July 1991. C.GRANDJEAN and C.GONNIER. Preliminary specifications for the ISP 28. CEA report—PHEBUS CSD 112/90, March 1990.

REVIEW OF B9+ BENCHMARK RESULTS

[3] [4] [5] [6] [7]

103

C.GRANDJEAN and C.GONNIER. Technical specifications, Complementary CEA report—PHEBUS CSD 117/90, Sept. 1990. B.ADROGUER. ISP 28. SFD PHEBUS test B9 +. Final specifications for calculations and reported results. CEA report—PHEBUS CSD 115/90, September 1990. S.BOURDON, P.VILLALIBRE, B.ADROGUER and G.GEOFFROY. Analysis of the severe fuel damage test PHEBUS B9+using ICARE code. 27th National Meeting, AIChE Minneapolis (1991). R.GEOFFROY Post test examinations from B9+ test. CEA report—to be published. B.ADROGUER, A.COMMANDE and C.RONGIER. Preliminary Comparison Report of the ISP 28. CEA report—PHEBUS CSD 122/91, May 1991.

Discussion following the presentations of SESSION II Summary of the chairman Mr. A.Meyer-Heine

The three initial presentations clearly pointed out the state of the art in the areas of core degradation, aerosols physics and chemistry. From the discussions it appeared that instrumentation is one of the major points on which the Phebus project should concentrate if optimal advantage is to be made from the programme. At the same time each of the three speakers admitted the great difficulty of the task. Questions to Dr. Wright allowed him to emphasize the two major objectives of core degradation research: - Better knowledge of the threshold failure and its location, transition to debris bed for highly irradiated fuels, burn-up effects and a clear picture of the molten pool; - Improvement of knowledge about low volatile fission product behaviour, influence of the changing fuel morphology and pool crust effects. From Dr. Schöck’s presentation it clearly appeared that predictions of aerosol behaviour were primarily dependent on an accurate knowledge of the thermohydraulic conditions and of the source itself. An instrumentation, both time and space dependent, is thus of great importance for the Phebus programme. Dr. Nichols emphasized the importance of chemistry and stressed that it would be of interest to be able to have more measurements at the bundle exit. He admitted the great difficulty of such an objective. The second part of the session was devoted to Phebus CSD experiments performed in the past. The main questions were related to the feedback of this programme on the Phebus experiments. Mr. Gonnier and Mr. Adroguer made the following observations: - Experimentalists now know how to conduct experiments on fuel degradation, even with a certain quantity of molten fuel. - Instrumentation behaviour has been checked. - The understanding of UO2-Zr interaction (solid and liquid contacts) deduced from Phebus CSD will be useful to define the Phebus-FP tests. A major difficulty remains as far as Phebus-FP is concerned: how can be assured that a given quantity of liquefied fuel is obtained. The problem is being studied, but at present no solution has been found.

SESSION III CORE AND FP-BEHAVIOUR

Safety analysis needs and main phenomena to be studied J.Gauvain , CEA/IPSN Fontenay-aux-Roses and H.M.van Rij , CEC/JRC Ispra PHEBUS-FP Objectives, test matrix and representativity of the Phebus-FP experimental programme A.G.Markovina , CEC/JRC Ispra and A.Arnaud , CEA/IPSN Cadarache Phebus-FP test facility P.Delchambre , CEA/IPSN Cadarache and P.von der Hardt , CEC/JRC Ispra Phebus-FP instrumentation P.von der Hardt , CEC/JRC Ispra and G.Lhiaubet , CEA/IPSN Fontenay-aux-Roses CEA analytical activities: HEVA, PITEAS, mini-containments C.Lecomteand G.Lhiaubet , CEA/IPSN Fontenay-aux-Roses CEA support activities: EC shared cost actions and others P.Fasoli-Stella and A.G.Markovina , CEC/JRC Ispra Phebus-FP organisation of the project and international collaboration A.Tattegrain , CEA/IPSN Cadarache and P.von der Hardt , CEC/JRC Ispra Summary of discussion P.Fasoli-Stella , CEC/JRC Ispra

SAFETY ANALYSIS NEEDS AND MAIN PHENOMENA TO BE STUDIED J.GAUVAIN 1—H.M.van RIJ 2

1Commissariat

à l’Energie Atomique Institut de Protection et 2 Commission of the European Community Institute for de Sureté Nucléaire Centre d’Etude de Fontenay-aux-Roses Safety Technology Joint Research Center of Ispra 21020 92265 Fontenay-aux-Roses—France Ispra—Italy SUMMARY In severe nuclear accidents the reactor is damaged such that fission products (f.p.) are released from the fuel rods. Hence the potential of a source term, i.e., quantity, composition and timing of a f.p. release into the biosphere, exists. Although the probability of a severe accident is very low, the biological impact may be significant. Reactor safety analyses have identified possible severe accident scenarios. Presently the two main objectives of these analyses are the evaluation of the source term and the efficiency of prevention and mitigation due to operating procedures and installed hardware. These analyses are carried out with computer codes, containing mechanistic models and/or empirical relations. Many of these models and relations have been verified by various types of experiments and others are only assessed on a limited basis. In some experiments, phenomena have been identified which are not incorporated into the computer codes. Finally no integral experiment, (i.e. fuel bundle, reactor vessel, primary circuit, containment/auxiliary building) has been performed to date and as a consequence the models/relations are not validated when interaction between various parts of the reactor system occurs. The PHEBUS-FP test series will enable some of the outstanding questions to be answered. In the fuel region an investigation on the f.p. release rate as a function of fuel burn-up, various clad oxidation states, molten phase duration and cooldown rate will be made. Once the f.p. are released, the hydrogen to steam ratio, the structural components, the boric acid and control rod will have an influence on the chemistry of the f.p. Also it is observed that ceramic cladding will remain intact longer than the metallic part which will melt earlier. In the reactor coolant system (RCS) more information is required on physical and chemical processes related to f.p. transport. More specifically, growth and hygroscopicity should be studied in greater detail. In the containment no experimental data are available that combine f.p. decay heat, thermohydraulics and aerosol phenomena. On a longer timescale chemical reactions (e.g. iodine species), radiolysis and sump pH have an important influence on the possible source term. The additive to the sump and the usage of engineered safety features might reveal important information concerning accident management aspects. The PHEBUS-FP project is not intended to provide information on f.p. behaviour into the biosphere and hence this aspect is not considered in this presentation on reactor safety analyses needs. 1 INTRODUCTION In the unlikely case where a severe accident would occur in a nuclear reactor the core would be damaged such that fission products would be released from the fuel rods. These fission products (f.p.) will partially deposit in the reactor coolant system (RCS) and the remainder will be transported to the reactor containment building, where they eventually deposit on surface areas as the time progresses. Gravitationnal settling is the dominant aerosol removal mechanism from the containment atmosphere. In the case where there would be containment leakage or intentional containment depressurisation through a filtered venting system, a small amount of these f.p. would be released into the biosphere. Although the probability of containment leakage is extremely low, its biological and economical impact may be significant.

SAFETY ANALYSIS NEEDS AND MAIN PHENOMENA TO BE STUDIED

107

For this reason reactor safety analyses are performed which identify the main likely scenarios for severe accidents. The two main objectives of the severe accident part of these safety analyses are the evaluation of potential source term and the evaluation of the efficiency of prevention/mitigation due to operating and emergency procedures and installed hardware (i.e. engineered safety features). These analyses are carried out with computer codes containing either mechanistic models or empirical correlations. Many of these models and correlations have been verified by various types of experiments conducted all over the world but others are only assessed on a limited basis. In some experiments, phenomena have been identified but not sufficiently to be incorporated into the computer codes. Finally no integral experiment, (i.e. with fuel bundle, reactor vessel, primary circuit and containment or auxiliary building) has been performed to date and, as a consequence, the models or correlations are not validated when possible interaction between various parts of the reactor system occurs. The PHEBUS-FP test series will enable some of the outstanding questions to be answered. 2 FUEL BEHAVIOUR In reactor accident studies the first aspect is the fuel behaviour during the core degradation. On one side, when the fuel rods reach high temperatures, due to residual power, cladding oxydation power and lack of cooling, f.p. are released, starting from the most volatile species, together with structural material components, especially coming from control rods. The release rates of these f.p. and material components, mainly in vapour form, are depending on the hydrogen to steam ratio, the cladding oxidization state, the molten phase duration and also to the fuel burn-up and the cooldown rate. The PHEBUS-FP instrumentation will not allow a precise study of these release rates, which are measured in specific experiments as HEVA in France or at Oak Ridge in USA; but the different tests will aim to represent a set of different specific conditions representative of the various typical accident scenarios. On the other side, even when the core melt would occur the plant operators together with the accident management team will always try to cooldown the core in order to mitigate or stop the accident and, after a certain period of time, to reset the reactor in a safe state. For this purpose it is needed to know the final state of the core and its coolability which is related to core degradation. The fuel mechanical degradation has been shown to be a complicated non-coherent process, in which the core region undergoes a number of physical and chemical transformations, driven by the temperature history. The experimental data from in-pile experiments cover mainly the range of phenomena up to the onset of formation of liquefied components and are not sufficient to explain the phenomena observed in TMI-2 core examination. A metallic partial blockage is formed across the core by the melting, relocation and freezing of the cladding zircalloy, the control rod materials and eutectic alloys of these materials. This metallic relocation leaves the ceramic free-standing declad UO2 fuel pellets and ZrO2 oxydized cladding behind the high temperature upper core region. Thus, the metallic and ceramic debris, with melting points that differ by 600 K or more, become separated in space and behave quite differently in continued melt progression. A major question is whether a metallic core blockage develops in all melt progression sequences. Major current uncertainties in melt progression concern the failure threshold and failure location of the debris-supporting metallic crust under attack by the growing molten ceramic pool. PHEBUS-FP may try to extend fuel melt progression up to ceramic melt temperatures and prolong the time of permanence at high temperature so as to allow extended material relocation. However the possibility of extending the test to include more extreme conditions is clearly bounded by technological and safety constraints. Then this objective could probably not be reached with the same quality as the f.p. behaviour objective. An other aspect of the in-core behaviour of the degradated fuel is the chemical aspect. The different sets of conditions proposed for the test series will allow to investigate the effect of carrier gas content, boric acid, structure and control rod material on f.p. chemical form and then on the partition between vapor phase and condensed phase of these f.p. The chemical analysis of these f.p. during the tests should provide useful informations concerning the solubility of each species which is directly related to the radiological consequences in atmosphere (if rain) or in groundwater. In summary it should be noted that, due to the specificity of in-pile experiments, precise analytical study of the degraded core behaviour is not possible in such a test. But, on an other hand there is no other way to produce a f.p. source representative of an “actual” severe accident than an in-pile experiment. Consequently, as far as the fuel region is concerned, the main objective should be considered as the production of a representative source term for the RCS and the containment, although information from core degradation should be thoroughly investigated when not impeding the first objective. This representativity will be verified by the specific instrumentation devoted to the bundle.

108

THE PHEBUS FISSION PRODUCT PROJECT

3 FISSION PRODUCT BEHAVIOUR IN RCS During a severe accident the fission products escape from the vessel through a part of the RCS toward the reactor containment or, in some scenarios, directly to the environment through an auxiliary or secondary line (bypass sequences). In the nonbypass sequences the f.p. stay a long time in the containment where they will be depleted from the atmosphere by various deposit mechanisms onto surface areas. It has been shown that pipe retention in RCS is, in those accident scenarios, of secondary importance. But in case of containment bypass—and among the possibilities the most important is the steam generator tube rupture—the pipe retention inside the RCS becomes of primarily importance, because all that is not deposited in the circuit is released into the environment. For the study of pipe retention a lot of semi-scale or large scale analytical experiments have been performed in the past (MARVIKEN V in Sweden, LACE first part in U.S. and TUBA in France). But all these tests have been conducted with aerosols generated by plasma torches simulating f.p. or material aerosols. However it is assessed from analytical tests like HEVA that the size of f.p. aerosols released by actual fuel is considerably smaller (one order of magnitude) than those obtained with plasma torches. On an other hand there is no experiment available involving actual f.p. in pipes and then involving the potential effect of residual heat on the deposit. Finally a lot of codes have been written with the purpose of modelling the formation of aerosol from vapor and others dealing with either vapor deposition either with aerosol physics but they are not yet fully qualified. From the chemical point of view, there is a general consensus that the principal fission product vapour species entering the RCS will be CsI, CsOH and Te in an atmosphere with a high H2/H2O ratio, whereas little attention has been paid to steam dominated flows. It should be emphasised that the following phenomena are particularly important in the RCS: interactions of f.p. with control rod or structural material, effects of boric acid, reactions of tellurium and changes in the fission speciation during revolatilisation. PHEBUS-FP should provide an important addition to the existing knowledge of these phenomena. In particular, studies will be made to examine the high-temperature vapour phase chemistry of f.p. using on-line sampling techniques (e.g. gamma spectroscopy, grab samples). Then the interest of the PHEBUS-FP test series appears clearly for the study of actual f.p. behaviour in piping systems representative of the different parts of a reactor coolant system. The different flow pathes chosen for the different tests tend to represent the most likely pathway of the main accident sequences (hot leg, steam generator, pressurizer, injection line,…). The purpose of each test is then much more to study the major f.p. transport phenomenology than to try to represent such or such individual accident scenario. The set of results coming from these different tests should provide a comprehensive data base for an overall assessment of computer codes devoted to severe accident studies such as the ESCADRE system in France. 4 F.P. BEHAVIOUR IN CONTAINMENT The f.p. behaviour in the containement is mainly driven by aerosol physics for the “short-term” (within the first hours after core melt initiation) and chemical reactions with the principal contribution from iodine species for the “long-term” (several days after the core melt). A lot of analytical experiments have been performed in the past concerning either aerosol physics in containment (DEMONA in Germany, LACE second part in U.S., PITEAS in France) either specific iodine behaviour (ACE/ RTF in Canada, IODE in France): But all these experiments have been conducted with simulated f.p. and, as for the RCS, it will be important to obtain data concerning actual f.p. behaviour in containment (i.e. actual aerosol size distribution, actual chemical species, decay heat,…). It is planned to study the most likely configurations for the containment atmosphere (moisture, incondensible gas content,…) and for the sump water characteristics (pH, temperature, radiation level,…). PHEBUS-FP will be a unique opportunity to study the combined effect of f.p. decay heat, thermohydraulics and aerosol phenomena. On a longer time scale, the most important chemical reactions will be evaluated (mainly iodine transformation by radiolysis from ionic form in solution to gaseous form degasing from water and being more or less trapped on metallic or painted surfaces). As regard to the limited number of tests in front of the numerous possible combinations concerning containment parameters, CEA has proposed to add in each test two small vessels to the main containment in order to take profit of each single test for obtaining results concerning three different sets of containment parameters (mainly for chemistry studies) instead of only one, and that for a small installation modification. Morever it will be possible to study in these “minicontainments” such phenomena as rapid depressurisation or pool scrubbing (case of the pressurizer) which is not possible to envisage in the main PHEBUS containment.

SAFETY ANALYSIS NEEDS AND MAIN PHENOMENA TO BE STUDIED

109

5 CONCLUSION PHEBUS-FP appears to be a unique occasion to perform a series of representative global tests allowing the overall validation of severe accident computer code systems (such as ESCADRE in France, ESTER at JRC, or similar systems developped in USA, UK, Germany and Japan). The main purpose of PHEBUS-FP is not the simulation of a limited number of typical accident scenarios but the representation of the most important part of the phenomenology involved in the major different sequences in order to improve code validation. The different tests proposed as much as the instrumentation which has been decided will give the possibility to build a large and comprehensive database describing the fission products behaviour in the reactor core, the reactor cooling system and the containement for the most likely severe accident sequences liable to occur in a nuclear PWR. This database will allow a better assessment of codes which will be used for the renewed study of severe accidents in nuclear reactors. On an other band this test series is not only the chance for a new and more advanced code assessment but is a wonderful occasion of strong cooperative effort between all the experts around the world due to the large extension of the PHEBUS-FP agreement. In the time where nuclear energy has not the approbation of every people such an effort in favor of reactor safety should be certainly an important stone in the building of a renewed confidence in the nuclear domain.

OBJECTIVES, TEST MATRIX AND REPRESENTATIVITY OF THE PHEBUS-FP EXPERIMENTAL PROGRAMME A.Arnaud CEA-SEMAR, Cadarache A.Markovina JRC-STI, Ispra

ABSTRACT The PHEBUS-FP experimental programme is a logical continuation of the international effort undertaken several years ago to quantify the Source Term for LWRs following the occurrence of a severe accident. Several research projects, usually sponsored by groups of organizations, have been carried out, such as the in-pile programmes PBF-SFD, STEP and LOFT-FP and the large-scale out-of-pile programmes MARVIKEN V, DEMONA and LACE. Other projects of different scale, such as ACE, BETA, HEVA, FALCON, REST, etc., have been completed or are still under execution. Despite such extensive effort, no in-pile experiment has been performed in which all the main components and, therefore, all the major phenomena involved in the transport of Fission Products from the core to the containment walls are represented in an integrated system. The general objective of the PHEBUS-FP programme is to offer such a “full” integration of the most important phenomena. In order to achieve this objective, the experimental facility is designed to scale down the components of the primary circuit and the containment vessel. The adopted scaling factor of 5000 is the ratio of FP inventory in a commercial reactor and in the PHEBUS-FP fuel bundle. In each component of the experimental facility the thermohydraulic boundary conditions are controlled to maintain representative conditions. Where the reduced scale or other technological constraints produce inevitable distorsions, compensating measures are possibly introduced to guarantee that the phenomena under investigation can be correctly reproduced. The test matrix includes six tests, the first of which will be performed utilizing fresh fuel, while the other five tests will benefit of the use of preirradiated fuel with burn-up ranging from 23 to 33 GWd/ton. Several test parameters are changed from one test to the other for each of the three main portions of the experimental facility, namely the fuel bundle, the primary circuit and the containment vessel. For this reason the six tests of the matrix should be considered, in principle, independent from each other. The test parameters chosen for the fuel bundle are those which are expected to affect not only the FP release and speciation, but also the mechanical degradation of the fuel. In the primary circuit the configuration of the components utilized in each test is defined on the basis of a reference severe accident sequence. Therefore, the components represented are steam generator, pressurizer, relief tank, low pressure injection system piping, etc. In the containment vessel the test parameters are defined with respect to the postulated physical and chemical processes taking place in the atmosphere and in the sump water. Specific phenomena affecting the FP transport are also taken into consideration, such as the thermal resuspension from pipe walls and the effect of recirculation spray on FP chemistry in the containment. 1. INTRODUCTION For a long period of time, since the beginning of application of nuclear energy for commercial use, it has been conceived that the safety analysis of the plants should be restricted to the so-called “design basis” accident. The Reactor Safety Study (WASH-1400) in 1974 for the first time pointed out the need to better understand “severe accidents”, namely those which potentially involve extended damage of the reactor core. The probability of such events is very small but the consequences for the population could be extremely serious so that the corresponding risk, that is the probability multiplied by the radiological consequences, is by no means negligible. TMI-2 in 1979 and Chernobyl in 1986 have dramatically put into evidence that such accidents may happen and need to be investigated in order to reduce their consequences to an acceptable level.

OBJECTIVES, TEST MATRIX AND REPRESENTATIVITY OF THE PHEBUS-FP EXPERIMENTAL PROGRAMME

111

Several reviews of data from extensive experimental and analytical researches, which were undertaken in the years following TMI-2, were performed by qualified groups like the Special Committee chartered by the American Nuclear Society, and showed that the amount of radioactivity that could be released outside the reactor building in a severe accident is, in general, far less than had been estimated earlier. It was, however, found that, due to large design differences in the existing nuclear power stations and, therefore, to the complexity of the postulated severe accident sequences, no general conclusions on the Source Term could be drawn. There exist, in fact, a few accident scenarios, in particular those involving early containment failure or containment bypass, in which the Source Term could be higher than that anticipated. Fission product retention in the reactor coolant system (RCS) was not taken into account in WASH-1400. Further studies showed that important fractions of the materials released from the damaged core could be retained not only in the reactor vessel upper plenum but also in other parts of the RCS. Re-vaporization of the deposited materials could also occur due to continued decay heating. Materials not retained in the RCS enter into the containment as aerosols transported by the escaping mixture of hydrogen and steam. It was believed for several years that the behaviour of fission products (FP) in the containment is dominated by the aerosol physical depletion processes, principally agglomeration and settling. It was soon found that the long-term release to the environment (several days from the accident) is strongly influenced by chemical processes, leading to the possible formation of highly volatile compounds of important isotopes. There is, at present, a consensus view in the international community that the quantification of Source Term is plant- and scenario-specific and has to be performed with analytical tools which include the most important phenomena which determine the FP behaviour. Such phenomena may become complex and, in order to be correctly described, may require a good understanding of the circuit thermohydraulics, of the aerosol physical processes, of the chemical transformations and of the effect of decay radioactivity. Extensive research performed up to date has provided a valuable database on separated effects and mechanisms. The PHEBUS-FP project has been proposed to produce a few qualified integral tests, in which not only the various effects and mechanisms are mixed together, but also the main components involved in an accident sequence are represented in the experimental facility. Together with a significant support programme performed in France by CEA and in the countries of the European Community, sponsored by the JRC-Ispra, it is believed that the PHEBUS-FP project may provide an important contribution for the solution or mitigation of existing issues in the prediction of LWR Source Term. 2. OBJECTIVES The PHEBUS-FP experimental programme is a logical continuation of the international effort undertaken several years ago to quantify the Source Term for LWRs following the occurrence of a severe accident. The experimental programmes, performed up to date, have been conducted along several research directions, namely: - aerosol physics in intermediate or full-scale facilities, including reactor vessel, primary circuit and containment, with simulated sources (e.g. MARVIKEN V, DEMONA, LACE); - fission product release at high temperature from high burn-up fuel in out-of-pile tests (e.g. SACHA, HEVA, ORNL); - fuel degradation processes in out-of-pile tests using electrically heated rods (e.g. REBEKA, CORA, ORNL); - in-pile severe fuel damage experiments and FP release and speciation (e.g. PBF-SFD, LOFT-FP, STEP, NRU, PHEBUSCSD). These experiments are either separate effect tests or partially integral tests with respect to the complex phenomenology taking place in a reactor system during a severe accident. No in-pile experiment has been performed to reproduce, in a single integrated system, the main phenomena which influence the Source Term from the beginning of the heat-up phase of the reactor core up to the fission product release into the environment. The general aim of the PHEBUS-FP experiments is to offer such a “full” integration of the phenomena during severe accidents. More specifically, the objectives assigned to PHEBUS-FP are the following: 1) to perform experiments of integral nature to verify: . the validity of analytical models developed on the basis of separate effect results, . that no major phenomena have been omitted from the models;

112

THE PHEBUS FISSION PRODUCT PROJECT

2) to study in-pile the late phases of mechanical fuel degradation and the FP and aerosol transport, deposition, resuspension and chemical transformations in the core region, primary circuit components and containment building under severe accident conditions; 3) to improve the understanding of Source Term phenomena and to make available a technical and scientific database to be utilized for: . . . .

severe accident analysis, Source Term evaluation, accident management, post-accident recovery actions;

4) to provide qualified experimental data to be utilized for model improvement and code validation in Source Term analysis.

The experimental programmes which could be considered as more closely related to PHEBUS-FP are the large-scale out-ofpile programmes MARVIKEN V, DEMONA and LACE for the FP/aerosol behaviour in the primary circuit and containment building and the in-pile programmes PBF-SFD, LOFT-FP, STEP and PHEBUS-CSD from the point of view of fuel degradation and FP behaviour in the core region and primary circuit. The role of PHEBUS-FP with respect to these experiments is at the same time confirmatory and complementary. 3. COMPONENTS AND PHENOMENA TO BE SIMULATED IN THE PHEBUS-FP FACILITY The PHEBUS-FP experimental programme is defined with consideration of two major aspects: - the reference severe accident scenario; - the phenomena to be investigated. The LWR severe accident conditions to be simulated are those of the basic PWR severe accident sequences plus some specific phenomena of interest also for BWRs, Namely the tests are planned with reference to the following guiding scenario: - Large Break LOCA (AB sequence) in the hot leg of the primary circuit. General requirements: . low pressure in the core and RCS . no components in the RCS . containment building - Small Break LOCA (SD sequence) in the cold leg of the primary system with the secondary side feedwater system in/out of operation. General requirements: . low/intermediate pressure in the core and RCS . steam generator in the RCS . containment building - Low Pressure Injection System (LPIS) Interfacing LOCA (V sequence). General requirements: . low/intermediate pressure in the core and RCS . LPIS components in the RCS . auxiliary building - Steam Generator pipe rupture (containment bypass sequence). General requirements:

OBJECTIVES, TEST MATRIX AND REPRESENTATIVITY OF THE PHEBUS-FP EXPERIMENTAL PROGRAMME

113

Figure 1: Source term main phenomena.

. intermediate pressure in the core and RCS . primary and secondary side of the steam generator in the RCS . auxiliary building - Transient initiated accident (TMLB') with or without inclusion of a volume of water (to simulate postulated TMI-2 conditions in the pressurizer). General requirements: . high pressure in the core and RCS . pressurizer (and relief tank) in the RCS . containment It is not intended to simulate any complete accident sequence, but to make reference, for each PHEBUS-FP test, to one or two risk dominating accident sequences in order to identify the typical components and thermohydraulic conditions affecting the FP/aerosol physico-chemical phenomena under investigation. Fig.1 shows schematically the main phenomena determining the radioactive release to the environment. This scheme reflects the phenomenological domains covered by the Source Term computer codes developed up to date. The phenomena falling within the scope of PHEBUS-FP are those within the dashed box. More specifically the PHEBUS-FP experiments should provide information and contribute to a better understanding of the following issues: - FP release from degrading fuel: . effect of fuel burn-up and FP release rate . effect of different cladding oxidation degrees on the release in molten phase during Zr diffusion into UO2 (eutectic formation) . effect of prolonged molten phase on the release of less volatile FPs . effect of changing cooling rates on the late phase FP release – FP speciation upon release from fuel:

114

THE PHEBUS FISSION PRODUCT PROJECT

. . . .

effect of different degrees of clad oxidation effect of different H2/steam ratios effect of control rod material effect of boric acid

- Fuel mechanical behaviour: . study of late phase of fuel degradation, including metallic blockage potential and ceramic molten phase - Reactor cooling system—physical processes: . . . . .

vapour and aerosol interactions (heterogeneous nucleation) aerosol formation, transport and growth during revolatilization processes behaviour of hygroscopic fission product compounds (e.g. CsI and CsOH) aerosol retention in primary system components decontamination by pool scrubbing

- Reactor cooling system—chemical processes: . . . . . .

interactions of fission products with control rod materials the effect of boric acid (chemical and aerosol) the reactions of tellurium (with other fission products and reactor materials) changes in the fission product speciation during revolatilization transformations in water pools reactions of fission products with structural materials

- Containment: . . . . .

iodine reaction kinetics in solution effect of radiolysis on the oxidation of I− in solution mechanisms of formation and trapping of organic iodides (e.g. by paints) effect of pH on solution chemistry effect of Te behaviour on the overall iodine chemistry

Since most of the above-listed phenomena require decay heat and correct chemical compositions of components, PHEBUS-FP may significantly contribute to resolve or to alleviate some of the present issues. 4. TEST MATRIX In Fig.2 a schematic representation of the PHEBUS-FP experimental facility is shown. This can be divided into three parts: - the fuel bundle which simulates the reactor core; - the primary circuit which includes components of the RCS; - the containment vessel which simulates the containment building. In Table 1 the basic characteristics of the PHEBUS-FP facility are illustrated. A simplified representation of the PHEBUS-FP test matrix is given in Table 2. The experimental programme consists of six tests, the first of which (FPT-0) will utilize fresh fuel, while the other tests (FPT-1 to FPT-5) will utilize preirradiated fuel at medium to high burn-up. Table 1: Basic characteristics of the PHEBUS-FP experimental facility FUEL BUNDLE N° of fuel rods Fuel rod lenght

20 1m

OBJECTIVES, TEST MATRIX AND REPRESENTATIVITY OF THE PHEBUS-FP EXPERIMENTAL PROGRAMME

FUEL BUNDLE Geometry Control rod Shroud Ø i=72,5 mm dense ZrO2 Spacer grids Inlet flow 5 g/sec System pressure Linear power 17,5 w/cm

PWR 1 cylindrical

12,6 mm pitch square array central 80% Ag–15% In−5% Cd

inner layer material 2 H2/steam max. flow rate 35 bar 150 w/cm max. test

Zircaloy variable composition max. preconditioning

PRIMARY CIRCUIT Components steam generator pressurizer and relief tank components of LPIS complex structures Material Neutral pipe L= ~12 m Total pipe lenght

pipes of different size

Inconel + SS Ø i=30–40 mm ~22 m

CONTAINMENT VESSEL 10 m3 1,77 m 5 bar 5 m2 recirculation 5 m3/h 1 K/mn max. cooling rate 500 mm diameter

Volume Internal diameter Pressure Condenser surface Spray type Depressurization rate Wall temp, control 0,1 K/mn Sump water pot 500 mm

max. operating

max. at TPN max. heating rate height

Table 2: PHEBUS-FP simplified test matrix Test n°

System pressure (bar)

Fuel bundle

Primary circuit

Containment vessel

SG secondary cold/hot as FP-T-0 or min. pipe line SG second. hot+V seq. components Pressurizer+relief tank dry Pressurizer+relief tank with water SG second. hot+ steam dryer or complex structures

Tight pH neutral Tight Initial pH acid Tight/vented Initial pH acid then alkaline Tight Initial pH alk. Tight Initial pH alk.

Burn-up (GWd/ton) Flow conditions H2 /H2O FP-T-0 FP-T-1 FP-T-2

low low low

Fresh precond. LOW 20–24+precond. LOW 22–26+precond. HIGH

FP-T-3 FP-T-4

high high

24–28+precond. LOW 26–30+precond. LOW

FP-T-5

interm./low

28–33+precond. HIGH

Vented Initial pH acid then alkaline

115

116

THE PHEBUS FISSION PRODUCT PROJECT

Figure 2: Schematic representation of the PHEBUS-FP experimental facility.

The system pressure is changed from low (2÷5 bar) to intermediate (~20 bar) up to high (~35 bar) level. The maximum operation pressure of the containment vessel is 5 bar; therefore, for the intermediate and high pressure tests a pressure reducing valve will be installed in the primary circuit. The inlet flow is controlled in order to have either H2-rich or steam-rich environment at the time FPs are released from the fuel. In Fig.3 the flow paths of the six tests are shown. The components of the primary circuit are changed from FPT-0 to FPT-5 while the overall designs of the test train and of the containment vessel are the same for all the tests and the containment vessel is a fixed structure. Since PHEBUS-FP is an integral experiment of a complex nature, more than one parameter is modified from one test to the other and each test should be considered, in principle, as self-standing. Therefore, it is worth illustrating the test parameters for each of the three parts of the experimental facility, namely the fuel bundle, the primary circuit and the containment vessel. The logical process by which each test will be analyzed is shown schematically in Fig.4. Extensive design and analytical work has to be performed before each test is finalized. This process may involve a revision of some objectives when contradictory with unforeseen constraints. In another paper of this Seminar the preparatory analytical work performed in collaboration with several European organizations in the frame of the Shared Cost Action programme is illustrated. This work was devoted to the analysis of PWR risk dominating severe accident sequences to identify the important FP/aerosol phenomena and to the assessment of the capability of a scaled-down facility to reproduce the identified phenomena. 4.1 Fuel bundle In Table 3 the test matrix of the fuel bundle is shown. The degree of clad oxidation is combined with the H2/steam ratio to obtain different conditions for clad oxidation and for FP speciation upon release from the fuel. It is also understood that the fuel mechanical behaviour during the entire degradation transient is strongly affected by the conditions in which the heat-up phase takes place.

OBJECTIVES, TEST MATRIX AND REPRESENTATIVITY OF THE PHEBUS-FP EXPERIMENTAL PROGRAMME

117

Figure 3: Configurations of the PHEBUS-FP primary circuit.

Low clad oxidation is obtained under oxygen starvation conditions. If combined with high system pressure, providing a negative pressure gradient across the cladding, this situation leads to a high potential for Zr dissolution through the UO2 with formation of a liquid eutectic compound. Vice-versa, high clad oxidation combined with low system pressure tends to maintain the geometry of the bundle, apart from the ballooning and rupture of the cladding which are local effects and require much higher temperature before melting and material relocation take place. The central control rod is at the moment foreseen for all the tests, but consideration will be given in the future analyses to perform one test without this component. Table 3: Fuel bundle test matrix

Boric acid, which is present in the main coolant and in the ECCS water of a LWR to control reactivity in normal and emergency conditions, will be injected in the fuel bundle. It is foreseen to run some tests with and some tests without boric acid. Studies are under way not only to design a system of injection sufficiently representative of the conditions encountered in a reactor, but also to understand better the behaviour of this compound under the conditions specified for each test and, in particular, the effect of system pressure. The fuel utilized for the experiments is from BR-3 reactor. Homogeneous enrichments (5 and 6.85%) for each fuel bundle is ensured. The burnup for the five tests utilizing preirradiated fuel will range from 23 to 33 GWd/ton. The experimental phase will be preceded by a preconditioning phase of 9÷15 days to build up a sufficient inventory of short-lived FPs. A liquefied amount of fuel is anticipated in each test. According to the degree of interaction between Zr and UO2 (eutectic formation) this may require to push the fuel temperature up to the melting point of ceramic uranium (3100 K). 4.2 Primary circuit The various configurations of the primary circuit are shown in Fig.3. Two options are left open for test FPT-1 and FPT-5 pending the results of FPT-0. In fact, if it is found that the FP retention in the primary circuit is high, then FPT-1 will be run with a configuration in which no major component is present. In the opposite case FPT-1 will utilize the same configuration of FPT-0, namely with inclusion of a steam generator. The secondary side may be switched from hot to cold condition to assess the retention potential in two different thermohydraulic conditions. Table 4: Primary ciruit test matrix

118

THE PHEBUS FISSION PRODUCT PROJECT

Figure 4: Work stages to arrive at the final specification of each test.

In the case that FPT-1 will have no major component in the primary circuit, then FPT-5 will be run including two components, one simulating the primary side, the other the secondary side of the steam generator. In the opposite case the primary side of the steam generator will not be included in FPT-5. In one of the two high pressure tests (FPT-3 and 4) a volume of water will be included to simulate the conditions postulated for the TMI-2 pressurizer. The test conditions to be satisfied in this case, which is of interest also for BWRs, are under study. Resuspension of FPs in the reactor results from a mechanical or a thermal process. Mechanical resuspension is produced when a steam spike is generated, i.e. when the damaged core slumps in the water located in the lower cavity. Reproduction of this phenomenon does not require radioactive fission products. In addition, it is not possible in the PHEBUS-FP experimental circuits to reach sufficiently high Reynolds numbers needed for such a simulation. Therefore, mechanical resuspension will not be investigated in the PHEBUS-FP programme. Instead, the study of thermal resuspension is foreseen within the PHEBUS-FP test matrix because in this case it is essential to generate the actual FPs in order to be representative. 4.3 Containment vessel The experimental investigation in the containment vessel will involve: - the physical behaviour of aerosols and the balance between phenomena such as thermophoresis, diffusiophoresis and settling during a first period of the transient; - the chemical behaviour of FPs and in particular the partitioning of iodine between the atmosphere, the sump water and the dry or wet paints during a second period of a few days. Therefore, the following parameters will be taken into consideration: - pressure, humidity, temperature and FP/aerosol concentration in the atmosphere (the first three parameters may be controlled through steam and non-condensible injection or venting);

OBJECTIVES, TEST MATRIX AND REPRESENTATIVITY OF THE PHEBUS-FP EXPERIMENTAL PROGRAMME

119

- temperature, pH, dose rate and FP/aerosol concentration in the sump water (also in this case the first two parameters may be directly controlled). Two other experimental phases may be foreseen in the late period of the test transient: - slow depressurization of the containment, typical of accident management measures; - recirculation spray to study the possible change in the chemical equilibrium. Table 5: Containment vessel test matrix Test n°

Initial sump conditions

FP-T-0 FP-T-1 FP-T-2 FP-T-3 FP-T-4 FP-T-5

Neutral Alkal. Acid Alkaline Alkaline Acid

Depressurization Acid

Slow No Slow Slow No No

Spray Slow

No No Yes Yes No No

5. REPRESENTATIVITY ASPECTS At present the design and analytical activities are concentrated on FPT-0 which has been utilized, in general, as the leading test to design the experimental facility. Therefore, a presentation of some representativity aspects may be restricted to this test. In general the approach followed in the definition of the geometry and boundary conditions of the experimental circuit is based on the following criteria: - The system component dimensions are scaled down so as to guarantee similar flow rates and concentrations; - the thermohydraulic boundary conditions are controlled to provide values close to those calculated for nuclear power plants; - when there exist unavoidale geometrical and technological constraints, design counter-measures are introduced, to the extent possible, to limit the corresponding experimental distorsions. The PHEBUS-FP fuel bundle is intended to provide a prototypical “FP/aerosol source” and a correct PWR geometry and material composition. Since the fuel cluster of PHEBUS-FP reproduces a typical PWR configuration (a square array with a 12. 6 mm pitch) and the active length is 1 m, the scaling-down criterion is given by the FP inventory ratio and results to be 5,000. By this factor the bundle inlet mass flowrate is scaled down. The primary circuit configuration is affected by some geometrical constraints, imposed by the distance between the test train and the “caisson”, housing the main components, and by the shielding and handling requirements. Since the approach chosen is to study phenomena taking place in the primary system components and not to reproduce a severe accident sequence, a section of the primary circuit, namely the part between the test train outlet and the first instrumented location (point C) is designed to be as “neutral” as possible with respect to the FP/aerosol transport phenomena. Therefore, the wall temperature will be maintained high (700°C) and the diameter (40 mm) is chosen so as to minimize deposition (best compromise between turbulent deposition and settling). The design of the steam generator is relatively simple since the application of a scaling-down factor of 5000 leads to a configuration in which one of the ~ 4,000 U-tubes of the SG is reproduced. Scoping calculations have demonstrated that the height of the SG can be reduced to 5 m without impairing the phenomena under consideration, as they take place mostly in the first few metres of this component. The volumes of the containment vessel (10 m3) and of the sump water pot are scaled down in order to have the correct FP/ aerosol concentrations and radiation intensity. A serious problem is generated by the distorsion in the surface/volume ratio. This problem, associated to decontamination requirements, leads to a particular solution for the internal geometry of the containment vessel. The lateral walls of the vessel are designed to be as “neutral” as possible with respect to FP/aerosol phenomena. The wall temperature is controlled throughout the entire test transient so as to be in equilibrium with the internal atmosphere. The heat and mass transfer are realized and controlled by an internal structure, called “condenser”, which provides the correct painted and unpainted surfaces and can be replaced between one test and another.

120

THE PHEBUS FISSION PRODUCT PROJECT

6. CONCLUSIONS The PHEBUS-FP experimental programme is established to provide an important step forward in the understanding of Source Term phenomena. This project offers a “fully” integrated test facility which includes the simulation of the reactor core, of the primary system components and of the containment building in the thermohydraulic and geometrical conditions expected during a LWR severe accident. Each of the six planned tests has to be considered self-standing, since several parameters are changed from one test to the other. In this way a broad range of phenomena will be investigated. Extensive design and analytical work is planned for each test, to take account of representativity aspects and to minimize experimental distorsions. The preparation of the first test FPT-0, utilizing fresh fuel, is progressing and has reached the stage in which most of the design options have been frozen. In the meantime the preparation of FPT-1 is being undertaken. REFERENCES (1) A.ARNAUD, A.MAILLUAT, A.V.JONES, A.MARKOVINA (1990). Overview of the PHEBUS-FP experimental programme, XVI Annual Meeting of the Spanish Nuclear Society, Oviedo, 14–16 October 1990. (2) PHEBUS-FP: Gran Rapport Descriptif, to be published. (3) Internal report, Test matrix of the PHEBUS-FP programme, Revision 2, March 1991.

PHEBUS FP TEST FACILITY Ph. DELCHAMBRE CEA IPSN/DRS P.von der HARDT JRC STI/IPTD

SUMMARY After a rapid survey of the main modifications realized on the existing Phebus facility, Phebus-FP experimental circuits are described, with mention of constraints to take into account and reasons of main choices. Starting from the fission products source (fuel bundle), we follow the fission product path via simulated apparatus (steam generator for the first test) and containment up to the atmosphere tank. Particular attention is drawn to the containment tank, unremovable part and a major point for each test. Nature and sizes of the different components are given and allow to have a good overview of the experimental circuit, and, even to make some precalculations. Lastly the presentation of the planning of a complete test cycle describes the test sequence and shows why the test rate will be approximately once a year. 1. INTRODUCTION The experimental aim of Phebus-FP is to study, during a PWR severe, out of dimensioning, accident the fission products outgoing from the fuel, their behaviour in the reactor circuits up to the break, in the containment vessel and, in case of containment failure, their release into atmosphere [1]. The Phebus facility which had allowed to carry out Loss Of Coolant Accident and Severe Fuel Damage tests, must undergo significant modifications to realize this new program [2], [3]. 1—The use of largely irradiated fuel (nearly 30 000 MWD/T) involves important transfer delay and, consequently, loss of the short life fission products. To reintroduce them one has to reirradiate the fuel at nominal power during two weeks, shortly before the test. This continuous operation excludes the use of the existing 700 cubic metres water tank to cool the driver core. Consequently this tank has been replaced by a 40 MW heat exchanger coupled with an atmospheric cooling tower. In addition the progressive driver core burn-up has to be balanced by the addition of a graphite core reflector with adjustable thickness. 2—A new building, nearly as large as the existing reactor hall is coupled with it. It contains the new core cooling circuits and principally the experimental Fission Product circuits. Auxiliary buildings are also erected to contain ventilation, control and electrical supplies. Moreover, as requested by the safety authorities, important additional work has been realized to reinforce the reactor building against earthquake. 3—The in-pile cell which contains the test train is also modified, for two reasons. First, it must have now two successive ways of operation: water cooling of the experimental fuel during reirradiation phase, shifting to fission products release conditions through complete cut-off of water flow. Second, it must allow to reach high temperatures all along the fission product path in the outlet towards the experimental circuit. After this rapid survey of the main modifications of the Phebus facility, the paper will describe more in detail the experimental circuit, only indicating auxiliary circuits (waste, cooling,…) and give some information about the different phases of a complete test cycle. 2. EXPERIMENTAL CIRCUIT The program requires the implementation of an experimental circuit to simulate the entire leakage path, from the core to the environment. There are several severe accident scenarios to be studied, following position and nature of the break. That leads

122

THE PHEBUS FISSION PRODUCT PROJECT

Figure 1—Experimental circuit

to different experimental circuits. In fact, the main difference between one and the other concerns the component to pass through for the fission products before reaching the break: steam generator, pressurizer with its relief tank or…nothing. 2.1 The circuit Therefore the experimental circuit consists of: (see fig 1) —a fuel bundle inside a removable test train which represents the PWR core. Coming from the partly molten fuel bundle inside the test channel, the FP pass through the vertical volume above the bundle, into the horizontal line that forms the link from the reactor to the experimental circuit proper. This horizontal line is needed to cross the reactor pool and the concrete wall between reactor and FP halls: it is foreseen to get that line as neutral as possible for FP behaviour. The experimental circuit is situated entirely inside a containment called “caisson” or REPF 501. At the point “Y” the cooling circuit of the test section separates from the flow path of the FP. This cooling circuit, required for the re-irradiation phase, is out of service during FP release. Through the valve VAPF 501, closed during re-irradiation, and open during the test phase to let the FP enter the experimental circuit, the FP reach the “C” point, a strongly instrumented area. Between the “C” and the “G” point, a second strongly instrumented area, a simulated reactor primary circuit component is placed which will be different at each PHEBUS FP test. For the first test FPTO, this component is a simple steam generator (SG). In the other tests, it will be replaced by another representative component. After the “G” point the FP enter the vessel REPF 502, representing the reactor containment. The line for gaseous waste links the REPF 502 and the atmospheric vessel REPF 503, which is outside the caisson. 2.2 The constraints The experimental circuit has: - to be representative, and calculations have given the best compromises to respect main phenomenologies, - to stay calculable by codes, - to confine safely the radioactive fluids. In this regard, the FP released by the bundle represent 35 000 curies, the first hour after release and 3 000 curies three months after. That explains why a caisson surrounds the experimental circuit up to final filters after the containment tank, and why this caisson is inside a concrete shielded compartment with a thickness of 80 centimetres, - to be equipped with adequate instrumentation, to satisfy the analytical and interpretation requirements of the tests [4],

PHEBUS FP TEST FACILITY

123

- to enable handling, assembly and dismantling, for each test, inside the caisson REPF 501. Dismantling has to be possible by remote handling, prior to, or after decontamination. Other constraints, linked to the installation geometry have to be taken in account: - the caisson with the dimensions Ø 7.25×8.5 m contains the components of the experimental circuit, as well as the REPF 502 vessel, the gaseous waste line with filter and iodine trap, elements not directly linked to the test, like ventilation ducts and filters, organic liquid loops for heating and cooling of the circuit, etc; - the fact that certain measuring devices have fixed places which cannot be modified between two tests; - fixed places for structural elements of the caisson: The stair, the air lock, the different transfer systems both at the periphery and at the ground and the ceiling, the remote handling system, etc. The descriptions which follow will give information about the different components. They will be more detailed for the containment, in so far as a very important and unremovable part of the experimental circuit. 2.3 Main components 2.3.1 The test train Device functions

The test train allows: - cooling of fuel during the 9 to 14 days irradiation phase while providing the continuity of LOCA loop (tightness, pressure loads take-up), - shifting from LOCA conditions to FP conditions through complete cut-off of water flowrate, - achieving the experimental phase which should lead to partial melting of fuel (20 %) and release of fission products to FP circuit and cell, - injection of steam, hydrogen or helium, under the bundle during the FP release phase. General description The test section essentially consists of a long tube (zircaloy and stainless steel above fuel, diam. 112/96, 4.5 m length) whose lower section is closed by the foot valve assembly and whose upper section is closed by the connector holder and gripping head. From bottom to top, we find: foot-valve, fuel bundle, transition zone, rising line and outlet to horizontal line. - The foot-valve allows shifting from irradiation phase to FP release phase. It insulates the internal part of the test train and bundle from the water cooling system (LOCA loop). The tightness of this valve is crucial to avoid melting UO2/water reactions. - The fuel bundle (cut view of fig. 2) Twenty PWR type fuel rods (only one metre height) and a Ag-In-Cd central rod are held in a bundle (12.6 pitch) made up of two AFA 5×5 spacer grids, re-cut-out (zircaloy), positioned on either side of the mid-core and secured to the four external tierods and the guide tube around the Ag-In-Cd rod. The fuel is fresh for the first test (UO2 enriched to 4.5 %) and is preirradiated at nearly 30 000 MWD/T for the followers (BR3 fuel). The pressure tube which surrounds the bundle is protected by a zirconia layer against high temperature of the fuel. Between foot-valve and bundle are piped-in the various steam, hydrogen and helium feed-lines. - The transition zone is a double conical part (inconel), above the bundle in which the temperature gradient will be important (2 000 to 700°C). The channel diameter is reduced to 48 mm, then to 30 mm up to the outputz elbow. A steam injection inlet is provided here in case of total plugging of the bundle. - Rising line: the channel tube (inconel 600) and associated heat insulator extend up to the heat insulated outlet elbow. It is heated at a temperature slightly in excess of 700°C. The temperature of 700°C chosen for the upper part of the test train and the horizontal line is the best compromise between technical possibilities and a sufficient level to avoid large aerosol formation.

124

THE PHEBUS FISSION PRODUCT PROJECT

Figure 2—PHEBUS FP In-pile section HORIZONTAL CROSS SECTION ON REACTOR CORE MID-PLANE (-7500mm)

The inconel has been preferred to stainless-steel because of possible spallation problems with steam at 700°C. 2.3.2 The horizontal line It is a double coaxial pipe between the outlet of the test train and the separating valve VAPF 501. The external pipe is fixed and insures the tightness, the internal one, removable, is fitted with heat tracing cables. The internal diameter is 30 mm. After the test it is planned to withdraw the straight part of the internal tube, between the outlet bend of the test train and the “Y” point. The tube will be examined while being withdrawn, by gamma spectrometry, after which it will be replaced by a new tube. For the three following components Fig. 3 shows main sizes and configuration. 2.3.3 THE “C” POINT is a highly instrumented extension of the horizontal line. Since it is the first point on the fission products path to be instrumented, the pipe and the instrumentation remain heated at the temperature of 700°C to avoid any thermal gradient problem. 2.3.4 THE STEAM GENERATOR is, in the first test, the representative component of the reactor primary circuit. Technically speaking the steam generator is a heat exchanger, with a single, representative U-tube on the primary side (int. diam. 20 mm, inconel), which forms the continuation of the FP circuit. Its secondary side is connected to an organic liquid loop, to control the primary temperature. The steam generator can be equipped with two condensate tanks of 5 1 each, at its bottom end, the contents of which will undergo post-test analysis. The steam generator design foresees the sectioning and recovery, by remote handling, of the primary tube, for post experiment analysis. As this component can collect by condensation a large amount of deposits, it is surrounded by an iron biological shielding (thickness=10 cm). 2.3.5 THE “G” POINT corresponds to the possibility to measure fission products after the steam generator and at the inlet of the containment. Analogous to the “C” point, it is a strongly instrumented part of the circuit. Between the “G” point and the REPF 502 a separation valve VAPF 701 is foreseen, of the full section type. The main pipe, at low temperature (<150°C), is in stainless steel, its internal diameter is 30 mm.

PHEBUS FP TEST FACILITY

125

2.3.6 THE CONTAINMENT TANK REPF 502 (see chapter 3). 2.3.7 THE ATMOSPHERE TANK is a 100 cubic metres tank, partially inertized (3 % of oxygen) by nitrogen before the experimental test to avoid hydrogen hazard, its absolute pressure is 0.7 bar at the beginning of each test. During the test, according to the scenario, it receives fission products and gas coming from containment through the filters of the exhaust line (max. abs. pressure:<1.5 bar). After the test the REPF 502 is emptied completely in the atmosphere tank and, after three months for iodine decay, all gases are released to stack. 2.3.8 THE CAISSON (REPF 501) Fig. 4 As said previously, the caisson contains the main parts of the experimental circuit, and, consequently, the involved instrumentation. To recover samples, may-packs, impactors, etc…, a remote power manipulator and a revolving crane on the same structure are provided. Two shielded windows coupled with two light remote manipulators each, allow to operate by direct sight. Otherwise it is possible to use a closed circuit TV camera. Under the caisson, directly connected, is a small shielded cell to operate first dismantling and gamma counting on samples. PHEBUS F.P. CAISSON 501

3. CONTAINMENTS The containments REPF 502 is a non-consumable component of the experimental circuit Phebus PF; as mounted in vessel REPF 501, it simulates the containment of a PWR in severe accident conditions. It has been designed from representativity studies and considerations. 3.1 Dimensioning criteria 3.1.1 Containment geometry This containment is dimensioned from representativity studies. Considering the inventory of FP available in the test bundle (source term), a 5 000 ratio has been defined relative to the reactor case in order to ensure that concentrations in containment are complied with. This led to a useful volume of 10 m3 for REPF 502. The walls (16 mm thick steel) can be heated externally, the containment is thermally insulated and is surrounded by a biological protection (10 cm steel). After specifying the volume, other problems had to be solved in order to obtain a behaviour representative of the phenomena in the containment: 1) On the one hand, its geometry, inevitably cylindric, yields too large a surface/volume ratio incompatible with the expected thermohydraulic characteristics. 2) On the other hand, painted surfaces are required inside the containment. And, direct painting of walls sets very complex problems of decontamination, destruction of these paints and maintaining an original surface condition from one test to the other. Thermohydraulic calculations of the containment showed that getting a correct relative humidity and a realistic condensation rate requires: - either to maintain 75 % of internal walls at the same temperature as the gas during the test; - or to maintain all walls at the same temperature as the gas and introduce an internal component with a well-defined surface whose temperature can be controlled.

126

THE PHEBUS FISSION PRODUCT PROJECT

Figure 3

The second solution offered the advantage of painting the surfaces of this component in so far it was designed as removable. As regards the aerosol physics, calculations showed that, though the 10 m3 volume is correct as far as the aerosol concentration is concerned, the weight of aerosols laid down with time by gravitational sedimentation was too important in the Phebus containment compared to the reactor case. For this given volume, it should then be necessary to envisage a maximum containment height. Knowing that REPF 502 is mounted in a vessel with defined dimensions and that, thought it is non-consumable, it should be able to be replaced in case of incident, its overall height is limited by the installation handling means. Therefore the dimensions of this containment are: - volume=10 m3 - internal height (without sump)=4400 mm, (overall=5520 mm) - inside diameter=1770 mm. This tank is provided, on top, by a 800 mm diameter opening in order to facilitate any possible post-experimental intervention, and to allow installation of equipments.

PHEBUS FP TEST FACILITY

127

Figure 4

3.1.2 Sump geometry Obtaining a representative chemical behaviour (iodine+aerosols) in the containment requires that several phenomena regarding the sump are complied with: 1) radiation level and associated radiolysis; 2) FP concentration in water; 3) presence of organic matters (paints) and pH monitoring. These requirements affect the sump dimensioning. The former requires a sufficient dose rate in water, hence a sufficient height of water. For a floor surface area of 2.5 m2, this required quantity of water is incompatible with the latter requirement which is a realistic FP concentration in water. The solution is then to limit the sump to a part of the containment floor, with the following geometry for the cylindrical can: height=500 mm, diameter=600 mm. To fully comply with the latter requirement, a washing system will be installed on the remaining floor surface; thanks to the generation (intermittent or continuous) of a water film by circulation of sump water, this system allows to drain the products sedimented on the floor towards the sump and to provide a sufficient surface of exchange with the atmosphere.

128

THE PHEBUS FISSION PRODUCT PROJECT

3.1.3 Containment wall thickness It was initially set at 20 mm (conservative value relative to the hydrogen risk). To improve the wall heating and cooling performance, their thermal inertia had to be reduced and therefore the steel thicknesses had to be optimized. Calculations on the hydrogen risk (blast and detonation) have been made. In fact, the maximum hydrogen quantity admissible in FP circuits (3 m3) requires that the containment withstands a blast and/or a detonation. The results from calculations lead to the following thicknesses: - cylindrical shell: 15 mm - both elliptical bottom sections: 20 mm - sump: 8 mm. Internal surface condition: mirror polish. This option has been requested in order to obtain a surface condition with a minimum roughness; this restrains the incrustation of sediments and provides a softer decontamination of the containment internal walls. EQUIPMENT OF REPF 502 Definition of main pick-offs

a) primary circuit input (level -3210=hot branch). This inlet, 30 mm diameter, enters the containment by 200 mm approximately and is equipped with an 80 mm diameter spraying cone positioned upward in order to restrain the impact of aerosols against the opposite wall. b) output to filters and atmosphere tank (depressurization line). This 25 mm diameter line located in the containment median plane allows to depressurize the containment at a maximum rate of 5 m3/h TPN. This dimensioning rate value corresponds to procedure U5. As regards the discharge circuit, taking the hydrogen risk into account has modified its process and restrained its dimensioning. In fact, the iodine filtering system installed on this line is efficient only for a fluid with a temperature below 70° C and a rate of humidity below 40 %, which requires that a high-performance condenser-dryer system be provided upstream of this filtering system. Moreover, the filters and condenser cannot withstand a hydrogen detonation, all the more as the elimination of water vapour increases this risk. This is why a bypass dilution circuit has been designed to draw gas (essentially nitrogen) from REPF 503 (storage containment) and to reject it upstream of the condenser. Partition between the dilution rate and the discharge rate of REPF 502 is such that the detonation risk in the circuit is cancelled. From this process, the maximum discharge rate has been dimensioned by taking the maximum possible overall size of the filtering system and the dilution rate required for this rate. c) line designed for injection of additional gas into the containment. It is possible to inject gas into the containment in order to adjust the rate of humidity and the pressure. The device foreseen is featured by: injection line inlet level: -3210 (same as for hot branch) type of gas: steam, CO2 or neutral gas, this inlet enters the containment as close as possible to its centreline and is provided with a spraying nozzle directed upwards. d) Instrumentation pick-offs Some twenty small diameter (15 to 20 mm) openings are provided along the walls for the passage of instrumentation (Maypack, impact separators, filtered capsules, etc…) and circulation loops (spraying system, fountain, condensates, etc…). Spraying system

A spray system is achieved by recirculation water from the sump; this water can be injected at a rate of 1 m3/h at the top of the containment via spraying nozzles dimensioned to provide a maximal speed and mean diameter of droplets of about 1 mm. The direction and the number of nozzles should permit the containment volume to be sprayed while preventing the internal walls and structures from being affected.

PHEBUS FP TEST FACILITY

129

Internal condensing structure Design

Obtaining representative thermohydraulic conditions in the containment fuarantees that the physico-chemical phenomena are progressing suitably both or aerosols and the other FP’s during the accidental sequence. It is therefore important to control the condensation rate in the containment. Furthermore, the partition of iodine between the gas and the liquid through radiolysis and reaction with paints is a major phenomenon which manages the type of wastes discharged outboard; painted walls inside the containment is therefore required. Considering the decontamination problems and since the containment is a non-expendable item of the circuit precluding any painting of the REPF 502 walls, it is foreseen to install (inside the containment) an internal, necessarily removable, painted structure whose surface and thermal response are comparable to the internal structures of a reactor containment. It is a steel structure, with a defined surface area, equipped with a temperature control system for simulation of the thermal response of the reactor structures. Geometry

This structure is hung from the top opening cover. Its geometry incorporates three cylinders, 160 mm diameter, each designed as follows: - condensing upper section made up of two concentric steel cylindric shells inside which an organic cooling fluid flows. The external shell is 1 mm thick, the internal shell is 3 to 4 mm thick. This externally painted condensing section is approximately 1500 mm high. - lower section, so-called dry, made up of a steel cylindric shell, 1 mm thick, accommodating a cooled receptacle of approximately 1 1 for on-line collecting of condensates from the upper section. A small valve, at the base of the receptacle, allows to drain the condensates sequentially, either towards the sump or towards the sampling capsules outside the containment. An insulating material is provided between the cylindric shell and the cooled receptacle in order to ensure the thermal equilibrium of the external shell with the atmosphere. This structure therefore provides a condensing surface area of approximately 3 m2 and a dry surface area of 2 m2, i.e. a total painted surface area of 5 m2. 4. TEST CYCLE A Phebus test is made up of three phases: - a construction phase during which the assembly of different new experimental circuit components is realized, - an experimental phase at the beginning of which the installation is operational and which consists of several stages: preparation, irradiation, transient, FP release and containment experimentation, - a post-experimental phase organized with respect to the sample recovery priorities, which leads to the initial installation status and therefore includes the decontamination and dismantling of the FP loop. 4.1 Construction phase As it is an unexperimental phase, it has to be as short as possible, all the new components are available on site before its beginning, and a maximum of prefabrication possibilities will be used. Nevertheless its duration, due to the caisson exiguity, to the large number of connections, mechanical and electrical for instrumentation and to thermal insulation placing, is estimated to three months. 4.2 Experimental and post-experimental phases They are described in the following table with their estimated durations. A particular point needs explanations: Transient 1 and 2. As the needed driver core power to heat and melt the experimental fuel is low, it is necessary to wait nearly 36 hours for resorption of xenon poisoning. This time is used to modify

130

THE PHEBUS FISSION PRODUCT PROJECT

the configuration of the installation circuits and to set the experimental circuits in the initial conditions required by the scenario. The complete test cycle is one year. 5. CONCLUSIONS The objectives of the PHEBUS FP programme, i.e. the study of late phase fuel damage, release, transport and depletion of fission products, required development, design and manufacture of a large new test facility and substantial modifications of the existing PHEBUS reactor plant. This work is now well advanced and most of the new components are available or under final assembly. Subject to confirmation by detailed on-site assembly and pre start-up commissioning planning, the first test FPTO is foreseen for autumn 1992. 6. REFERENCES 1 2 3 4

() A.G.MARKOVINA and A.ARNAUD, “PHEBUS FP. Objectives, representativity, test matrix”, these proceedings. () Ph. DELCHAMBRE and P.von der HARDT, “The PHEBUS FP Project”, EUR 12195 EN (1989). () E.F.SCOTT de MARTINVILLE, Ph. DELCHAMBRE and P.von der HARDT, “The PHEBUS FP Project Status Report 1989– 90”, EUR 12926 EN (1990). () P.von der HARDT and G.LHIAUBET, “PHEBUS FP. Instrumentation”, these proceedings.

Table 1 EXPERIMENTAL PHASE INITIAL STATE

PREPARATION duration=2 months

IRRADIATION duration=9 to 15 days

TRANSIENT 1 duration=24 hours

*

power-on operation

*

* *

thermal balances fuel restructuring

*

bundle instrumentation check-up

*

test scenario defined

*

* *

reactor loaded experimental circuits installed, tested experimental circuits installed, tested experimental circuits installed, tested experimental circuits installed, tested

* *

* * *

*

thermohydraulic test (scenario) checking the γ detectors sampling rates preadjustments loading of FPT0 bundle

*

instrumentation check

*

final writing of test sheets

*

circuit check-up

checking reactor safety system+thresholds * LOCA preparation * FP vessel windows closure * *

isolation of LOCA and FP vessels general instrumentation check-up

TRANSIENT 2 duration=12 hours

FP RELEASE (power on) duration=3 hours

END OF FP EMISSION (cooling) duration=1 hour

AGING duration=4 days

*

*

bundle monitoring

*

reactor shut-down

*

containment experimentation

*

foot valve closure and tightness check horizontal line drainage

*

FP circuits monitoring

*

bundle cooling

*

gas wastes circuit management

*

LOCA circuit prerequisites

*

containment monitoring

*

*

experimental circuits preheating at 240 °C

*

reactor+LOCA monitoring

*

FP/REPF 502 circuits isolation LOCA decrease from 240°C to 150°C FP circuits temperature decrease from 700 to 150 °C

*

PHEBUS FP TEST FACILITY

POST-EXPERIMENTAL PHASE FIRST POSTEXPERIMENTAL ACTIONS duration=2 weeks *

gas drainage

*

FP circuits cooling down

* *

liquid wastes drainage preparation of remote operation May-Pack recovery test bundle removal post-experimental gamma measurements

* * *

SAMPLING RECOVERY duration=2 months

DECONTAMINATION duration=2.5 months

DISMANTLING duration=1.5 months

*

*

*

FP circuits dismantling

*

disposal of solid wastes, HEPA filter+iodine trap

* *

gas wastes disposal liquid wastes drainage

recovery of REPF 502 samplings * removal of internals from REPF 502

*

SG dismantling

horizontal line internal extraction and gamma measurements * test bundle analyses * FP loop decontamination * recovery of samplings on FP * REPF 503 recirculation circuits

131

PHEBUS FP Instrumentation P.von der HARDT, JRC, STI/IPTD, CADARACHE G.LHIAUBET, CEA, IPSN, DPEI, FONTENAY-AUX-ROSES

ABSTRACT Design and development of the Phebus FP experimental instrumentation were carried out in two phases: during the preliminary concept phase of the test facility, 1988–1989, a first scheme was set up on the basis of experience available at Cadarache and from earlier in-pile and out-of-pile tests elsewhere. It enabled R&D projects to be defined in time for those instruments which were not, or not fully, qualified for the projected operating conditions in PHEBUS. During the second phase, 1990–91, the requirements on instrumentation for control, analysis, and interpretation of the tests have been defined. An instrumentation plan was then conceived compromising between above mentioned requirements and various constraints (feasibility, detectability, space, funding). This plan contains a generic part intended as a basis for all tests up to 1997 and a specific part for the first irradiation, FPTO, in autumn 1992. In the PHEBUS FP programme and in this paper, “instrumentation” is used as a collective term designating all apparatus and operations specified for the acquisition of experimental data. The overall instrumentation plan in its present form includes the following items: - thermal-hydraulic measurements in test train (in-pile section), FP circuit and containment vessel, using C and K type thermocouples, ultrasonic thermometers, pressure transducers, thermocouple signal correlation and other flow meters, hygrometers, etc. - on-line FP monitoring by 10 to 15 spectrometers, ionisation chambers and mass spectrometers - off-line (post-test) analysis of aerosol morphology, elemental/ isotopic FP composition and FP chemical speciation using a series of samplers: impactors, filters, capsules, iodine adsorbers…. - post-irradiation examens of the (severely damaged) fuel bundle. The paper reviews this plan and a number of development items involved. It does not cover the test fuel characterisation plan which is usually included into the instrumentation issues. 1. INTRODUCTION This paper treats instruments and methods used to extract experimental data from PHEBUS FP tests. The objectives assigned to the test instrumentation are closely linked to the Programme targets. They can be summarized as follows: a. Compile those data which are necessary to pre-calculate each test, b. Supervise and conduct the experiment during its various phases in order to maintain its controllable parameters inside the limits defined by pre-calculations and by safety criteria, and to determine time and duration of sampling, c. Acquire, process, and compile those data which are necessary to interpret each test and to reconstitute the detailed course of events, d. Acquire, process, and compile any additional information required for code validation and model improvement. Experimental data acquisition of each test uses a number of methods, viz:

PHEBUS FP

133

- a qualified compilation of as-fabricated dimensions and material data of the experimental equipment, supported if necessary by specific laboratory tests, characterization of the test fuel, - on line instrumentation for thermal-hydraulic and FP data, - sequential sampling of FP-containing gas, liquid and solids, at various points of the circuits through global “grab” samples or selective devices (filters, aerosol impactors, iodine speciation adsorbers), with post-test instrumental analysis. - post-test examination of circuit components and of gaseous and liquid wastes, - post-irradiation examens (PIE) of the in-pile section with its fuel bundle. Pre and post-test instrumentation calculations are required for most of the measurements in order to translate the raw results into data useful for test interpretation (fig. 1). 2. OVERALL INSTRUMENTATION PLAN /1 This plan has been developed in 1988–1991 through an iterative process using the following scheme -

review of earlier-in-pile and out-of-pile experiments with FP release /2/, general recommendations on instruments and methods available /3/, /4/, /5/, compilation of the analytical requirements, elaboration of an overall instrumentation plan by the Project team, review by an ad-hoc Instrumentation Group, review by external experts, and by the PHEBUS FP Analytical Group.

Simultaneously detail design, dimensioning calculations, and qualification tests clarified feasibility and accuracy problems. A number of questions, particularly on possible data accuracy, were still open at the time of writing of this report. The following chapters describe the instrumentation, mainly for the first test, in some detail. 3. DETAILS 3.1. In-pile devices 3.1.1. Thermal-hydraulic instrumentation 3.1.1.1. Requirements Re-irradiation phase: The instrumentation of the in-pile device (test train) will supplement the information on flow rate, temperatures, and fuel bundle power output, supplied by the high pressure water loop (LOCA) controls, reactor instrumentation, and neutronic precalculations. Experimental phase: The purpose is to determine the evolution of the following parameters versus time: Fuel bundle and inlet (lower end): total power rating of the test bundle

determined by pre-calculation and verified by post-test interpretation

- nuclear power rating of the test bundle distribution of heat sources in constant geometry and after fuel relocation, fluid inlet (base of test bundle) temperature, pressure, flowrate, composition (H2O, H2),

134

THE PHEBUS FISSION PRODUCT PROJECT

Fig. 1. Role of the analytical-experimental instrumentation

- bundle boundary conditions (part under flux): tube temperature, shroud (heat insulation) temperature field, - test bundle temperature field (including control rod). Fuel bundle outlet: - pressure, temperature, flowrate, - composition (H2O, H2). 3.1.1.2. First test, FPT 0 The instrumentation is summarized in table 1. It features in particular centre line thermocouples in 12 of the 20 unirradiated fuel rods of this test /6/. They exit the fuel bundle at the bottom end leaving the upper zone free from obstructions. Other high temperature thermocouples measure fluid temperatures at various points in and above the bundle. Two ultra-sonic thermometers are provided. Their intrinsic advantages, with respect to noble metal thermocouples, are /7/: . the potential to measure at several axial positions, . the potential to compensate neutron-induced decalibration, . better high temperature reliability.

PHEBUS FP

135

3.1.2. Fission product instrumentation 3.1.2.1. Requirements, experimental phase It is required to determine the evolution vs. time of the effluent at bundle outlet, viz. fission gases (Xe, Kr), elements released ex-core (physical state of vapours or aerosols, chemical forms, characteristics of any aerosols), and transferred radioactivity. 3.1.2.2. Proposed instruments (see table 3) Vertical deposition coupons are mounted above the fuel bundle, providing sampling at two different time intervals. The design and operating principles had been used by the LOFT FP tests /8/. A gamma spectrometer facility has been designed looking at the vertical outlet above the coupons, i.e. 1 m above core midplane. 3.1.2.3. Comments Time resolution and accuracy of the post-test analysis of the coupon deposits are expected to be poor with respect to the requirements. First detectability studies of the in-pool gamma spectrometer showed a limited number of isotopes with satisfactory signalto-noise ratio, viz. 90Kr, 132Te, 131I, 132I, 133I, 136I, and 137I 3.1.3. Post-irradiation data acquisition 3.1.3.1. Objectives The post-test examens are designed to supply the following information: a. End state of fuel bundle and control rod, i.e. axial and radial material distribution. b. Inventory of remaining fission products in the fuel, c. Recovery and analysis of the sequential coupon, d. FP deposit analysis in the upper part, e. Control of possible damage to shroud, pressure tube, and other components in the fuel area. 3.1.3.2. Non-destructive examens One week after the end of the irradiation the test train is transferred to the examen and conditioning station PEC. The first operation will be a gamma scan over the entire length, with three Table 1. Instrumentation Summary: TEST SECTION—POINT A+B Parameter

Instrument

Location; Comments

FPT0

FPTx(c)

Temperature

Ultrasonic thermometers

2 (14 measurements)

2

K type TCs W/Re TCs K type TCs W/Re TCs W/Re TCs

in the lower part in the rod centreline on AIC rod on guide tube on stiffener plates

Two outer corners of the bundle, seven axial positions 2 12 2 4 8

2 −(1) 2 4 8

136

THE PHEBUS FISSION PRODUCT PROJECT

Parameter K type TCs W/Re TCs K type TCs fluid " " "

Instrument

in the ZrO2 insulator temp. gradient zone " K type TCs W/Re TCs K type TCs K type TCs K type TCs Pressure pressure transducers differential pressure transducer (2) hot wire flow meters Flow rate pump rate, weighing injected quantity (H2O) On-line γ activity γ spectrometer(a) Deposit sequential coupons (a) Post test activity γ spectrometer fission chambers (+post test analysis) Fuel relocation

Location; Comments

FPT0

12 1 4 bundle inlet temp, gradient zone " upper part safety & heaters in the gas feeding lines bundle inlet/outlet injected quantity (H2, He) 1 at level −6500, at bundle outlet temp. gradient zone post test examination: PEC along the bundle length in coolant water

12 1 4 2 4 4 4(2) 10 & 11 2 1 1 1 − 6 + 4

FPTx(c)

2 4 4 4 ≤ 21 2 1 1 1 6 + 4

(a)

Research in progress (b) Development in progess (c) Not decided Instrumented fuel pins under discussion (2) Two thermocouples could be used for flow rate measurement by TC correlation technique, in the test section upper part, if requested (1)

repetitions after 90° rotations. These gamma scans are expected to supply a first impression about the degree of axial fuel displacement in the bundle area, and of FP deposits in the upper part of the in-pile section. They will establish decaycorrected axial count rate profiles of 131I, 134I, 134Cs, 95Zr-Nb, 110mAg,…. The second operation, tomographic radiography /9/, should give a detailed image of the fuel bundle damage. 3.1.3.3. Destructive examens After transfer to a hot laboratory the in-pile device will be sectioned by a number of radial cuts. The first cuts will recover the sequential coupon (to be transferred for further analysis) and separate the fuel bundle section from the upper and lower extensions. The fuel bundle section, about 1 m (FPT 0) and 1.2 m long (FPT 1), respectively, is than transferred for further sectioning. Axial position and number of radial cuts in the fuel area will be defined by the image obtained through gamma scan and radiography. Between 10 and 20 cuts are expected to be required for a detailed 3D picture of the geometry of fuel, cladding, control rod, zirconia shroud, and structural parts. This picture will be obtained by visual inspection and photography of the polished surfaces. Computerized image analysis (enhancement/processing) might be used for the analysis and for construction of 3D pictures to be coloured later on, on the basis of more detailed elemental, isotopic and chemical analysis. Transversal gamma scanning of each slice will be the next step, again preferentially connected to a tomographic computer treatment for three-dimensional mapping of a selected number of isotopes of different volatility. This exercise will include the search for, and if required, quantification of, fission products in the zirconia shroud. Five to ten slices are then selected for detailed examens, viz.: - optical and electronic microscopy coupled with energy-dispersive spectroscopy for the determination of microstructures and phases of areas of interest, e.g. fuel-cladding and control rod material-structure interactions, molten fuel areas,…. - microsonde, microdrilling with ensuing radiochemical analysis, electron spectroscopy for chemical analysis,…for determination of the chemical composition in selected areas, - additional analyses resulting from discussions between the evaluating teams and the laboratories in charge of the examens.

PHEBUS FP

137

3.1.3.4. Sequential deposition coupon The exposed surfaces of the instrument are covered with vapour and aerosol deposits. Their post-test analysis should enable the following data to be obtained: . average density and granulometry of the aerosols during the exposure period, . elemental, isotopic, and chemical composition. The analytical sequence will follow a scheme similar to those to be used for impactor plate samples, circuit tube cut-outs, filters, and deposition coupons, see chapter 3.5.3. 3.1.4. Other methods The reconstitution of events in the fuel bundle, i.e. fuel damage and FP release vs. time, will also use instrumentation data from other areas of the facility. The event “control rod failure” e.g. will not only cause a variation of the fission chamber signals but also of the aerosol monitor (paragraph 3.2.2.1.), fuel clad failure will be seen by gamma and mass spectrometers, clad oxidation by mass spectrometer and hydrogen meter, etc.… 3.2. Circuits 3.2.1. Thermal-hydraulic instrumentation In the configuration for the first two tests (FPT 0 and FPT 1) the experimental circuits contain a simulated steam generator. They are instrumented for temperature, pressure and flow rate (Figure 2). 3.2.2. Fission product instrumentation 3.2.2.1. Aerosol instruments The main instruments, both at points “C” and “G”, are modified ANDERSEN MKII impactors and filters. These samplers are mounted inside heated boxes with temperature control, adjusted equal to, or slightly above, the respective main line temperatures. They are operated at strategic points in time during the FP release phase of the test. Sampling duration and flow rate are pre-determined by calculation with the aim of collecting between 1 and 100 mg of aerosols per instrument. The instruments are recovered after each test and transferred for post-test analysis (see paragraph 3.5.). An on-line optical aerosol density monitor is incorporated at the inlet to the experimental circuit. The instrument measures the attenuation of light across two path lengths during the bundle release phase. It has been adapted by EG & G, Idaho Falls, from a similar instrument used in PBF test SFD 1-4 /10/, and is expected to supply semi-quantitative information on the instantaneous aerosol population and to support readings of other instruments following events like control rod burst, zircaloy melt-down and gross fuel relocation. 3.2.2.2. Grab samples Three “bulbs” each are connected to the main line, at points “C” and “G”. They are 30 cm3 capsules preceded ‘by particle filters. All six capsules are mounted inside temperature controlled boxes, operating at 150°C and 700°C, respectively. They will be filled with main circuit gas at selected points in time during the FP release phase, by draining from the circuit towards the exhaust system REPF 502—REPF 503. Filters and capsules are recovered after each test and transferred for post-test analysis of solid, liquid and gaseous samples (see paragraph 3.5.)

138

THE PHEBUS FISSION PRODUCT PROJECT

Fig. 2. Instrumentation plan for the FPT0 test Primary circuit

3.2.2.3. On-line gamma spectrometers An overall scheme had been developed early on in order to accommodate the necessary space inside the FP building for collimators, GeLi detectors with accessories and shielding, liquid nitrogen supply, etc. Out of the total possible number of positions those to be equipped with fully operational instruments are selected for each test. Two characteristic problems of on-line gamma spectroscopy are: - isotope detectability in a “noisy” environment - discrimination between deposited and moving gamma emitters in the component in front of the collimator. Detectability of interesting isotopes can be assessed using a code system developed by CEN Grenoble (Fig. 3). The code system has been validated by calibration experiments and through evaluation of HEVA Out-pile FP release test data. The discrimination between deposited and moving gamma emitters is required in order to follow the kinetics of gas-borne aerosols.

PHEBUS FP

139

Fig. 3. Flow sheet of detectability calculations

The method selected at present consists in collimating, at the same circuit location, onto two pipe sections with different surface-to-volume ratios /11/. Tests simulating gas volume and wall deposits with radioactive sources confirmed the validity of the approach. 3.2.3. Mass spectrometry On-line analysis of hydrogen and noble gases in the test fuel bundle effluent line had been used with same success in the PHEBUS severe fuel damage (CSD) experiments. The instrument was a QUADRUVAC Q 100, connected to the main circuit via a two stage pressure reducing system. For PHEBUS FP a similar instrument will be used to monitor H2, Xe, Kr and fission product components inside containment vessel REPF 502 (see paragraph 3.3.3.) Apart from the two noble gases the instrument is expected to detect I2 and ICH3.

140

THE PHEBUS FISSION PRODUCT PROJECT

Provisions are made to connect mass spectrometers at points “C” and “G” where the range of potential species includes Ag, In, Cd, CsOH, SnO2, and boric acid. A number of problems make the application of the instruments very difficult: - high aerosol, low vapour population density, resulting in plugging and detectability problems, - installation near the point of sampling required, resulting in operating and maintenance problems, due to high temperatures and radiation. Investigations are presently continued into the feasiblity of meaningful measurements with an expensive apparatus. 3.2.4. Post-test sample recovery The following inventory of samples has to be recovered from the circuits at the end of each test: -

6 gas-filled bulbs and their pre-filters, 4 liquid-filled bulbs, 4 impactors, 3 thermal gradient tube liners, the steam generator U tube, the removable internal of the horizontal line.

The two latter items are stored in the carroussel (“lazy susan”) for several months, the other samples are transferred to the CECILE hot cell and on to hot laboratories for post-test analysis (see paragraph 3.5.) 3.3. Containment vessel 3.3.1. Thermal-hydraulic instrumentation 3.3.1.1. Scope The containment vessel REPF 502 has been designed for investigation over a time up to one week, of FP behaviour after release from the point of leakage in the Reactor Cooling System. The main phenomena, aerosol agglomeration and settling and iodine species migration between gaseous and liquid phases, are controlled by thermalhydraulics, sump pH, radiation fields, FP concentrations,… These parameters have to be measured or controlled in order to enable the events in the vessel to be understood. 3.3.1.2. Instrumentation plan Fig. 4 The vessel is instrumented with a large number of thermocouples on the main walls, the condenser walls, in the gas volume and in the sump. Pressure and humidity are measured, together with condensate flow rate and sump water level. Vessel wall, sump, and condenser temperatures are controlled by independent organic liquid loops fitted with own instrumentation for temperature, flow rate, and pressure. Their thermal balance will be used to interpret the behaviour of the vessel: injected and lost energy, rate of condensation, sump boiling rate,…

PHEBUS FP

141

3.3.2. FP instrumentation 3.3.2.1. Aerosol instruments Impactor and gas phase filters are similar to the instruments used in the circuit. Their operation will depend on the aerosol density kinetics in the vessel, determined by pre-calculations, with the aim of collecting not more than about 10 mg per impactor plate or filter. Three instruments have been considered for continuous aerosol density measurement or a discontinuous determination with increased time resolution: . a granular bed impactor “IMLIGRA” /12/ with a dedicated gamma spectrometer following the build-up of activity in the different stages of the instrument, . the INSPEC inertial aerosol spectrometer /13/ which can prepare a large number of filter plates with particles spread out according to their size, . a light extinction aerosol density meter (photometer) with an adjustable optical path. All three instruments are still in the development phase. More aerosol samples will be collected on the filters before the gas grab samples and on a sequential coupon placed in the lower part of the containment vessel. The sequential coupon is designed to take deposition samples at different points in time and to be transferred for analysis after each test. 3.3.2.2. Grab samples The standard 30 cm3 capsules are preceded by standard filters whereas the four condensate and the four sump water capsules are gravity-filled with soluble and insoluble matter. 3.3.2.3. Iodine speciation instruments Radiolysis in the container vessel sump and the presence of painted surfaces are expected to generate volatile species of iodine which would build up in the vessel atmosphere. The samplers provided are of the Maypack type. The materials to be used are presently under test. The choice includes: . glas fiber mats for the two filters, . Ag-plated Cu wire or Ag membranes in the I2 stages, . KI-impregnated active charcoal or Ag-impregnated zeolite in the ICH3 stages. 3.3.2.4. On-line gamma spectrometers As for the circuits these instruments are assigned an essential role for following the FP kinetic in the containment vessel. One instrument will be collimated onto the condenser and, alternatively, onto the vessel gas space, the second onto the painted coupon in the sump and, alternatively, onto the sump water space. In either case the surface measurements will overlap with the bulk activity of the surrounding medium. The distance between instrument and containment vessel implies possible further losses of accuracy. The same remark applies to a projected third instrument looking at an “on-line” iodine speciation filter, presently in the development phase. The readings will, moreover, be affected by the high background radiation of the inlet filter. 3.3.2.5. Miscellaneous Condenser structures and painted coupons out of the sump will be recovered after each test for post-test analysis.

142

THE PHEBUS FISSION PRODUCT PROJECT

Fig. 4. Instrumentation plan for the FPT0 test Containment vessel

Gaseous and liquid effluents, from containment vessel REPF 502 are to be recorded in order to establish an overall FP balance. 3.3.3. Mass spectrometry An instrument for amu ≤ 200 will be used, similar to the one applied for the CSD tests. It will essentially detect H2, Kr and Xe, the concentration of volatile iodine species is probably below detection limit (see paragr. 3.2.3.). 3.4. Outlet system and atmosphere vessel The experimental instrumentation of these systems is redundant with the measurements taken in the containment vessel. It is therefore reduced to a post-test gamma scan of absolute filters and condenser, post-test analysis of a water sample from the

PHEBUS FP

143

condenser and a safety-related analysis (hydrogen and fission gases) of the atmosphere tank contents prior to its release to the stack. 3.5. Post-test sample analysis 3.5.1. Overall flow charts The handling and analysis flow sheets shown on fig. 5–7 have been developed on the basis of the following input data: -

a compilation of requirements, for the interpretation of PHEBUS FP tests, literature review on techniques used in earlier tests /2/, a review of analytical techniques /5/, discussions with various laboratories.

The flow sheets are preliminary and will be finalized in future discussions. 3.5.2. Examination and conditioning cell The post-test analyses imply numerous transports over distances between 2 and 1500 km (Cadarache, Grenoble, Saclay, Karlsruhe, Winfrith,…) according to the detailed test plan. In view of the large number of samples (150 per test) and of the delicacy of solid aerosol deposits the overall operation represents the risk of: . loss of material due to handling and transports, . loss of sample identification . loss of I131 due to long transfer times. The CECILE* facility was designed, in collaboration with CEN/ Grenoble, to avoid such losses. The cell is equipped with manipulators, teleoperated tools, a video system and an annexed gamma spectrometer, as well as with the necessary ventilation, with filters, a storage well, container docking and manual inlet/outlet locks, lights, shielded windows, etc…. 3.5.3. Instrument analysis survey Among the large number of instruments which could be used for post-test FP sample analysis the following preliminary selection is proposed for the first PHEBUS test: - gamma spectroscopy for initial identification and finger printing of a sample (to be cross-checked by the receiving laboratory), and for quantitative isotopic/elemental analysis. The instrument will be used for all samples of the programme, eliminating those with obvious malfunctions from further analysis. It is usually quite sensitive and accurate; - solid samples (aerosol deposits, filters with insolubles from liquid sampling) should be prepared for the scanning electron microscope with energy dispersive spectroscopy and automated particle recognition attachments (SEM/EDS/PRC). This combined instrument yields results on the morphology of the deposits and on their elemental composition; - selected samples will be transferred towards instruments for more accurate elemental analysis: X-ray fluorescence spectroscopy (XRF) or inductively coupled plasma emission spectroscopy (ICPES) and/or for chemical speciation analysis: X-ray diffraction (XRD), electron spectroscopy for chemical analysis (ESCA), or one of the mass spectrometers (SIMS, SSMS,…); - neutron activation analysis (NAA) is an inexpensive and accurate method for isotopic/elemental analysis of non-gamma emitters (e.g. 90Sr), fissile particles, and decayed iodine;

* Cellule d’Examen, de Controle de l’Iode, de Lotissement et d’Expédition

144

THE PHEBUS FISSION PRODUCT PROJECT

Fig. 5. Flow sheet for solid samplers (filters, coupons, pipe samples, TGT liners)

- alpha (and beta) spectrometry and autoradiography can be used for searching activities in selected debris particles. 4. DEVELOPMENT ITEMS 4.1. Scope The present instrumentation plan, in particular its specific form for FPT 0, is limited to devices sufficiently qualified for the anticipated operating conditions. Research, development, and qualification will continue in order to improve data quality, as the Programme proceeds.

PHEBUS FP

145

Fig. 6. Flow sheet for Maypack (iodine speciation samplers)

4.2. Survey Specific laboratory test programs have been operated a.o. in the following areas: -

flow rate measurement by thermocouple signal correlation and ultra-sonic flow meters (feasibility, accuracy), sequential coupon (design, aerodynamics, aerosol impaction), miniature fission chambers for control rod and fuel displacement detection (detectability calculations), gamma spectrometry (detectability, discrimination methods, design, data acquisition and processing), mass spectrometry (detectability, connection, controls), on-line aerosol instruments (design, feasiblity, accuracy), impactors (design, calibration, accuracy, handling), thermal gradient tubes (detectability calculations, design, interpretation) iodine speciation filters (filter material selection, design, calibration, concept of an on-line instrument), post-test FP sample analysis (detectability, calibration, accuracy, handling problems).

At present most of these programs are sill active. So far the test results indicate problems for the following instruments precluding their application for the first in-pile test . high temperature ultra-sonic flow meters, . high temperature mass spectrometry, . aerosol impactor with on-line gamma spectrometer monitoring.

146

THE PHEBUS FISSION PRODUCT PROJECT

Fig. 7. Flow sheet for sampling bulbs (liquid and gaseous grab samples)

Major criticism of the present instrumentation layout concerned an insufficient time resolution of the transient chemical speciation development of fission products and other releases from the test fuel bundle. For the time being, this problem can partly be solved by a limited increase of the number of filters (and capsules) at points “C” and “G”. 5. CONCLUSIONS The experimental instrumentation for the Phebus Fission Product Programme has been designed using as an input data the scientific objectives on one hand and available technologies on the other hand. A number of “second best” solutions, determined by unavoidable constraints of space, operating conditions, timing and funding, have to be incorporated. Future improvements will be possible as results from the first test and from ongoing R & D activities become available. In its present stage, however, the plan developed will adequately meet the objectives set out by the Programme. 6.

PHEBUS FP

147

REFERENCES 1 2 3 4 5 6

7 8 9 10 11 12 13

. R.ZEYEN, “Phebus FP, Experimental instrumentation”, to be published (1991). . A.TANI and P.v.der HARDT, “Bibliographic study for post-test analysis of samples from fission product release tests”, to be published (1991). . G.HAMPEL, G.POSS, and H.K.FROHLICH, “General and preliminary thermo-hydraulic, hydrogen, and aerosol instrumentation plan for the Phebus FP project”, EUR 12397 (October 1989). . G.HAMPEL and G.POSS, “Development of advanced instrumentation of the Phebus FP project. Preliminary studies”, EUR 12396 (October 1989). . B.R.BOWSHER and A.L.NICHOLS, “PHEBUS FP, SCA-1, Review of analytical techniques to determine the chemical forms of vapours and aerosols released from overheated fuel”, EUR 12399 (December 1989). . J.Y.BLANC, B.CLEMENT, R.AGAISSE, and G.BARTHELEMY, “Temperature measurements in French nuclear safety experiments: Results from PHEBUS Severe Fuel Damage Program”, 3rd Intern. Symp. on Temperature and Thermal Measurement in Industry and Science, Helsinki, Finland, September 1990. . H.S.TASMAN, M.CAMPANA, D.PEL, and J.RICHTER, “Ultrasonic thin-wire thermometry for nuclear applications”, in “Temperature, its Measurement and Control in Science and Industry”, Vol 5, 1991–1996, Amer. Inst. of Physics, New York 1982. . “Post-irradiation examination data and analyses for OECD LOFT Fission Product experiment LP-FP-2”, OECD-LOFT-T-3810 (December 1989). . L.STEINBOCK, “Tomography of nuclear fuel experiments with an electronic line scan camera”. Nuc. Engrg. Des. 118 (1990) 9–16. . D.A.PETTI et al., “Power Burst Facility (PBF) Severe Fuel Damage test 1–4, Test results report”, NUREG/CR-5163, EGG-2542 (April 1989). . T.W.PACKER and B.H.ARMITAGE, “Application of gamma-ray spectro-scopy…” to be published (1991) as an EUR report. . D.BOULAUD, “Use of granular beds in the inertial impaction regime for aerosol size distribution measurement”, to be published in J.Aerosol Sci. (1991). . V.PRODI, “Characterisation of nuclear aerosols”, J. Nucl. Mater., 166 (1989) 189–198.

CEA ANALYTICAL ACTIVITIES: HEVA, PITEAS, MINICONTAINMENTS C.LECOMTE and G.LHIAUBET IPSN/DPEI, CEA, Fontenay aux Roses, France

SUMMARY This paper presents analytical experiments concerning fission product transport; these experiments have been conducted in France since 1985 in order to develop and validate the modelisation of fission product behaviour during severe accidents, as it is realized in the French system ESCADRE. The main experiments which are described are respectively HEVA for fission product and material release from the fuel during core degradation and PITEAS for aerosol behaviour in the containment. Finally, the analytical tests for iodine chemistry are described and a proposal for a thorough iodine chemistry program in PHEBUS-PF is presented. 1. INTRODUCTION The ESCADRE system is the French code system used by IPSN, as technical support to the French safety authorities, for the overall analysis of severe accident sequences liable to occur on pressurized water reactors. Its main objective is to determine, qualitatively and quantitatively, the potential source term to the environment in case of severe accident. It also allows to study the efficiency of various preventive or mitigative measures. In order to fulfil this goal, it is necessary to predict quantitatively the fission products’ location, at any time and for every containment failure mode. For this, the thermal-hydraulic properties of the carrier fluid which governs fission product phenomena have to be computed, either for the circuits, or for the containment. Then, the fission products characteristics—i.e. the physical and chemical properties— are computed, at each stage from the core to the containment and to the environment. The organization scheme of the ESCADRE system is represented in Figure 1. The different codes belong to two categories: • Thermal-hydraulic codes: - VULCAIN for the primary circuit thermal-hydraulics and core degradation; VULCAIN also calculates the fission products release during core degradation; - JERICHO for containment thermal-hydraulics; JERICHO calculates pressure, temperature, atmospheric composition in the containment, from mass and energy flow rates coming from the other modules; it can also describe hydrogen deflagration phenomena; - ECROUL is a module based on mass and energy balances which can derive the time between a given core degradation state and the beginning of corium-concrete interaction; - WECHSL calculates corium-concrete interaction and the resulting gas and energy flow rates to the containment. • Fission product codes: – VULCAIN, already mentioned, for which fission product release and core thermalhydraulics are strongly coupled; – SOPHIE, which stands for vapour fission product behaviour in the pipes; – AEROSOLS CIRCUIT, concerning the specific depletion of aerosols in pipes (primary or secondary circuit, safety injection lines); – AEROSOLS/B2 which calculates the behaviour of aerosols in the containment; – IODE, which is devoted to the chemistry of iodine compounds under radiation in the containment.

EA ANALYTICAL ACTIVITIES

149

Fig. 1. Organization of the code system ESCADRE

– Finally, the code ALICE can evaluate the activity transport and the dose rate due to the fission products in different buildings, as a function of time, while CONRAD gives the radiological consequences in the environment, by taking into account the atmospheric dispersion of the fission products. As mentioned earlier, the ESCADRE code system is an operational tool aimed at predicting the main parameters describing the potential source term in the case of severe accident; the strategy adopted by IPSN has been to develop an engineer code, the degree of detail in the modelisation being adapted to both the final needs and the possibility of obtaining pertinent data from the experiments, giving the general instrumentation limitations. So far, the code system ESCADRE has helped to prepare a number of experiments; it also relies on the results of an important program concerning behaviour of fission products and aerosols. A scheme of the different French programs related to the behaviour of fission products is shown in Figure 2: analytical and global experiments complement each other to validate fully the codes used for safety analysis. 2. THE HEVA PROGRAM The first phenomenon of interest when studying fission product behaviour is the release of fission products from the fuel during the core degradation; the approach adopted in France was not to develop a very mechanistic description of fission

150

THE PHEBUS FISSION PRODUCT PROJECT

product release, which would always suffer a tremendous lack of accuracy, but rather to measure the global result of fission product escape from the element, under conditions as realistic as possible. This was the reason for the initiation of the HEVA program, which extended from 1983 to 1989. The three main areas of interest were as follows: • fission product and material release rates; • chemical species identification; • fission product aerosol size distribution (as a function of temperature). The experimental set-up was a furnace located in a hot cell (Figure 3); the sample was high burn-up (36 000 MWd/t) fuel pellets, with cladding and, in some experiments, control rod material. These pellets were re-irradiated for eight days in the SILOE reactor in order to have a realistic amount of short life fission products, then transferred in the HEVA cell and heated up to 2400 K in a steam or hydrogen gas flow. Eight tests have been conducted under various conditions. Each experiment was designed to investigate one parameter at a time. For example, for the HEVA 08 test, the thermal-hydraulics conditions were the same as HEVA 05 (only fuel) and HEVA 07 (only AIC). For HEVA 08, control rod material was introduced in the form of silver, indium and cadmium pellets located under the fuel sample. Some results are given in Figures 4 to 7 as examples. Figure 4 shows the metallographic aspect of the central pellet and the lower pellet after the test; in this case, a complete cladding oxidation was observed, together with fuel fragmentation. Figure 5 gives the fission product released fractions, while Figure 6 gives the same information for the control rod material; high cadmium release and low silver release can be observed; kinetics can be found in Figure 7. A large working program is still going on to perform the chemical analysis of numerous samples from the HEVA experiments and interpret the results so as to develop confident release correlations adapted to various situations: oxidizing or reducing atmosphere, influence of control rod material, effect of heating-up kinetics. In the future, the VERCORS program will bring new information for fission product release at higher temperatures (up to fuel melting). 3. THE TUBA EXPERIMENTS The aim of the TUBA experiments, which have been in progress since 1988, is to derive validated deposition laws for aerosols deposits in tubes, with inner diameter 20 mm (tuba, simulation of SG tubes) and 400 mm (TRANSAT, simulation of large pipes) Validated correlations are not available for this range of diameters, although a lot of data exist for smaller diameters (instrumentation lines).

EA ANALYTICAL ACTIVITIES

151

Figure 3: HEVA experimental facility

The TUBA experimental facility shown in Figure 8 consists of a system of steam generator simulating tubes crossed by a steam/air gas flow. The first section allows to prepare the aerosol injection at a given temperature. In the test section, a cooling circuit creates a temperature gradient between the carrier gas and the walls of the tube; gas and wall temperatures can be controlled up to about 800 K. The test matrix includes the effect of laminar or turbulent flow, the effect of thermal gradient magnitude and the effect of steam condensation on the walls. Up to now, the effect of thermophoresis under laminar conditions has been investigated under conditions representative of an accident sequence involving the steam generator in a PWR. Figure 9 shows a typical result of TUBA measurement; the duration of the test was two hours, with a wall temperature of 300 K and a gas temperature of 674 K at the beginning of the test section. The experimental CsI plated mass is compared to the models of Brock and Derjaguin respectively. Future work will include new experiments under diffusiophoresis conditions (with a variable ratio of steam to non condensible gases) and interpretation of the test measurements, to include new correlations in the ESCADRE system. The TRANSAT experiments, which are currently at the design stage, will use a system of large pipes (300 mm) crossed by a steam and air gas flow, with gas and wall temperature control and aerosols injection; these experiments will be specific for large pipes like safety injection, primary circuit or venting lines. The aerosol deposition velocity will be measured as a function of thermal gradient and steam condensation, among other parameters. These experiments are designed to meet the greatest possible number of reactor events; the measurements could start from 1992. 4. THE PITEAS PROGRAM The PITEAS program was initiated around 1984 in order to identify precisely some characteristics of aerosols’ behaviour in the containment in the presence of steam; namely, experiments were designed to study the effect of steam condensation, aerosol diffusiophoresis, on soluble particles growth velocity, evolution of droplet size distribution and finally removal of airborne aerosols.

152

THE PHEBUS FISSION PRODUCT PROJECT

Fig. 4. Metallographic aspect of the central pellet and lower pellet after test completion

Fig. 5 HEY A 08: fission product released fraction

The experimental facility, shown in Figure 10, is a 3 m3 vessel with aerosol source and steam moisture control; the wall temperature is also controlled. The pressure can be monitored up to 5 bars and the temperature up to 140°C. Examples from two different types of experiments are given below: • PITEAS PDI05: the objective of this experiment was to investigate the diffusiophoresis of soluble CsI aerosols when the walls of the vessel were cooled from 120°C to 100°C. The temperature transient is shown in Figure 11; Figure 12 shows a comparison between experimental results and computed ones using the Schmitt and Waldmann equation. The prediction could be greatly improved by using different particles/vapour relationships. • PITEAS PCON 01: this experiment was realized to study the effect of vapour condensation on CsI aerosols; the suspended concentration versus time is shown in Figure 13.

EA ANALYTICAL ACTIVITIES

153

Fig. 6. HEVA 08: control rod released fraction

Fig. 7. HEVA 08: kinetics of control rod material release

The results will serve to validate the formalism introduced in the AEROSOL code of the ESCADRE system. 5. IODE ANALYTICAL EXPERIMENTS The contribution of iodine to the source term in the case of hypothetical severe accidents on pressurized water reactors is of extremely high importance in the short term of an accident, as it is the determinant for counter measures which could eventually be taken. The chemical form of this iodine source term determines the possible contamination paths and the possibility to improve the safety, i.e. reduce the releases by appropriate measures; for example, for iodine combined with cesium or other

154

THE PHEBUS FISSION PRODUCT PROJECT

Fig. 8. TUBA experimental facility

compounds to form aerosols, the depletion of suspended aerosol concentration in the containment as a function of time, combined with as high as possible leaktightness of the containment, is the key factor to limit possible releases. On the contrary, gaseous forms of iodine like I2 and ICH3 are less subject to retention by filtration materials; the concentration of I2 in the gaseous phase can be reduced by the use of spray in the containment, while ICH3 is insensitive to it. The global scheme of the iodine research program at CEA is the following: • analytical experiments designed to develop models and give input data to the codes, • semi-global experiments with simulated iodine source, aimed at code validation by means of the variation of numerous parameters,

EA ANALYTICAL ACTIVITIES

155

Fig. 9. Results of the TUBA TT29 experiment

• global experiments with real iodine source (PHEBUS), designed to improve the validation and solve the uncertainties derived from the use of simulated sources in the analytical experiments. A number of analytical experiments have therefore been realized at CEA/IPSN to investigate the physico-chemical behaviour of iodine under conditions representative of a severe accident. At first, these experiments were small scale experiments, designed to evaluate separately the different mechanisms involved in the chemical transformations and physical transfers of iodine. Generally, glass bottles with one or two compartments were used; these bottles could be placed in a radiation generator, with an iodine/air/steam flow. Painted, bare steel or concrete plates could be inserted in the bottle. The measurements so far have addressed the iodine deposition/resuspension velocity on various surfaces under different atmospheric compositions, the molecular iodine radiolysis velocity and the organic iodine formation. First results have exhibited a very high potential for iodine trapping from the paints; the specific paint used was shown to be a very sensitive factor; also, submerged paintings appeared to have a strong influence. Radiolysis experiments have confirmed the major role of pH in the sump to govern the rate of gaseous iodine formation. An important experimental result is the sensitivity of both I− and IO3− to radiolysis, as evidenced in Figure 14. A new series of experiments named CAIMAN is currently under consideration: these experiments would be realized under gamma irradiation, with iodine injection, paintings, and wall temperature control. The objective is to measure the global behaviour of iodine and the partition between gas, water, paintings, and leakages, the chemical forms being taken into account. The influence of dose rate and pH will be essential topics.

156

THE PHEBUS FISSION PRODUCT PROJECT

Fig. 10. The PITEAS facility

6. MINI-CONTAINMENT PROPOSAL Given the high number of chemical reactions involved in iodine chemistry, the fission products emitted during the PHEBUS FP tests could be a unique way to explore the behaviour of iodine in representative conditions that could not be modelled during analytical experiments, as the composition of fission products released from the primary circuit during a severe accident is not precisely known, as far as chemical form and kinetics are concerned. For that reason, CEA has proposed to consider the possibility to add in some PHEBUS FP tests two vessels around 300 litres each in volume; these vessels would be in parallel to the main PHEBUS FP containment and would be dedicated to iodine chemistry for a set of experimental parameters: pH, paintings, rapid venting, pool scrubbing, temperature, in particular. Such data would be particularly relevant for the validation of analytical experiments, as it is the case for the part of the PHEBUS measurements made in the REPF 502 tank; the number of parameters involved in iodine chemistry makes special minicontainments necessary, in order to increase the amount of information obtained from the five PHEBUS tests. With regard to semi-global tests (tests performed with a simulated iodine source), other fission products and structural material interact with iodine chemistry, either directly or indirectly, by interfering with one of the numerous reactions involving iodine (more than a hundred depending on the author); their presence in the PHEBUS FP facility is the unique feature of these experiments and the opportunity to study iodine chemistry in the most representative conditions possible. Compared to other experimental facilities making use of simulated fuel, the PHEBUS FP facility has several advantages: • chemical representativity of the initial fission products in the fuel; • representative kinetics of fission products and structural material release; • representative initial form of iodine in the sump.

EA ANALYTICAL ACTIVITIES

157

Fig. 11. Temperature transient in PDI05 experiment

Fig. 12. CsI plated mass in PITEAS PDI05 experiment

The effect of pH and paintings is also recognized as an important parameter for the formation of gaseous forms of iodine; usually measured values of iodine partition coefficients between an aqueous and a gaseous phase differ by three to four orders of magnitude for pH values varying between 4 and 9. Nevertheless, the effect of temperature and paintings on this partition coefficient is not yet clear; values for temperatures and dose levels representative of a severe accident are still needed and could be investigated here. 7. CONCLUSION A number of experimental programs concerning fission product aerosol behaviour has been conducted at IPSN for some years in order to improve the knowledge on these topics; safety needs have been fulfilled in terms of development of a qualified database and qualified tools to estimate the potential for radioactive releases in the case of a severe accident. The PHEBUS FP

158

THE PHEBUS FISSION PRODUCT PROJECT

Fig. 13. CsI mass concentration versus time in PITEAS PCON01 experiment

Figure 14: Radiolysis of IO3− and I−

program will contribute to improve the models and give a high degree of confidence to the physical laws and computational tools which have been derived from analytical experiments in conjunction with international programs, by ensuring a wide range of validity to the operational means used for safety analysis in different countries.

CEC SUPPORT ACTIVITIES: EC SHARED COST ACTIONS AND OTHERS P.FASOLI-STELLA, A.MARKOVINA Commission of the European Communities Thermodynamics and Radiation Physics Division, Joint Research Centre, Ispra, Italy

SUMMARY The Phebus FP programme has a central role in the research of the European Commission in the area of Source Term. The Joint Research Centre of the CEC is directly involved in the project work and test analysis and is putting in a strong effort to stimulate the widest cooperation of the most qualified European laboratories. Some of these have been active from the beginning in the project working groups, others are performing research funded through the Shared Cost Actions (SCA) in support of the project itself or in areas which are closely related. The paper illustrates the guidelines in choosing and performing the different activities related to the Phebus FP Project and summarizes the main results of all these cooperations in order to better identify how to continue in the future. 1. INTRODUCTION The Commission of the European Communities funds a research programme on reactor safety which is partly performed at the Joint Research Centre (JRC) and partly in the National European Laboratories through the Shared Cost Actions (SCA). A considerable part of this research is focussed on severe accidents and in particular on the study of the Source Term, which with the programme 1988–1991, became one of the main subjects for the amount of the resources involved and because it is organized to allow and stimulate the widest cooperation among the European laboratories. As is well known, the most important initiative in this field was to join the French Phebus FP programme in 1988 with a substantial funding, as well as with participation of the JRC staff in the project and analysis work. Phebus FP is a programme of integral in-pile experiments, which are very expensive and difficult and are performed with the final objective of validating models and computer codes to be used for the calculation of the Source Term in case of a severe accident. The phenomenology to be modelled is however so complex that the codes presently available ask in reality not only for validation, but also for substantial developments which, to the extent possible is being performed during the programme preparation. This leads to the need for additional separate-effect or small-scale experiments which have to guide in model development and in many cases also in a better identification of the most important phenomena to be studied in the in-pile tests. The JRC is actively involved in the definition of the scientific objectives and test matrix of the project, as well as in the test calculations and in coordination of the EC partners contribution to the project. In the long term, a best estimate package of computer codes, validated through the numerous experiments already performed or in progress (including Phebus FP), developed in close cooperation and with the substantial contribution of the EC Member States, based on a large consensus on the modelling and numerical techniques adopted, should result from this effort and be the “European” tool for severe accident and source term calculations. As an essential part of its effort, since 1987 the JRC has been launching a number of diversified SCAs in support to the Phebus FP project (e.g. instrumentation, scoping calculations, specific aerosol/FP researches) or more in general aiming at enriching or complementing the project itself (e.g. modelling, aerosol/FP physics and chemistry). Table 1 gives the complete list of SCAs launched in the period 1987–1990. The paper will describe first the work done and still under way to define and continuously update the test matrix (section 1) and the support given to the instrumentation (section 2), then it will review the experimental work launched through the SCAs (section 3) and the effort on model and code development (section 4). The review will concentrate in particular on the actions which are not described in other parts of the seminar.

160

THE PHEBUS FISSION PRODUCT PROJECT

2. THE TEST MATRIX DEFINITION In other parts of this seminar, the test matrix is discussed in detail. Here only the approach followed and the support actions will be illustrated. The first step was the consideration of basic accident scenarios in large modern PWRs: -

large break loss of coolant (AB), small break loss of coolant (S2D), special transients (TMLB’), containment by-pass (V), steam generator pipe rupture.

In 1987, before the signature of the Phebus FP agreement, the JRC funded a first set of SCAs to obtain sequences calculations performed by national institutions on national reactor designs. The objective was to identify the main phenomena and range of parameters characterizing different risk dominating accident sequences (Phase A). The 16 sequence calculations made available by 10 teams organized in 5 national groups, for different reasons (mainly difficulty in modelling some phenomena) were only related to the first 3 types of scenarios (AB, S2D, TMLB’) and did not take into consideration the presence of the Steam Generator in the primary circuit. The results of this exercise, jointly analysed by CEA and JRC (1) gave a picture of characteristics and timing for fuel melting, FP release, hydrogen production, FP/aerosol transport and deposition in the core region, cooling systems, containment building.They allowed also the identification of the components or systems to be simulated in the experimental circuits (pressurizer, low pressure injection system components, etc.) in order to reproduce phenomena relevant to class of sequences. Immediately after the conclusion of phase A, a new exercise was launched by the JRC. (phase B), namely the calculation of the proposed Phebus FP tests to verify to what extent the dominant phenomena could be represented in the proposed lay-out and experimental conditions. This procedure allowed a verification of the adopted scaling criteria. Table 1 SCAs Source Term 1987–1990 Topic and beginning

End

Organisation

Title

Instrumentation 1987

1988

Battelle

General Thermohydraulic, Hydrogen and Aerosol Instrumentation Plan for Phebus FP.

1988

Siemens(KWU)

1988

Battelle

Scoping calculation 1987 (Ph. A and B)

1989

Radionuclides Measurements for Phebus FP. Development of an Advanced Instrumentation for Phebus FP Project: Preliminary Studies. Un. Pisa/ENEL

1988

AEA Winfrith

1989

CIEMAT/UPM

1989

Siemens/GRS/I KE

1989

ENEL

Best-Estimate scoping calculations for Phebus FP Tests and Nuclear Power Plant Analysis. Preparatory Studies of in-pile Integral Tests on FP in Severe Accidents and Pretest Calculations through Detailed Codes for Phebus FP Scoping Calculations. Preparatory Studies for Phebus Fission Products Project. Influence of the Aerosol Coming from the Core-Concrete Interaction on the Aerosol

Preparatory Studies of in-pile integral tests on FP Transport in Severe Accidents (e.g. Phebus FP)

CEC SUPPORT ACTIVITIES

Topic and beginning

End

1989

ECN

Instrumentation 1988

1990

1990

AERE Harwell

FP Chemistry 1988– 1989

1991

Modelling 1988

1991

IKE

1990

UPM

1990

SCK/UPM

Informatics 1989 IKE

Organisation Title Depletion Mechanism in a PWR Containment. Contribution to Phase B of Phebus FP Scoping Calculations. Lavoro Ambiente In-situ High Resolution Particle Sampling by Authomatic Large Time Sequence INSPEC. Development of Existing Measurement Techniques for Application in the Phebus FP Project. AEA Winfrith Chemistry Studies in Support of Phebus FP: Multicomponenl Aerosol Behaviour. ENEL Development of Computer Models on the Chemical and Physical Behaviour of FP and Aerosol in the Primary Coolant System of a LWR Plant during a Severe Accident. Modelling and Code Development for the Improved Description of FP and Aerosol Release during LWR Core Heatup and Degradation. Modelling the Chemical Behaviour and Resuspension of Cs and I species in the Reactor Coolant system under Severe Accident Conditions. Modelling the Chemical Behaviour of Te species in the Reactor Pressure Vessel and the Reactor Cooling System under Severe Accident Conditions. CISI Code Package Architecture (ESTER).

Development of Tools for the Interaction of Codes for Source Term Predictions (ESTER).

Modelling 1989– 1990

AEA Winfrith

Thermochemical Data 1989

1991

AEA Winfrith

1991 Thermochemical Data 1990

ECN

Thermodynamic Data. AEA Winfrith

V.U.B.

Laboratory Measurement of Thermochemical Properties of Selected Reactor Materials and Fission Products with the mass spectrometric Knudsen cell method. Thermochemical Data Acquisition.

ECN

161

Development of a New Module for treating Chemical and Physical Aspects of FP Behaviour in the Primary Circuit (VICTORIA). Thermochemical Data Acquisition. Thermochemical Data Acquisition.

162

THE PHEBUS FISSION PRODUCT PROJECT

Topic and beginning

Organisation

Title

Computation of PWR sequences 1990

UPM

ESTER modules + Chemistry Modelling 1990

CEA/IPSN Cadarache

Analysis of Severe Accident Sequences in the Spanish ASCOIINPP with MELCOR code. Adaptation of the codes JERICHO, AEROSOL B2 and JODE to ESTER Informatic Structure.

IKE

VINCOTTE

UPM

AEA Winfrith

End

Implementation of selected KESS modules into ESTER code package and contribution to the improvement of FP models for the core region. Consistency Quality Assessment of Modules or Codes coupled through ESTER. Analysis, Improvement and Validation of Models to Quantify Phenomena Related to FP/Aerosols. Development of Models to follow Vapour-Aerosol Reactions and Iodine Chemistry.

Pool scrubbing 1990 ENEA Casaccia SRD Culcheth

Convenio Proyecto LACE

Validation of Pool Scrubbing Models.

Participation in the SPARTA Programme. Status of Research and Modelling of Water Pool Scrubbing.

The circuit lay-out was defined at that time with a scale down factor of 2000 for the flow rate and the containment volume; boundary and initial conditions derived from phase A were specified separately for the fuel bundle, primary circuit, and containment vessel, in order to perform independent calculations. The teams were invited to use to the extent possible the same codes used in phase A (with the necessary modifications to take into account differences in geometry, test control and test procedures); each team could explore by sensitivity calculations or using different analytical tools, specific aspects concerning the dimensioning of the experimental circuits. The comparison of phase B calculations was not easy, because many calculations were performed under different assumptions or boundary conditions, however, important points to be taken into account were observed: - for the bundle the results were largely different after the onset of rapid oxidation; the heat generated in the bundle by fission and oxidation was mainly lost through the shroud; - The pipe walls in the upper plenum would be colder than in a reactor and too short to allow settling. The tellurium reaching the containment could be less than in a reactor due to chemical absorption; - in the containment the thermal inertial has a very important effect. The results obtained from phase B clarified a lot of important problems and limitations and led to a number of modifications of the circuit, such as the diameter of neutral tubes in the primary system, the size of the steam generator, the size and the characteristics of the containment. In particular, because it was found that the inventory ratio compared to a reactor is of the order of 5000 instead of 2000 (due to the limited length of the pins) it was decided to reduce the containment volume from 25 m3 to 10 m3. A large part of these results have been analysed in much detail (2), (3), (4). In spite of the limitations, which are due to technological constraints and also to the the lack of knowledge of some of the processes, the scoping calculations have shown that FP and aerosol phenomena which can be simulated in Phebus FP cover a wide range of parameters. Therefore an important data base will be obtained for code validation. In conclusion, the work performed in phases A and B in 1987 to 1989 was very important because:

CEC SUPPORT ACTIVITIES

163

1) it stimulated the interest and the participation of the European groups in the Phebus FP project; 2) it gave basic elements for the dimensioning of the experimental circuits and of the test matrix; 3) it allowed a first comparison of the codes and a clearer identification of needs and priorities in model and code development. Because, as mentioned above, only a limited number of scenarios have been analysed in phase A, the test matrix resulting from this procedure is still preliminary and incomplete. In order to obtain data on the main phenomena characterizing other accident scenarios a new set of calculations has been launched at the end of 1990 to study containment by-pass, steam generator tube rupture sequences in an NPP as well as the pool scrubbing phenomena (when a volume of water is present in one of the components) and a large break loss-of-coolant at cold leg sequence (which was not considered in phase A). These calculations will be performed with the US MELCOR code by the University of Madrid. 3. INSTRUMENTATION The success of a complicated experimental programme, like Phebus FP, depends to a large extent on the instrumentation and the analytical techniques which will be adopted. For this reason it was considered important, just from the beginning of the project, to undertake studies able to make available the best existing equipment and methods of measurement. The support provided through the SCAs may be divided into three parts: - review of the state of the art of on-line instrumentation and post test analytical techniques, directly applicable to Phebus FP; - assessment of advanced instrumentation potentially applicable to Phebus FP; - development and testing of selected instruments. The first part was performed by three organizations: - BATTELLE-Frankfurt to review thermohydraulic, hydrogen and aerosol instrumentation (5); - SIEMENS KWU to review fission product measurement instrumentation and post irradiation techniques as well as the behaviour and measurement of iodine species during and after the test (6); - AEA Winfrith to review analytical methods that will characterize the chemical forms of vapours and aerosols released from overheated fuel (7). The relevant instruments and methods, such as pressure and temperature transducers, liquid level and humidity sensors, flow velocity measuring methods, diverse hydrogen probes and aerosol mass concentration and size distribution measuring systems, have been analysed with a view to their specific measuring task and their location in the test facility. Each method and instrument has been described in detail under various aspects such as measuring principle, measuring range, technical design, evaluation model, calibration procedure, accuracy, experience, commercial availability, etc. Special attention has been paid to the behaviour of the instruments under the influence of radiation and recommendations have been made concerning the installation in the Phebus FP facility. A review was made of a wide range of analytical techniques developed to determine the elemental and chemical composition of vapours, liquids and aerosols. To study the chemical characteristics of an aerosol, no single technique can provide all the desired data and the most suitable approach involves a judicious combination of several methods. Post-test analyses can be wide ranging, and more sophisticated techniques should be used to determine the various reaction products, chemical interactions and deposition/attenuation mechanisms. The second part of the activity was assigned to BATTELLE Frankfurt (8). An extensive literature search served to identify the advanced measuring techniques to be proposed for Phebus FP. The identified methods include on-line determination of relative and absolute mass concentrations and particle sizes, on-line droplet size determination, local flow velocities, aerodynamic particle sizer and liquid level measurements. Emphasis has been put especially on the present state of radiation resistant optical fibre waveguides, which might be integrated in several measuring instruments(9). As a result of these studies, the third part of the activity was performed. The organizations involved in this research were: - Lavoro e Ambiente-Bologna; - AEA Harwell A high temperature sampling device with aerodynamic inertial separation of aerosols (INSPEC), developed by Prof. Prodi, was subjected to a feasibility study to assess the possibility to install it in the Phebus FP facility. The design of the unit was

164

THE PHEBUS FISSION PRODUCT PROJECT

upgraded to resist to height temperature (up to 700°C) and pressure (up to 35 bars) and a prototype was successfully tested in a furnace (10). The gamma-spectroscopic method of separating contributions to the gamma-ray yield from gas and deposits in pipes, already applied to differentiate between gamma-rays emitted by UF6 gas and deposits of uranium in a gas centrifuge, was transferred to the study of fission products released from irradiated fuel. This research demonstrated the capability to make real-time measurements on a test facility like Phebus FP (11). A study was performed on Maypack components to assess their effectiveness in the separation of inorganic from organic iodine species. A number of selective absorbers have been tested in various conditions of organic/inorganic ratios and temperature. It has been demonstrated that charcoal, used at ambient temperature, or silver loaded zeolite, used at any temperature up to 140°C, are adequate absorbers for organic iodine (methil iodide). 4. EXPERIMENTAL WORK Around the Phebus FP programme which foresees complex tests and integral in nature, a number of small scale and separate effects tests on aerosol physics and aerosol/FP chemistry have been launched to give support to model development/validation and give guidance in the definition/execution of the Phebus FP tests. From the beginning special emphasis has been put on the chemistry of fission products, aerosols and reactor structural materials during a severe accident, because a number of observations had shown that chemical reactions could drastically influence the Source Term to the environment depending on the production of more or less volatile or reactive products. Phebus FP dealing with real FPs is expected to give important information, however chemistry is one of the fields where the data and the modelling capabilities are still very limited. It was therefore decided to support two main actions: I) Acquisition of thermochemical data at high temperatures II) Execution of a large number of integral tests (complemented by separate effect experiments) to study the behaviour of multicomponents aerosols within the small-scale upper plenum, primary system and containment of a facility named FALCON (AEA Winfrith). If for a more accurate evaluation of Source Term the modelling of chemical behaviour of FP/aerosols is necessary, this asks for the knowledge of thermochemical data, in particular, the study of energy changes in chemical reactions, in order to predict the most stable chemical forms of FPs in specified accident conditions. For FP and their compounds at very high temperatures these data are in many cases not available. Therefore, first critical reviews of existing data, then experimental work to obtain thermochemical data using different types of sophisticated measurement techniques for selected gaseous species and condensed-phase compounds (specific FP compounds, hydroxides, mixed FP-bulk material compounds) relevant to LWR severe accident Source Term have been funded by JRC (through SCAs). A first priority group of about 40 compounds systems has been identified by a group of chemistry experts convened by the JRC in January 1990 (12). At present three laboratories in Europe, AEA-Winfrith, ECN-Petten and the Vrije Universitaet of Brussels are working on a harmonized critical review and measurement programme. A number of advanced and sophisticated measurement techniques including gas-phase, matrix isolation infrared spectroscopy, Knudsen cell mass spectrometry, calorimetry, differential thermal analysis and thermogravimetry are applied. A large number of priority thermochemical data will be available by the end of 1991 and will be used in the codes to perform more realistic evaluation of Source Term. A lot of effort was devoted from 1987 to the chemistry small scale integral studies performed in the FALCON facility at AEA Winfrith. The FALCON programme is presented in another session of the seminar, and a number of publications are already available (13), (14), (15), (16). One of the main objectives of the FALCON programme is to obtain experimental data for the assessment/development of mechanistic models for Source Term codes. At the same time, the programme is giving indispensable data for the preparation of Phebus FP (chemistry aspects, instrumentation, etc.). In FALCON simulant and trace-irradiated fuel samples can be heated up to 2000 K also in presence of typical bulk materials found within the core and additives (boric acid). The experiments can be performed under a wide range of thermohydraulic conditions, the containment chemistry can be modified including the sump pH and addition of painted surfaces. On-line and post-test analytical techniques can be applied. The facility is very flexible and makes it easy to evaluate the influence of different parameters.

CEC SUPPORT ACTIVITIES

165

A first series of FALCON tests focussed on FP and aerosol behaviour in the cpntainment was concluded in 1988 and demonstrated the influence of the aerosols originated by control rods, reactor structure materials and by boric acid on the chemical and physical form of the products reaching the containment. Later FALCON has been modified to study more realistically the processes occurring in the primary system and in the containment. The second series of the FALCON integral tests have indicated that the bulk components of the core (e.g. control rod materials, boric acid, zircalloy cladding and structural materials) can generate significant quantities of aerosol. The interaction of these aerosols with FP vapours (released from simulant and trace irradiated fuel samples) has been studied in experimental conditions which, to the extent possible, simulate the accident thermohydraulics and the oxygen potential. The study concentrated on aerosol distribution in the components of the facility and on the measurement of size, morphology and composition of a large number of individual aerosol particles generated during each test. A series of twenty tests has been performed and extensively documented. It is worthwhile reporting some of the preliminary conclusions: the aerosol transport properties can be substantially modified by the interaction between FP vapours and bulk material aerosols (e.g., reactions of CsI and CsHO with boric acid); the particles of aerosols generated during the tests can be considerably different in composition, size and morphology; iodine is preferentially retained on the painted internal surfaces and in the alkaline sump. As mentioned above, separate effect experiments have been performed to study more in detail specific aspects, such as aerosol nucleation, characterization of control rod aerosols, characterization of boric acid aerosols, formation and transport of mixed aerosols, vapour-aerosol reaction kinetics. All these experiments have been accompanied by theoretical studies to assess the validity and to improve the FALCON experimental tests. The experimental results have been also used to validate or improve some of the code models and represent a data base which will be used in the future to reduce the uncertainty in the prediction of source term. This programme gave also the possibility of testing or developing measurement techniques which are very important for the Phebus FP tests. Chemistry studies will probably be continued in the near future with the help of the SCAs which will be started at the end of this year. Another experimental programme, which is being considered to complement the Phebus FP experiments, is the STORM project, which has been proposed by ENEL-Milan to be executed in collaboration with JRC at Ispra. As indicated by the acronym, Simplified Testing Of Resuspension Mechanisms, the objective is to study in separate effect tests the mechanisms of resuspension of aerosols in the components of the primary circuit. Resuspension, which could be very important for accident management, is one of the phenomena on which the uncertainty is still very large due to the lack of data and modelling. It was shown in another experiment, namely the LACE programme, that under specific conditions of flow composition and themohydraulics, the deposition in the primary circuit of a reactor may amount to a large percent of the material released from the core region and become therefore available for a later resuspension. Since the mechanisms under consideration are only of aerosol physics nature and therefore do not require a representative chemical composition of fission products, it was considered inappropriate to include this study in a complex in-pile experiment like Phebus FP. The experimental facility is at present in the stage of design. A preliminary test matrix has been defined and the experimental phase is expected to start in 1993. JRC intends to invite other partners to participate in this programme. 5. MODEL AND CODE DEVELOPMENT The discrepancies found in Phebus FP scoping calculation results as well as the awareness of the need for model improvement and extension (particularly on the chemistry side) convinced the JRC to start a number of actions in this field. It was first decided to work on the improvement of existing models or computing codes with particular emphasis on the modelling of the chemistry of FP released from degraded core (IKE, Stuttgart) and the transport of FPs/aerosols in the primary system (ENEL-Milan, UPM-Spain). The objective of work performed by IKE was the improvement of the subsystem modules dealing with FP release in the German KESS system. Different aspects were considered: FP chemistry with both homogeneous equilibrium and reaction kinetics approach, inclusion of FP chemistry within the fuel rod during fuel degradation and in the coolant channel, effects of FP release on core thermohydraulics (17). The work related to the primary circuit included several aspects of the modelling. UPM made a state of the art report (18) on chemistry of Cs and I into vapour/hydrogen atmosphere, emphasizing the time dependent reaction kinetics which could be important for dilute mixtures when the reaction time is longer. A kinetic model has been developed and then applied to phase B Scoping calculations with the conclusion that instantaneous equilibrium assumption could be inaccurate. Another part of the work referred to particle resuspension (UPM and ENEL) which could enhance radionuclide release to the containment if

166

THE PHEBUS FISSION PRODUCT PROJECT

the flow passing through the primary system after the aerosol depletion is highly turbulent. This is a process not well understood and not modelled in the usual transport codes, even if a few theoretical models (force balance and energy balance) do exist (19). In addition, ENEL is studying the modelling of deposition under 2-phase flow and turbulent conditions, deposition on bends, elbows and pipework irregularities. A study on the behaviour of Te and its compounds in the reactor core and primary circuit was also performed at the UPM and the Mol Centre SCK (Belgium). UPM prepared first a state of the art report (20) on the behaviour of tellurium and its compounds in the RPV and RCS for LWR severe accidents reviewing the release and transport experiments and the models implemented in the transport codes (VICTORIA, RAFT, etc.). There is experimental evidence that Te strongly interacts with inconel 600 and SS pipe walls, but this phenomenon is in competition with Te/aerosol interaction. An improvement of Te chemistry model was introduced in the RAFT code. During 1989 the decision was taken to concentrate on a more advanced approach for the modelization of FPs/aerosols in the reactor core and primary circuit. Because the USNRC code VICTORIA became available to the JRC and EC organizations, it was therefore decided to work on development and validation of this code (contract with AEA-Winfrith) with the objective of including also the modelling work done by IKE, UPM, ENEL, etc. described above. The code VICTORIA calculates the behaviour of FPs during the release from fuel grains and transport through the fuel open porosity, their chemical interaction with other reactor materials, the transport through the reactor coolant system and the formation of aerosols (condensation) in the bulk coolant, the deposition on structure surfaces and potential re-entrainment. Also US laboratories sponsored by NRC, are working on VICTORIA. The development/validation activities, are from 1990 harmonized and decided in common between AEA, NRC and JRC. The AEA Winfrith activity was focussed in 1990 on the assessment of the quality of thermodynamic data base on the modelling of aerosol deposition and on a modelling approach for heterogeneous chemistry. It is planned to extend the control rod modelling, cladding behaviour, chemistry and numerical efficiency for multicomponent aerosols. As mentioned above, the long term objective of this JRC. programme is to make available to the Member States a European tool for severe accident calculations using mechanistic models, qualified data, efficient numerical techniques, extensively validated. It will be a long and difficult job for the JRC to give orientations and coordinate all the necessary work and to make the more appropriate choices. It was then decided to launch a SCA (CISI and IKE) for the design and construction of a modern informatic architecture allowing an easier coupling of modules coming from different sources (national laboratories or the JRC itself) or dealing with the different part of an accident scenario. This structure, called ESTER, which is described in detail in another presentation, would be the framework of future developments and could also allow the partners to combine and compare different modules to make safety analysis. A first pilot version of ESTER is expected to be available in 1991. At the end of 1990 additional contracts have been passed to improve modelling aspects and to adapt existing programmes to ESTER. One of the contracts, passed to CEA/IPSN, foresees the restructuring of JERICO, AEROSOL B2, IODE to constitute a fully integrated containment code which will then be adapted to ESTER with a data base coupling the three codes. Another activity which would have to result in an important contribution to the modelling is related to pool scrubbing. The final objective is to compare and assess the models already available in the existing computer codes SPARC (USNRC), BUSCA (SRD) and SUPRA (EPRI) and to implement a best estimated code. The code will be validated with the experimental data obtained by the Spanish facility of the LACE España Consortium and by the ENEA facility SPARTA as well as data available from the literature. 6. CONCLUSIONS The previous sections are focussed on the work that the JRC has launched and coordinated from 1987 in the Member States to give support to Phebus FP and to help in reaching the main objective of this programme, namely the development of a best estimate, validated package of computer codes for Source Term calculations. This initiative has already produced an enormous amount of qualified research work contributing also to a common approach in the complex problems of severe accident analysis. The JRC has also put in a lot of effort in promoting the participation of the most qualified experts of the EC Member States in all the groups dealing with the Phebus FP programme and in general with Source Term research. EC experts are regularly participating in the Technical Group and Scientific Analysis Working Group contributing to the solution of technical problems of the project and to the definition of the test objectives and execution of test precalculations. The success of this cooperation is indicated by the continuous increase in the number of European participants.

CEC SUPPORT ACTIVITIES

167

The Ad Hoc Working Group on Source Term, set up by CGC-5, is also organized by the JRC. This group is a forum where the progress and the needs in Source Term research are discussed in order to keep the Phebus FP programme, and all the related work, in close relation with what is performed elsewhere. References 1

2 3 4

5 6 7 8 9

10 11 12 13

14 15 16 17 18 19 20

() A.Markovina, P.Fasoli-Stella, A.Mailliat: Review of the Major Predicted Phenomena during FP Transport and Deposition on the RCS and Containment Building under Severe Accident Conditions ICHMT Seminar on FP Transport Process in Reactor Conditions Dubrovnik, 22–26 May 1989 () A.Markovina, A.Mailliat: Scoping Calculations in Support of the Phebus FP Experimental Programme Seminar on the CEC Contribution to Reactor Safety Research Varese, 20–24 November 1989 () I.Shepherd et al: Review of Scoping Calculations in Support of the Phebus FP Programme EUR Report in publication () A.V.Jones, E.Bonanni, A.Markovina: Principal Results of the Phase B Verification Studies in Support of the Phebus FP Project Aerosol Behaviour and Thermal Hydraulics in the Containment OECD-CSNI Workshop, Fontenay-aux-Roses, 26–28 November 1990 () G.Hampel, G.Poss, H.K.Fröhlich (Battelle-Frankfurt): General and Preliminary Thermohydraulic, Hydrogen and Aerosol Instrumentation Plan for the Phebus FP Project October 1989, EUR 12397 EN () E.Schuster, K.Nopitsch (Siemens): Radionuclides Measurements for the Phebus FP Project—Preliminary Study September 1989, EUR 12398 () B.R.Bowsher, A.L.Nichols (UKAEA, Winfrith): Review of Analytical Techniques to Determine the Chemical Forms of Vapours and Aerosols Released from Overheated Fuel December 1989, EUR 12399 EN () G.Hampel, G.Poss (Battelle-Frankfurt): Development of Advanced Instrumentation for the Phebus FP Project-Preliminary Studies October 1989, EUR 12396 EN () G.Hampel, G.Poss, B.R.Bowsher, A.L.Nichols: Advanced Instrumentation and Post-Irradiation Examination Concepts for the Analysis of Aerosols and Vapours in Source Term Experiments Seminar on the CEC Contribution to Reactor Safety Research Varese, 20–24 November 1989 () V.Prodi, F.Belosi (Lavoro e Ambiente, Bologna): In-Situ High Resolution Particle Sampling by Large Time Sequence Inertial Spectrometry September 1990, EUR 13236 EN () T.W.Packer, B.H.Armitage (AEA-Technology, Harwell): Application of Gamma-Ray Spectroscopy to the Differentiation between Mobile and Deposited Fission Products in Pipes August 1990, EUR 13215 EN () A.L.Nichols (AEA Technology, Winfrith): Fission Product Chemistry in Severe Nuclear Reactor Accidents Report of a Specialist Meeting at JRC-Ispra, 15–17 January 1990 September 1990, EUR 12989 EN () A.M.Beard, C.G.Benson, B.R.Bowsher, S.Dickinson, A.L.Nichols (AEA Technology, Winfrith), I.R.Beattie, P.J.Jones, J.S.Ogden, N.A.Young, (University of Southampton), E.R.Buckle (University of Sheffield): Fission Product and Aerosol Behaviour within the Containment April 1990, EUR 12844 EN () B.R.Bowsher: Fission Products and Aerosol Behaviour in the Containment Seminar on the CEC Contribution to Reactor Safety Research Varese, 20–24 November 1989 () C.G.Benson, B.R.Bowsher, A.L.Nichols: The FALCON Programme Multicomponent Aerosol Behaviour in the Primary Circuit and Containment OECD-CSNI Workshop, Fontenay-aux-Roses, 26–28 November 1990 () A.M.Beard, C.G.Benson, B.R.Bowsher, A.L.Nichols, F.Sabathier: Multicomponent Aerosol Behaviour AICHE Annual Meeting, 11–16 November 1990, Chicago, USA () K.D.Hocke, A.Paller, A.Schatz: Modelling and Code Development for the improved description of FP and aerosol release during LWR Core Heat-Up and Degradation EUR Report in publication () S.Fernandez, J.M.Huron, J.V.Lopez: Analysis of the Kinetic Behaviour of Iodine and Caesium Isotopes in the Primary Circuit of LWR Severe Damage Fuel Accidents EUR Report in publication () R.Bolado, E.Hontañón: Aerosol Resuspension in the Reactor Cooling System of LWR in Severe Accident Conditions EUR Report in publication () C.González: Modelling the Chemical Behaviour of Te Species in the Reactor Pressure Vessel and the Reactor Cooling System under Severe Accident Conditions EUR Report in publication

PHEBUS FP Organisation of the Project and International Collaboration A.TATTEGRAIN CEA IPSN/DRS P.von der HARDT JRC STI/IPTD CADARACHE

ABSTRACT PHEBUS FP developed from the initial French design study into a European project, with the signature of the CEA-CEC Contract in 1988, and further into an international programme by agreements with overseas partners during the past two years. The programme is supervised by a Steering Committee which reviews the technical-scientific options and the results. The executive body under the Committee, the Project Group, includes a CEA and CEC manager as well as three (CEA) project leaders for design & manufacture, experiment operation, and interpretation of test results. The Steering Committee can request expertises from the two working groups: - the Analytical Group (SAWG), elaborating test objectives, carrying out reactor calculations and test precalculations, - the Technical Group (TG), assessing the designs proposed and the results obtained by the Project Group. A third group looks into financial aspects of the CEA-CEC contract only. The two working groups, SAWG and TG, play an important role in the exchange of information and of expertise between all partners. Besides documents issued by the Project, annual seminars will be, in future, a more formal and efficient way of information transfer. The paper reviews the internal Project organisation and the collaboration network, through CEC/JRC inside the European Community and through CEA overseas. 1. INTRODUCTION The TMI-2 accident in 1979 and the Chernobyl accident in 1986 confirmed the importance of a sound data base on phenomena related to severe accident conditions in nuclear power plants. Many of the research programmes triggered or influenced by these two events have been carried out through international agreements. The reasons are obvious: . the high costs of both experimental facilities and computer code development encourage the sharing of resources, . many countries and organisations meet with identical problems of severe accident analysis, the quantification of their consequences, management for mitigation, design improvements,… . accident consequences are not limited by national or institutional boundaries. The French CEA at Cadarache had developed a tradition of collaboration in the scope of reactor safety related experiments in the CABRI/SCARABEE reactors (fast reactor fuel) and the PHEBUS reactor (thermal reactor fuel). Staff of the DRS (formerly, DERS) Department implemented a large design study for a fission product transport experiment, in 1986–88 /1/. The Commission of the European Communities (CEC) on the other hand had been operating research programmes in the area of severe accidents, both through programmes of the Joint Research Centre (JRC) at Ispra, Italy /2/ and through actions coordinated by Headquarters at Brussels, Belgium /e.g. 3/. The two parties therefore decided to co-sponsor the execution of the PHEBUS FP Programme and to share technical tasks and responsibilities /4/. The CEA-CEC Convention was signed in July 1988, covering the total duration of the Programme which had been estimated to 9 years. The two parties also agreed to invite EEC Member State organisations to actively contribute to PHEBUS FP, and to encourage international participation. Today, three years later, good progress can be reported in both areas, see enclosed figure 1.

PHEBUS FP

169

2. INTERNAL ORGANISATION 2.1. Scope CEA and CEC as major partners of the Programme defined, by means of the Convention of July 1988, the groundrules of their (“internal”) collaboration and also possible forms of integration of “external” partners. EEC Member State organisations obviously belong to the “internal” part. 2.2. Responsibilities CEA holds the technical and financial management responsibility for preparation and execution of the in-pile test programme, in accordance with decisions taken by the Steering Committee (see paragr. 2.3.). CEC participates in the programme execution by engineering staff seconded to Cadarache and by calculations carried out at JRC Ispra. CEC also manages supporting activities, either in its own laboratories or in those of EEC Member States /5/. 2.3. Structures The Programme is guided by the PHEBUS FP Committee (Steering Committee, SC), meeting twice a year. The SC operates in two configurations: a CEA-CEC Committee surveying essentially financial and organisational aspects of the Programme, and the full configuration including associated countries. The SC receives advice from three working groups; viz. 1) a financial group controlling the expenses of both partners according to the Convention and reporting to the CEA-CEC configuration of the SC, 2) an analytical group (SAWG) with the following tasks: -

elaboration of a test matrix, analysis of Phebus test pre-calculations and interpretation, definition of calculations to be performed, proposal of Shared Cost Actions of interest for test analysis, analysis of results obtained through Shared Cost Actions.

3) a technical group (TG) with the following tasks: - analysis of the design of the experimental equipment and the instrumentation to be used, - analysis of the experimental procedures proposed by the experimenters, - assessment of the test results obtained in terms of adequate performance of experimental equipment, instrumentation and experimental procedures. Chairmanship and secretariate of above-mentioned groups change each year between CEA and CEC. The executive body under the SC is the Project Group. It has the mandate to -

manage the Phebus FP Programme between SC meetings according to SC decisions, inform the SC and ask decisions for issues beyond day-to-day programme management, schedule the work of the different teams working for Phebus FP in close collaboration with their managers, keep all those working for FP informed, prepare the SC meetings, together with the working group chairman.

Details on the operation of groups and committees are shown in table 1.

170

THE PHEBUS FISSION PRODUCT PROJECT

3. EXTERNAL RELATIONS International participation, as said before, is encouraged in order to make the best expertise available to the Programme. This is achieved by reviewing the Phebus FP design papers on a “lessons learned” basis using the experience gained in in-pile and out-of-pile tests and by participating in Phebus FP test precalculations. Proposals concerning the test matrix, on the other hand, are accepted and discussed on group and committee level. Typical recent discussion points have been, for example: the non-condensing steam generator for the first test, instrumented fuel rods for the second test, the introduction of boric acid,…. Organisations from non-EC countries gain access to Phebus FP by contributing to the funding and to support activities of the programme. The level of contributions determine the type of partnership, i.e. participation or association. Associated organisations hold memberships in Steering Committee, Analytical Group and Technical Group, receiving all docupents processed by these groups. All participating organisations will be informed by annual seminars, planned as first results of each test become available. With the present schedule, the first of these seminars will be held in spring 1993. TABLE 1 Phebus FP Organisation Structures Committee, group

Membership Associates (per country)

Other participation

CEA

CEC CEA CEC

Associates (per country)

Steering Committee, “internal” configuration Steering Committee, “external” configuration Financial Group Technical Group

3

3

–

Programme managers Working group chairmen

–

2

3

3

1

Programme managers Working group chairmen (except Financial group)

Invited experts possible

2

2 2

2 2

– 1

– Experts

2 3 to 4

Analytical Group

3

3

1

Experts

3 to 4

Project Group

1

1

–

Programme managers Analytical group chairman Programme managers Experts Technical group chairman Programme managers Experts Task and project leaders

Typical annual meeting frequency

–

4. REFERENCES 1 2 3 4 5

. Ph. DELCHAMBRE and P.von der HARDT, “The PHEBUS FP Project”, EUR 12195 EN (1989) . W.KRISCHER, ed. “Reactor Safety Research. The CEC Contribution”, EUR 12343 EN (1990) . G.N.KELLY, M.OLAST and J.SINNAEVE, “The CEC Research Program on Methods for Assessing the Radiological Impact of Accidents (MARIA)”, Nucl. Technol., Vol. 94, (May 1991) 161–172 . J.BUSSAC and H.HOLTBECKER, “The Phebus Fission Products In-Pile Test Programme”, NUCSAFE 88, Avignon, France (October 1988) . P.FASOLI-STELLA and A.MARKOVINA, “CEC Support Activities: EC Shared Cost Actions and Others”, these proceedings.

GLOSSARY CEA CEC DRS EEC IPSN IPTD JRC

Commissariat à l’Energie Atomique Commission of the European Communities Département de Recherches en Sécurité European Economic Community Institut de Protection et de Sûreté Nucléaire In-Pile Test Division Joint Research Centre

PHEBUS FP

Fig. 1. PHEBUS FP Programme—Contractual Network (May 1991)

SAWG SC SCA STI TG TRPD

Scientific Analysis Working Group Steering Committee Shared Cost Action Safety Technology Institute Technical Group Thermodynamics and Radiation Physics Division

171

Discussion following the presentations of SESSION III Summary of the chairman Mrs. P.Fasoli-Stella

This session included many presentations, therefore the time for discussion was limited even if the papers stimulated many questions and comments. The first paper, presented by Mr. Gauvain, was an overview of the main phenomena to be studied to describe the behaviour of Fission Products and to define the Source Term in case of Severe Accidents. The comments from the audience dealt with a series of technical aspects such as the necessity to take into account the bulk material aerosols which can have an important influence on FP behaviour, the impact on FP release of the high burn-up of the fuel and, in particular, of the high pitched values which are in the interest of the utilities. The speaker shared the view that these aspects are very important even if, for instance, the effect of fuel structure cannot be studied in an in-pile experiment, but has to be tackled first analytically. Other comments were more general such as the need to know more about the probability of containment failure or by-pass in order to clarify better the importance of FP retention in the primary circuit. Another participant expressed doubts about the possibility of describing such a large number of physical and chemical mechanisms associated with aerosols/FP behaviour without very complex codes which risk not being easily understood and correctly run. The speaker expressed the view that very sophisticated codes could be needed to describe an accident sequence, while, provided that the mechanisms are well understood, global safety evaluations can probably be performed with lower precision. The second part of the session was entirely devoted to the different aspects of the Phebus-FP programme: the test matrix, the hardware, the instrumentation plan, the support programmes, the organization of international cooperation and of the project. Concerning the test matrix, presented by Mr. Markovina, a few questions came from the audience, such as how to study the effect of oxidation level on FP release and of the tellurium (Te) behaviour and whether it is the best strategy to have a control rod in all the tests. Concerning oxidation, the speaker confirmed that this parameter is very important because it affects the chemistry of the system and Te release is related to oxidation level. About the number of tests with control rod, a final decision has not yet been taken on all the tests, also because the test matrix will be influenced by the test results. A general comment referred to the fact that in in-pile tests like Phebus-FP it is foreseen to proceed from rod integrity to a likely debris bed stage. To get information about the interim stage of the fuel, small-scale integral experiments are needed. Several comments (not only in this section) were stimulated by the instrumentation plan presented by Mr. von der Hardt, in particular on oxygen potential and on high temperature measurements, on the need for on-line instrumentation and on the software to be developed for sophisticated measurements. The use of an instrumented fresh fuel rod in the bundle was again recommended by a participant. An integral test like Phebus-FP relies heavily on post-test analysis and on sampling. It is very important to have a strategy on sampling frequency or to have a sort of on-line monitoring system (e.g. photometer) to decide when to take a sample. The speaker recalled that guidance for sampling can rely not only on a photometer but also on gamma spectrometers and on-line aerosol monitors. In any case, instrumentation and sampling are limited by financial and space constraints. Concerning the French support programme presented by Mrs. Lecomte, questions were asked about the HEVA results (transferability to Phebus-FP) and, in general, on scaling criteria for HEVA and Phebus-FP, on the possibility of studying fuel swelling without melting in HEVA as well as on the objective of the TUBA facility. The speaker confirmed that these experiments are available to Phebus and are used for model validation and definition of correlations. With respect to the mini-containments and to the doubts expressed by a participant on the practical possibility of executing such complex tests during a Phebus-FP experiment, Mrs. Lecomte reaffirmed the interest of taking the opportunity for experiments under chemically realistic conditions. The presentation on the support actions, in particular, on scoping calculations by the CEC, made by Mrs. Fasoli-Stella gave the opportunity for additional comments on the shroud problems. It was discussed how to compensate the radial heat losses during the test and on the importance of high temperature instrumentation. In conclusion, this long session gave a fairly detailed view of the status of the Phebus-FP programme and of all the activities which are being executed in an international frame preparing and supporting the foreseen experiments.

SESSION IV ANALYTICAL ACTIVITIES

Survey of source term codes M.R.Hayns , AEA Technology, Harwell and S.R.Kinnersly , AEA Technology Winfrith ESCADRE and ICARE code systems M.Reocreux and J.Gauvain , CEA/IPSN Fontenay-aux-Roses ESTER—a European Source Term Evaluation System A.V.Jones and I.M.Shepherd , CEC/JRC Ispra FPTO test precalculations A.Mailliat , CEA/IPSN Cadarache, A.V.Jones and I.M.Shepherd , CEC/JRC Ispra Summary of discussion A.Tattegrain , CEA/IPSN Cadarache

SURVEY OF SOURCE TERM CODES M HAYNS1 and S R KINNERSLY2 AEA Technology Reactor Services 1Harwell Laboratory Didcot Oxon OX11 ORA UK

2Winfrith Technology Centre Dorchester Dorset DT2 8DH UK

SUMMARY The type, scope and role of current source term codes is reviewed. Issues and problems that are generic to most or all such codes are identified. These include numerical problems, fundamental physical and chemical databases, uncertainties in fuel degradation and thermal-hydraulics calculations, how to benchmark systems application codes against mechanistic codes and weaknesses in phenomenological modelling. It is concluded that further code development to address these issues is justified. This should be tightly coupled to experimental work within an internationally agreed framework. 1. INTRODUCTION A paper with this title could follow a number of different approaches. It could • Describe all current source term codes, their capabilities, strengths and weaknesses or • Review the history of the development of source term codes, explaining how we arrived at the situation today or • Assess the overall situation in source term modelling, identifying areas where codes are weak and where the challenges of the future lie. This paper adopts the final approach for the following reasons:• In this Phebus-FP seminar we are looking forward— history is interesting but not the reason we are here. • Catalogues of code capabilities are valuable but are usually long and detailed, are best studied in private in relaxing surroundings and are readily available from the owners of the code. The starting point is the following three observations and questions:(1) Source terms calculated in NUREG-1150 (date: 1990) were quantitatively consistent those in the Reactor Safety Study (RSS) WASH-1400 (date: 1975), even though many, many man years of source term research and code development had been expended in the years between the studies (Figure 1). QUESTION: Does that mean that source term code development should stop? (2) Source term experiments—most notably Phebus-FP-continue world-wide. QUESTION: Does that mean that source term code development must continue? (3) Source term codes are being used in reactor safety assessments. QUESTION: Does that mean that source term codes are good enough already? These questions prompt consideration of:(a) the type of source term codes available, their scope and role; (b) issues or problems that are generic to most or all current source term codes; (c) the code validation and development process. Finally, some conclusions will be drawn regarding future work on source term codes.

SURVEY OF SOURCE TERM CODES

175

Figure 1: Representative Comparison of NUREG-1150 and WASH-1400 Source Terms

2. SOURCE TERM CODES—TYPE, SCOPE AND ROLE The TMI-2 accident showed that a severe accident in an LWR does not necessarily give rise to a large source term. To make full use of this fact in safety assessments and thus to avoid excessive conservatism it is necessary to have:(a) practical tools for wide ranging safety assessments, (b) scientific understanding to underpin the use of those tools. The main tool developed for safety assessments is Probabilistic Safety Analysis. This requires many source term estimates. High accuracy is not generally needed as conservatisms or substantial uncertainties are acceptable provided they are incorporated properly into the PSA and the desired level of safety is demonstrated. Thus, for PSA studies, source term code accuracy can be traded off against other desirable attributes such as speed, transparency of modelling, user controlled

176

THE PHEBUS FISSION PRODUCT PROJECT

Figure 2: System Application Codes—Typical Attributes Attribute Fast running Good user interface Simple models Not state of the art science User input parameters From initator to consequences Modelling experiments is difficult

Comment Applications tool, many calculations Minimum needed to do the job -

for sensitivity, scoping, or conservative calculations Focus on source term Make best use of simplifications and approximations appropriate to plant scale

Figure 3: Mechanistic Codes—Typical Attributes Attribute Slow running Difficult user interface Detailed models State of the art science Hard wired, ‘best estimate’ parameters Restricted to part of accident sequence Modelling experiments is (fairly) easy

Comment Research tool, few calculations Best that can be done -

Calculations cannot be tuned to experiment Focus on phenomenology Make direct link with experiments

parametric models and the ability to follow an accident through from initiation to consequences in a single calculation. Computer codes in this category will be referred to here as systems application codes. A brief list of typical attributes is given in Figure 2. The quest (and need) for scientific understanding has lead to the development of a second category of source term codes. Here, the objective is accuracy to be achieved primarily through incorporating all significant phenomena or processes and using state of the art models. This means that speed and scope are usually traded off against completeness of modelling and incorporation of the most up to date scientific understanding. In general, such codes only model one part of a plant or accident sequence (eg fission product release from fuel and transport through the primary circuit). Computer codes in this category will be referred to here as mechanistic codes. Figure 3 lists briefly the typical attributes of such codes. The two classes of codes each have distinct and complementary roles. Figure 4 gives in diagrammatic from the main roles of each code. Note in particular:(a) the bridging role played by the mechanistic codes between experimental data and systems application codes. (b) the experiment design role of mechanistic codes. These roles follow on from the compromises made to achieve speed and applications flexibility for systems applications codes. In general, direct assessment against experiment is difficult or of doubtful value. Firstly, the code is usually tailored to plant configurations and cannot adequately represent experimental boundary conditions. Secondly, simplified models (typically including user specified parameters) do not readily allow the non-prototypic and small scale results of experiments (including Phebus-FP) to be extrapolated to plant scale and conditions. In principle, mechanistic codes can do these things. The last sentence started ‘in principle’ because in practice there are restrictions on what can be achieved. Some technical issues will be considered in the next section. It is worth noting here, however, that calculations of source terms require calculations of fuel degradation and thermal hydraulics processes as well as fission product release and transport. Thus mechanistic fuel degradation and thermal-hydraulics models are needed to bridge the gap from experiment to system applications code. In the containment, mechanistic codes (eg CONTAIN) are being developed that couple thermal-hydraulics and fission product (at least aerosol) behaviour. This is being driven by the (sometimes) tight coupling between thermal-hydraulics and condensation onto aerosols. The development of thermalhydraulics models adequate to build the bridge from fission product experiments to systems analysis code is therefore a well established activity. For the primary circuit, however, fission product transport and fuel degradation/thermal-hydraulics are not so tightly coupled and separate mechanistic codes have been developed. Development of fuel degradation/thermalhydraulics codes has tended to focus on areas that affect primary circuit or containment survivability (eg hydrogen-generation)—although there are still large uncertainties in late phase melt progression modelling. While containment survivability is, of course, a (if not the)

SURVEY OF SOURCE TERM CODES

177

Figure 4: Roles of Systems Application and Mechanistic Source Term Codes

major source term issue, consideration of what is needed from a fuel degradation/thermal-hydraulics code to bridge the gap between fission product experiments and system applications code is not always high on the development and assessment agenda. 3. SOURCE TERM CODES—GENERIC ISSUES Other papers at this seminar have reviewed the state of the art in our understanding of source term phenomenology and have considered priorities from the view point of reactor safety. This section addresses the question of which issues common to most or all source term codes need to be addressed in the future. Time and space are too limited to cover topics specific to individual codes (lists are readily available in most cases, though, and much in evidence when the time comes for applying for funding!) The first issue is simply whether the codes are numerically sound: Do answers depend sensitively on timesteps? Do they vary unrealistically (and perhaps randomly) when small changes are made to input parameters? Do results converge with decreasing timestep and finer nodalisation? Do results vary from computer to computer? etc etc. The many and varied models (usually nonlinear and many containing ‘switches’ or abrupt changes) in source term codes make it difficult to demonstrate satisfactory numerical behaviour by pure analysis. Code verification (demonstrating that the coding is consistent with the specification) is therefore not proof against problems. Difficulties often emerge by chance during applications. Both mechanistic and system applications codes are affected. In recent months the author has discussed or heard presentations of numerical problems with: VICTORIA (mechanistic, under development) CONTAIN (mechanistic, mature) MELCOR (systems application, under development) MAAP (systems application, mature)

178

THE PHEBUS FISSION PRODUCT PROJECT

All possibilities (mechanistic/systems application, mature/under development) are covered. There is no reason to believe that other source term codes do not have similar problems. They are expensive to put right and cast doubt on individual results produced by the codes. For mature codes, the best that can be done is probably careful use of sensitivity calculations in applications, reporting of all problems to the user community (via the code support team, please!) and back fit repairs to the code when justifiable. With codes under development, however, it is important that rigorous testing should be carried out to identify and eliminate problems. It takes time and money but ultimately is cheaper than wasting time and effort with an untrustworthy code. (A word about chaos: strange sensitivities might result from mathematical chaos implicit in the physical phenomena themselves, but the odds must surely be in favour of poor coding or modelling.) The second generic issue concerns the fundamental physical and chemical data used by the codes. Water properties are, for all practical purposes, the same in different thermal-hydraulics codes. Why, then, should thermodynamic data and physical properties vary between source codes? The practical answer is, of course, that the world stock of data is being improved and extended while codes have to use whatever is available when they are developed. However, it is reasonable to ask whether an aim for the future should be internationally agreed, assessed databases with agreed functional representations which should be available by default in source term codes. That is not to preclude having other data available as options. It would, however: (a) add credibility to reactor licensing applications (pity the poor regulator who has to ask whether every thermodynamic function in the code is acceptable when there is no international standard), (b) ensure that code comparisons really are what they claim to be (not investigations into differences in fundamental databases), (c) help codes keep up to date with latest data (provided they are written to accept easy input of international standards). Of course, data for high temperature chemical species will always be fewer and less accurate than for water so uncertainties in the database will always be much larger than for water properties. Nevertheless, an international standard database for source term codes seems, to the authors, to be a worthwhile aim. The third generic issue is aimed at fuel degradation and thermal hydraulics modellers. Is degraded fuel and thermal hydraulics research and development sufficiently well focussed on areas that cause major uncertainties in fission product release and transport? Source term calculations involve both fuel/thermal hydraulics and fission product behaviour. Developments in each area should be in step. For instance, models for quenching degraded fuel are being developed. Total hydrogen production is an important indicator of a successful model but hydrogen is predominantly concerned with containment threats. For fission product release during quench (which could be substantial) the fuel temperature transient is likely to be most important. How accurate do temperatures need to be ie: when is a quench model good enough for fission product release calculations? As an example of the value of asking such questions, scoping calculations carried out a few years ago by AEA Technology showed that a melt temperature accurate to 50°K is needed to calculate certain fission product releases from melt-concrete interaction to within a factor of 2. This both helped to set a target for melt-concrete thermal hydraulics modelling and cautions against development of highly accurate fission product release models which are not justified by the accuracy available from temperature predictions. The next issue is one of the methodology: how should a systems application code be benchmarked against mechanistic codes? This may seem a strange issue to raise. After all, both types of code have existed for several years now and the ‘two tier’ code strategy has a lengthy and honourable history. Nevertheless, as far as the author is aware, a thorough and complete benchmark exercise does not yet exist for any systems application source term code. Instead, there is an increasing tendency to try to assess systems application codes directly against experiments. This is no doubt due in part to the fact that current generation mechanistic codes are unable at present to carry out on a routine basis source term calculations from accident initiation to release to environment. Thus, benchmarking is usually confined to comparison of results for specific processes. Furthermore, although sensitivity studies are an important part of many applications of systems application codes, the inflexibility of most mechanistic codes makes it very difficult to benchmark the sensitivity behaviour. In principle, a recommended set of input parameters for a systems application code might give results that adequately reproduce a single calculation from an inflexible mechanistic code but the sensitivity calculation designed to show that the results are not near a cliff-edge may be grossly in error. Perhaps more flexibility in mechanistic codes is needed. With source term codes now being used in submissions to safety authorities, regulators are no doubt considering the question of the adequacy of benchmarking as it is currently done. Developers of source term codes— particularly mechanistic codes—need to address the issue, too. The final issue concerns phenomenological modelling. Where are source term codes generically weak because adequate models have not been developed (whether because of the lack of suitable experiments or otherwise)? Adequacy depends on your point of view, of course (eg. funder or researcher). However, some important areas are noticeably weak and there is at least a prima-facae case for continuing model development and, where necessary, experiments. Important areas of fission product modelling are listed in Figure 5. Such areas of weakness do not, of course, invalidate current applications of source

SURVEY OF SOURCE TERM CODES

179

Figure 5: Weak Areas in Fission Product Modelling In-vessel/primary circuit Vapour/aerosol chemical reactions Effect of liquid water in the RCS Optimum set of chemical species Aerosol particle shape factor prediction Aerosol deposition at high fluid Reynolds number Aerosol resuspension, including bulk resuspension Chemical reaction with and revaporisation from surfaces Containment Inhomogeneous atmosphere (plumes, stratification etc) modelling using ‘lumped parameter’ codes Condensation onto aerosols (including fast numerical scheme) Non-ideal solution modelling for fission product release from Core-concrete melts Aerosol particle shape factor prediction Decontamination due to water pool on molten core-concrete (churn-turbulent flow regime) Mechanical generation of aerosols from high pressure melt ejection and fuel coolant interaction

term codes. They do, however, impact on the scope and uncertainties of calculations. It is the need for increasing the scope and reducing uncertainties that should determine priorities for source term code model developments. 4. SOURCE TERM CODE VALIDATION AND DEVELOPMENT There is a phrase in English ‘It’s as long as a piece of string’. It means that the scope or duration of ‘it’ is not fixed and can be as big or as small as you choose. The phrase applies perfectly to source term code development. However, just as the length of a piece of string must be enough (but not too much) for its intended use, so must source term code development continue until the codes are fit (but not excessively so) for purpose. The latter requires two things: (i) specification of ‘fit for purpose’, (ii) measuring sticks to establish whether ‘fitness for purpose’ has been achieved. These are properly technical matters, although financial criteria are increasingly being used. The technical activity that is a key to both is code validation. Unfortunately, code validation, for all its common usage, is an ambiguous and ill-understood term. To some people, validation means proving that a code is accurate and improving the code until it is so. Others consider that the uncertainties in source term calculations are so great that the codes cannot be validated, only assessed (and so they do code validation under a different name). Since the question of when to stop developing source term codes (and why) is important it is equally important to be clear about code validation and its role in the code development process. There is ample evidence that this is currently not the case. This section attempts to clarify the situation and propose a framework or methodology which could be used to structure source term development and validation. To clear the ground for the proposed framework, some thought must be give to what the real purpose of source term codes is. Source terms can be estimated with simple methods, experimental data and expert judgement. There will be substantial uncertainties. The purpose of source term codes is not to provide an alternative accurate calculation of the source term (leaving philosophical considerations aside, that is a practical impossibility). Rather, it is to help reduce the judgemental uncertainty ranges in source term estimates to an acceptable level. If uncertainties obtained without using source term codes are both acceptable in range and credible (ie they are derived convincingly from physical evidence) then there is no need for code calculations. Otherwise, some evidence must be found to reduce or give credibility to the uncertainties. This almost always means calculations. Given the complexity of phenomena occurring during severe accidents, use of a source term code is almost obligatory. This view of the end use of source term codes has three important implications for their development and validation. Firstly, applications (or plant calculations) must be an integral part of the code development and validation process. Secondly, code validation is seen to be the process of providing evidence and reasoned arguments derived from the comparison of code calculations and experimental data to support the reduction (or otherwise) and justification of source term uncertainties through the use of source term codes. Thirdly, the target application must be well defined - one should really talk about validating a code for a specific application. Because fission product science is an active area of research, it is also necessary to consider how this activity affects the process of source term code development and validation. This is important because research calls for continuous code

180

THE PHEBUS FISSION PRODUCT PROJECT

development to keep abreast of latest results and to guide future research whereas application to, for instance, plant safety assessment calls for well validated codes stable over many years. The conflict can be resolved by developing and validating source codes in an interative framework closely integrated with experiments. Code validation must appear at the end of each iteration both to establish the capabilities of the code in applications and (in consequence) to guide future research and development. In fact, because a source term code has to include models for many processes, there is actually a need for two sets of iterations. One covers the development of individual models and involves separate effects experiments and the acquisition of basic data. An outer iteration considers the development of the source term code as a whole and involves integral experiments and applications. It is this outer iteration (which necessarily takes longer than the inner ones) that produces a code with an established capability that is stable over at least the lifetime of the next iteration (typically a year or more). Code validation comes at the end of an outer iteration and determines whether the code is fit for purpose (ie. whether development should stop). This iterative framework is summarised in Figures 6 and 7. It is not a revolutionary idea, some aspects are already evident in various source term code developments around the world. It is also idealised, taking no account, for instance, of the need to keep experimental teams employed during the periods of code validation at the end of each outer iteration. Some compromises are inevitably needed in practice. However, with funding shrinking and international cooperation increasing some sort of agreed overall framework for source term code development and validation is desirable. It does not exist at the moment. This proposal might be a starting point. Funding bodies should note that it provides a rational basis for stopping source term code development. Code developers should note that it provides a rational basis for continuing code development. 5. QUESTIONS, ANSWERS AND CONCLUSIONS Three questions were asked in the introduction. Answers will now be attempted before conclusions are drawn. Firstly, in the light of NUREG-1150 AND WASH-1400 should source term code development be stopped? Not surprisingly, the answer (for the authors at least) is no. The uncertainties in NUREG-1150 were determined by a complex process including a great deal of expert judgement. Current generation source term codes did not contribute much to the determination of source term and risk. This was at least partly due to the limited capabilities and running costs (in computing time and manpower) of the mechanistic codes which might have provided benchmark results for this benchmark risk assessment. An important implication is that mechanistic source term codes in particular should continue to be developed until they can reliably be used for such plant calculations. It should then be possible to put some of the expert judgement to the test. Apart from that, the NUREG-1150/WASH-1400 comparison is not evidence for or against source term code development. Secondly, since source term experiments continue across the world, must source term code development continue? The answer is a qualified yes. Yes, because some of these experiments are addressing areas of generally acknowledged weakness in the codes. Qualified, because some experiments, particularly integral experiments such as Phebus-FP and Falcon, should perhaps be best seen as providing data for integral code validation rather than code development. Experiments are a necessary input to source term code development either for building basic models or for assessing ‘fitness for purpose’. Their mere existence, however, is not sufficient justification to develop codes. (Note: there is one exception here, namely when code development is needed to assist in experiment design and specification). Thirdly, are source term codes good enough already if they are already being used successfully in reactor safety assessments? The answer is no. While details must depend on specific codes and applications, previous sections have highlighted generic weaknesses in code stability and robustness, important areas of modelling and the capability to benchmark adequately systems application codes by using mechanistic codes. The issue is really about the scope of current systems application codes and having confidence in their results and use. The codes are certainly good enough within (code dependent) limits and provided their results are used with due caution. Increasing consideration of severe accidents (including managed accidents) in reactor licensing and safety assessments and the development of advanced reactor designs makes it highly likely that the scope of source term codes will have to be widened and confidence in their results and application improved. Finally, some conclusions must be drawn. (a) Source term codes are available and being used for reactor safety assessments. (b) There is in principle a two-tier structure of systems applications codes and mechanistic codes but this has not yet been convincingly used in practice. (c) Weaknesses in fission product modelling, numerics and coding limit the scope of source term codes, reduce confidence in results and prevent complete and thorough benchmarking of systems application codes by mechanistic codes.

SURVEY OF SOURCE TERM CODES

181

Fig. 6. Code development strategy —outer iteration

Fig. 7. Code development strategy —example of inner iteration

(d) Weaknesses in degraded fuel and thermal hydraulics modelling dominate uncertainties in some areas of source term calculation. (e) Code development is justified, but (i) must be tightly coupled to experimental work with feedback both ways, (ii) priorities between fission product and degraded fuel/thermal hydraulics modelling must be properly assigned, (iii) must address robustness, numerical stability and other ‘usability’ issues as high priority,

182

THE PHEBUS FISSION PRODUCT PROJECT

(f) Increasing international cooperation in source term code development and assessment argues strongly for: (i) international standard databases for properly assessed fundamental physical and chemical properties, (ii) a common framework and strategy linking source term code development, experiments and code validation. Much has been done. There is much to do.

ESCADRE and ICARE code systems M.REOCREUX*, J.GAUVAIN** Institut de Protection et de Sûreté Nucléaire * Département de Recherches en Sécurité ** Département de Protection de l’Environnement et des Installations Centre d’Etudes Nucléaires de Fontenay-aux-Roses

SUMMARY The french severe accident code development program is following two parallel approaches: the first one is dealing with “integral codes” which are designed for giving immediate engineer answers, the second one is following a more mechanistic way in order to have the capability of detailed analysis of experiments, in order to get a better understanding of the scaling problem and reach a better confidence in plant calculations. In the first approach a complete system has been developed and is being used for practical cases: this is the ESCADRE system. In the second approach, a set of codes dealing first with primary circuit is being developed: a mechanistic core degradation code, ICARE, has been issued and is being coupled with the advanced thermalhydraulic code CATHARE. Fission product codes have been also coupled to CATHARE. The “integral” ESCADRE system and the mechanistic ICARE and associated codes are described. Their main characteristics are reviewed and the status of their development and assessment given. Future studies are finally discussed. 1. INTRODUCTION In the past years, both risk studies and TMI2 accident analysis have confirmed that severe accident prediction on Nuclear Power Plant (NPP) was a major significant safety issue. Such prediction for NPP is a considerable task. In a similar way than in the other accident areas, the first attempts have been concentrated on experimental simulations. More or less global and analytical experiments have been run which gave: - for the most analytical ones, information on basic physical mechanisms - for the most global ones, information on which qualitative phenomena can occur in a plant and how these phenomena are interacting each other. Deriving the NPP behaviour from such experimental information requires clearly the use of analytical tools. These tools play then a key role as linking experimental programs and plant evaluations. The development of these tools is the next step of all research program. French severe accident code development has followed this research plan and like many research plans in the field it has been realized on the basis of a two levels approach: - the first level consists in an integral code system named ESCADRE which characteristic is to give an immediate engineer answer to questions raised by safety studies. This is a complete system where all phenomena occurring during the accident in any place of the plant are predicted or at least evaluated when actual physical knowledge is not sufficient. The objectives of such a system are: . the assessment of the radioactive releases for validating emergency plans . the analysis of the efficiency of procedures and corrective actions . risk surveys for PRA program - the second level consists of mechanistic codes which till now have been focused on primary circuit behaviour (ICARE and associated codes). For these codes, the main characteristic is to reach a physical description sufficiently detailed for having a better physical understanding and by this way a better scaling capability from experiments to NPP. The aim is then to

184

THE PHEBUS FISSION PRODUCT PROJECT

Fig 1: ESCADRE general flowchart

improve the level 1 codes for which some engineering models are giving only partial answer to the basic and critical question of scaling, leading to evaluations which are difficult either to assess or to narrow. The objectives of such codes are then: . to provide physically based modelling of all important phenomenological events occurring in a plant during a severe accident, . to allow therefore, first the design and then the detailed physical analysis of separate effect experiments and system experiments (such as PHEBUS FP), . to provide assessed calculation tools for plant prediction and in consequence to provide the mean for scaling up from small scale experiments to scale 1 plant. This paper will describe, for each of these two systems, their main characteristics, the status of their development and assessment, and the studies planned in the future. 2. ESCADRE SYSTEM 2.1. Overall system structure ESCADRE system (1) (2) consists in several individual computer codes, describing specific physical phenomena in separate parts of the plant and sometimes during specific periods of the transient. ESCADRE links these various codes to provide an analytical tool under the form of a semi-integrated assembly of modules. The coupling of the codes which is a sequential coupling, is generally obtained through data files processing. The complete set of codes and their sequence are given on Figure 1. For the Reactor Coolant System (RCS) the beginning of the accident sequence up to the beginning of the core uncovery is predicted by the thermalhydraulic code CATHARE. The following part of the transient is treated for the RCS by the VULCAIN code up to the attack of the core support plate:it includes prediction of the thermalhydraulic, of the core degradation and of the Fission Product release. The initial F.P. core inventory is given by the PEPIN/IN VCO modules and after the release from the core, the transport in the RCS is predicted by the SOPHIE code (for the vapors) and the AEROSOLS/CIRCUIT (for the aerosols). For the containment, the thermalhydraulic behaviour is predicted by the JERICHO code, the core concrete interaction by the WECHSL code , the aerosols and iodine behaviour by the AEROSOLS/B2 and IODE codes respectively.

SCADRE AND ICARE CODE SYSTEMS

185

2.2 Main codes characteristics 2.2.1 VULCAIN VULCAIN (3) (4) is used as the thermalhydraulic driver for the primary circuit part of ESCADRE. It performs the calculation of the thermalhydraulic for the whole circuit (fluid dynamics and thermal behaviour of fluid and structures). The core degradation is also calculated including the thermal, mechanical and chemical aspects and the resulting material relocation. The fission products release from the fuel is finally deduced from the temperature history of the rods. Primary circuit modelling In standard, VULCAIN represents the RCS by two loops (broken loop and gathered unbroken loops). Each loop is represented by 7 or 8 volumes which distribution is depending of the transient. In the vessel an upper volume and a downcomer volume are connecting the two loops. The core is modelled in cylindrical geometry with a 2D noding. The pressuriser relief line is represented up to the relief tank by 10 control volumes. The cooling system have been recently added to the code capabilities. Thermalhydraulic models The main characteristics of the thermalhydraulic modelling are: - In the core, the calculation starts when a “water level” appears. The initial conditions are given by the CATHARE code. - The swell water level is determined by WILSON or CUNNINGHAM-YEH correlation. - Above the swell level and before the core collapse, an axial multi- channel steam hydrogen thermalhydraulics is considered. The 2D effects are taken into account by cross-flows between channels (2D noding) assuming the instantaneous redistribution of the flow rates in order to reach isobaricity in any horizontal core cross sections. The changes in hydraulic areas due to the thermal and mechanical deformations are calculated and introduced in the cross flows calculation. - In the loops, only steam hydrogen flows are considered. Steam condensation in the steam generators is calculated with direct reinjection of the condensed water into the downcomer or into the core. - The system pressure which is assumed uniform is calculated by means of an overall mass balance in the primary circuit. Thermal models The thermal behaviour of the rods and of the internal structures is calculated as follows: - For the fuel rods, 3 layers are considered i.e. UO2, gap, zircaloy cladding. Energy balance equations are solved for UO2 and cladding layers (one node per layer). In the gap, convective and radiative heat flux are calculated by classical models when the gap is open. A contact resistance is used when the gap is closed. - The control rods are modelled by an average material (absorber, stainless steel and zircaloy) for which a thermal balance equation is written. - For the internal structures, each component in the lower part of the vessel and in the upper part of the vessel is represented by one node with one average temperature. The thermal balance equation is solved using this average temperature and a convective heat transfer to the fluid. For the upper part components, an additional radiative heat flux from structure to structure is taken into account. - In the core region (core barrel, thermal shield, vessel walls), a thermal conduction calculation is performed for each component. - A lumped parameter approach is used when melting and material relo- cation occur. - The convective heat transfer to gas mixtures is calculated using classical heat transfer correlations (turbulent, laminar, forced and natural convection regimes). - For the radiative heat transfer, smoothing functions based on the temperature profiles are used for the transfers from ring to ring in the core lattice. Radiative exchanges are calculated between the external ring, the core barrel, the vessel walls and between the upper part of the core and the upper plenum. - Radiative flux to steam flow is taken into account in the energy balance.

186

THE PHEBUS FISSION PRODUCT PROJECT

- For the heat generation, the decrease of the average level of decay residual power is calculated by the code. The axial and radial profiles are given as input. When zircaloy cladding is oxidized by steam, the corresponding heat generation is added. Core degradation - The zircaloy cladding oxidation by steam is calculated by models which can be selected by the user (Cathcart, Baker-just, Urbanic-Heidrick). - The mechanical behaviour of the cladding (thermal expansion, tensile deformation, creep) is calculated using models derived from EDGAR experiments. The control rods mechanical rupture is based on empirical criterion deduced from UKAEA studies. - The loss of geometry model is based on simple criteria for material melting (melting temperature), relocation (refreezing temperature) and slumping (decladmg of the rod). These criteria are applied in every node for each material (UO2, Zr, ZrO2,…) separately. Core blockage due to meltdown, crust refreezing in the lower parts of the core are then evaluated. Fission product The fission product release rates versus temperature are evaluated by semi-empirical laws derived from CORSOR and assessed on HEVA program (5). 2.2.2 SOPHIE/AEROSOLS The behaviour of fission products after their release from the core is calculated by SOPHIE and AEROSOLS codes. SOPHIE is dealing with F.P. as vapors when the carrying gas is at temperatures higher than vapor/aerosol transition temperature. It describes the vapor transport in the circuit and the deposition on the structures. AEROSOLS codes are describing the transport and the deposition of aerosols either in the circuit (AEROSOLS-CIRCUIT) or in the containment (AEROSOLS/B2). For both of these codes, the thermalhydraulic conditions are given by input files. They are calculated for the RCS by VULCAIN, for the containment by JERICHO. The nodalization for these codes is based on control volumes which must be compatible (not necessarily identical) with the nodalization of the thermalhydraulic codes. Regarding F.P., each control volume can transfer with two structures (SOPHIE) or one structure (AEROSOLS).

SOPHIE A scheme of the phenomena modelled by SOPHIE (6) is given on Figure 2. The concentration of a vapor species in a control volume is obtained by the balance of the mass flux with adjacent control volumes (expressed as proportional to the flowrates) and the fluxes with different layers related to the structure. These fluxes are the following: - evaporation/condensation in the boundary layer: the diffusion of the vapor is calculated by using the analogy theory between mass and heat transfer; this flux is then proportional to the concentration difference between the bulk of the flow and the boundary layer. The condensation and the evaporation fluxes are taken proportional to the difference between actual vapor concentration and saturation concentration corresponding to the wall temperature. - adsorption/desorption in the first layer (deposit 1): the adsorption is taken proportional to the concentration in the boundary layer, the deposition proportional to the concentration in the first layer. - the exchanges with a second layer (deposit 2) where chemical interaction can occur with the structure material. The fluxes are proportional to the concentration of the layer from which they are issued. The formulation of SOPHIE code is the same for all the chemical species, the code has then to be run as many times as the number of species which are of interest. For each species a set of adsorption coefficients (reference adsorption velocity, activation energy) are being derived from CEA experimental data.

SCADRE AND ICARE CODE SYSTEMS

187

Figure 2: FIowchart of phenomena described by SOPHIE

AEROSOLS The AEROSOLS codes are calculating the distribution of sizes in discrete classes (up to 100) of a two components population of aerosols in the different control volumes (7). For a given class of size x, the evolution of concentration is obtained by a balance between: -

the sources of aerosols of size x in the control volume, the net contribution to class x by agglomeration, the net flux of size x aerosols by transfers with neighbouring control volumes or by leakages, the deposition of size x aerosols, the net contribution to class x by aerosols growth due to condensation.

The agglomeration model is based on the collisions physics. The physical phenomena which are acting are differential sedimentation, brownian and turbulent agitation. For the deposition, the phenomena which are taken into account are sedimentation, thermophoresis, turbulent diffusion, brownian diffusion, centrifugal impaction, diffusiophoresis. The aerosols growth related to condensation is described by the MASON equation. The differences between AEROSOLS-CIRCUIT and AEROSOLS/B2 consist mainly in the choices of the mechanisms which are considered to occur in each cases (i.e. RCS or containment). 2.2.3. JERICHO JERICHO (8) is the thermalhydraulic code for the containment. In the ESCADRE design, it is considered as the central part of the system. Its inputs are mass and energy flow coming from the primary circuit (VULCAIN) and from corium concrete interaction (WECHSL). Its outputs are the containment thermalhydraulic conditions for aerosols and iodine calculations (AEROSOLS/B2 and IODE codes). Containment modelling JERICHO uses a multicompartment description of the containment (up to 20 volumes). 3 types of structures are distinguished: internal structures inside a compartment, structures separating two compartments, external structures. Thermalhydraulic models In each volume a bipunctual representation is used with a liquid subvolume (sump water) and a gaseous subvolume (containment atmosphere). The gas phase is constituted of up to 6 components which are steam and 5 noncondensibles (N2, O2, H2, CO, CO2). - Pressure is assumed uniform inside a volume. - Thermodynamic nonequilibrium between the liquid phase and the gaseous phase is resulting from the mass and energy transfers between the sump and atmosphere: in the case of atmosphere humidity greater than 1, bulk condensation is calculated by a simple relaxation model; in the case of superheated liquid, same type of model is used for bulk

188

THE PHEBUS FISSION PRODUCT PROJECT

vaporisation. In case of humidity lower than 1, the evaporation from the sump at the interface is calculated by two models depending wether water is subcooled or superheated. - Flows between compartments and leak flows are calculated by Barré de Saint Venant or Bernoulli relations. - Inlet flows (mass and energy) are given by tables for water and noncondensibles. Models of partition of the incoming water between the liquid and the gaseous phase of the JERICHO volume can be chosen between 5 options. - Combustion of H2 and CO can be described by different optionnal models: continuous combustion, deflagration, virtual deflagration (i.e. without energy contribution of combustion). Thermal models - 1D conduction model is used for structure. - Heat transfer between the atmosphere and the structures is calculated: . either using convection heat transfer, . or using a film condensation model with 4 classical options: TAGAMI-UCHIDA, UCHIDA, CHILTON-COLBURN, COLLIER. Condensed steam is injected instantaneously in the liquid phase at a temperature equal to the wall temperature. - Transfer with the sump water uses convection correlation. - For the heat generation, the residual power law is imposed.

Injection, aspersion JERICHO has a detailed description of safety systems including: recirculation of sump water (injection in primary circuit), direct spray system, spray with recirculated water. In these last two cases, the condensation on droplets is modeled. 2.2.4 WECHSL-CALTHER WECHSL (9) is describing the corium concrete interaction and namely: concrete ablation by corium, corium chemical behavior, heat transfer between corium/concrete/atmosphere/water in the cavity. CALTHER (10) is dealing especially with the heat transfer calculation above the corium including concrete decomposition. The initialization of the codes is triggered after the lower head failure. WECHSL provides the energy source and the gas

SCADRE AND ICARE CODE SYSTEMS

189

Figure 3: Modelling scheme of WECHSL

emission for JERICHO. It determines the thermodynamic parameters which are needed for F.P. release from the corium. A picture of the main phenomena described in WECHSL is given on Figure 3. Concrete model The concrete ablation is driven by the fusion temperature. Concrete temperatures are obtained by 1D calculations axially and radially. The gases are released immediatly when the concrete reaches the fusion temperature. Corium model The corium is assumed to be constituted of two distinct layers: one oxide layer at the top, one metallic layer at the bottom. The gases which are released from the concrete are inducing corium swelling; the swell level is calculated by a mechanistic model. This level determines the part of the side wall where radial ablation can occur. Crust formation is calculated when the interface temperature becomes lower to the solidus temperature. The crust is assumed not to inhibit the gas release but introduces an additional thermal resistance. The chemical reactions which are described are oxidation of Zr, Cr, and Fe by H2O and CO2. The energy provided by these reactions are entering as sources in the thermal calculation. Cavity model Above the corium the cavity is modelled by CALTHER. The concrete walls ablation is calculated using a 1D representation. The modelling of the gas release from the concrete is based on 3 mechanisms depending on the temperature level: . free water release (saturation temperature) . water release from hydroxide decomposition . CO2 release from carbonate decomposition Thermal model The thermal modelling is very important in WECHSL-CALTHER as it determines most of the ablation phenomena. Heat transfer between corium and concrete is based on the boundary layer theory: on horizontal parts (see Figure 3) stable film, transition and nucleate boiling correlations are used; film stability is determined by the BERENSON criteria. For vertical parts only stable film regime is used. Heat transfer between the two corium layers is purely empirical. Between the corium and the atmosphere, only radiative heat transfer is taken into account in WECHSL. A special radiative model has been developed in CALTHER with transfer from the corium to the structures and heat absorption by the gases. Between corium and water in the cavity, heat transfer is calculated using usual film boiling, transition, nucleate boiling and natural convection correlations.

190

THE PHEBUS FISSION PRODUCT PROJECT

For heat generation, the contribution of residual power and the heat generated by corium chemical reaction are taken into account. 2.2.5 IODE IODE (11) is calculating the physico-chemical behavior of all iodine compounds in the containment and namely the chemical reactions and the numerous mass transfers (sources, leaks, transfers between sump/atmosphere, sump/structure and atmosphere/structure). Thermodynamics parameters are taken from JERICHO. Iodine sources are coming from the primary circuit on two forms: molecular iodine as vapor and CsI (I−) as aerosols. The aerosols are assumed to go immediately in the sump. Containment modelling for iodine The containment for iodine calculation is represented by only one compartment divided into two subvolumes: the sump (liquid subvolume), the atmosphere (gas subvolume). Iodine chemistry 8 species are considered in the atmosphere: I2, CH3I, HOI, I−, IO3−, AgI, R, CH3R. 11species are considered in the liquid solution: I2, CH3I, HOI, I−, IO3−, AgI, R, CH3R, Ag, CH3, I3− (R is an organic radical to be defined by the user). Chemical reactions between the different above compounds are summarized on Table 1. Activation energies and kinetics constants are determined from literature and from specific french experiments. Iodine physics For mass transfer between the sump and the atmosphere, diffusion is modeled by using the heat/mass transfer analogy theory. Partition coefficients for I2 and CH3I are taken from experimental values. For transfer from the atmosphere, two phenomena are modelled: the deposit related to condensation on structures (each species flux is proportional to the species concentration and to the condensation rate) and the adsorption on structures (3 constant adsorption velocities can be given in the input data). Table 1—Chemical reactions modeled in IODE

2.3 Assessment and application A significant effort has been done for validating ESCADRE with a special concern for the containment modules. For example the following tests have been used: -HEVA06: -EMIS: -MARVIKEN V: -CCE-LA4: -SURC4: -DEMONA B3: -BATTELLEF2:

Fission Product release under realistic hydrogen conditions (VULCAIN) (12) Aerosols behavior (AEROSOLS/B1) (13) Aerosols behavior (test2b) (AEROSOLS/B1) (14) Code comparison exercise for aerosols behavior (AEROSOLS/B2) (15) ISP on corium concrete interaction (WECHSL) (16) Containment thermalhydraulic behavior (JERICHO) (17) Multicompartment thermalhydraulic behavior (JERICHO) (18)

Besides this assesment a large analysis has been initiated on a validation matrix for ESCADRE (19). Starting from a review of the phenomenology and of the available experiments, a list has been tentatively selected in order to cover most of the phenomena and in order that the assessment of codes against such list should be as complete as possible. ESCADRE is the actual working tool for French Safety Analysis. In this context several applications on NPP have been performed and among them AB sequence (18),AF sequence and V sequence.

SCADRE AND ICARE CODE SYSTEMS

191

2.4 Future work Development of new modules is now concerning mainly the improvement of the lower head description (CALEBU code) for the late phase of the transient inside the primary circuit. It will include core support plate degradation and vessel failure. Concerning the system itself, an important effort is being undertaken to improve the links between codes. Till now the sequential coupling is obtained through manual data files processing. The development which is underway, consists in the automatization of the process using a data base (SIGAL structure) allowing either calculations by individual modules or complete calculation using all modules in a single run. The coupling which will be obtained, will remain sequential as it was designed in the first basic principles of the system. However this coupling will improve greatly user’s convenience and friendliness. For achieving this task, some rewriting of modules have been necessary. It has been the opportunity to make specific checks on the numerics of the code as for example mass conservation or energy conservation. Substantial improvements have then been obtained. Concerning the containment, the JERICHO, AEROSOLS and IODE codes are being implemented in the European Source Term Evaluation and Research code (ESTER). Most effort has now to be done on code assessment: the validation matrix (16) has to be finalized and applied in a systematic way in order to determine the exact code capabilities for reactor applications. 3 ICARE AND ASSOCIATED CODES 3.1 Overall system structure The objective of the second level of the two level approach is to have a complete capability of analyzing experiments and of making transposition to NPP. To reach this goal, an analysis of the phenomena to be predicted has been conducted and has allowed to determine the degree of physical detail needed for the mechanistic codes to be developed. For the primary circuit which will be only examined here, it appears that phenomena can be classified in large classes, described by specific codes, such as thermalhydraulic, core degradation, F.P. release and transport, chemistry. It appears also that significant interactions were existing between these classes. Therefore, a structure was necessary to handle the transient interactive coupling of the different codes during a calculation: this is achieved by the SIGAL structure which by a dynamic and structured data management allows such a flexible interactive code coupling. The review of the phenomena shows that for thermalhydraulic the CATHARE code was the necessary basis. For core degradation, ICARE code was developed and as starting basis for F.P., the code TRAPMELT was used to derive an optimized and improved code called TRAPF. CATHARE/ICARE/TRAPF constitutes then a complete mechanistic tool designed to predict the NPP primary circuit behavior during a severe accident as well as for analyzing simulation experiments. 3.2. Main codes characteristics 3.2.1. ICARE 2 ICARE 2 (20) is modelling the progression of core damage including: core heat up with chemical reactions, loss of geometry by melting and embrittlement, relocation of materials, fission product release. Core modelling An axial grid is used for space meshing. This grid is applied to fluid and structures. In the radial direction, the meshes are composed of several different structures (presently cylindrical structures) and fluid meshes corresponding to independent monodimensionnal channels (without crossflows). The structures meshes can be composed of different materials or materials mixtures (alloys, eutectics). By this noding capabilities, ICARE2 is able to represent large NPP core as well as specific core designs such as the ones of small scale experiments (PBF, CORA, PHEBUS).

192

THE PHEBUS FISSION PRODUCT PROJECT

Fluid dynamics The thermalhydraulic model of ICARE2 is derived from the 1D two-phase flow model of the CATHARE1 code. The liquid phase has been removed but new capabilities have been added for a better severe accident prediction: - non condensible gas has been introduced in the gas phase (generally hydrogen but possibly helium or argon). It is assumed that steam and non-condensible gas have same temperature and same velocity. Diffusion of noncondensible gas in steam is taken into account specially for hydrogen. - The geometry changes due to flowdown and relocation of molten materials are considered (flow areas and heat transfer areas). - In a blocked mesh by refrozen materials fluid equations are still solved but with minimum values given in the input data for fluid volume and for cross section areas. - Radial effects of fluid flows on temperatures gradient are approximated, in this no cross-flow approach, by the redistribution or inlet flowrates in the independent parallel channels according to flow blockages occurence. Thermal model The energy conservation is written in a lumped form for each structure mesh. The phase changes and the corresponding latent heat are taken into account. The numerical solution has been carefully chosen to correctly describe coupled phenomena (degree of implicitness) and nonlinearities (iterations). Heat generation is coming from several sources: - nuclear power for which axial and radial profiles, as well as power history are input data given. The power generated per mass unit at an axial level is then used during the transient to generate the nuclear power in a structure mesh in function of its mass of UO2. - electrical power generation for electrical experiments calculation (parallel heated rods made of several conductive structures in series) - the energy released by oxidation reactions (see below) and which are linked to the mass of zirconia formed or to the mass of stainless steel reacted within the structure mesh. - an automated power control model which gives the power history necessary to follow some prescribed temperature evolution (helpful for defining or analyzing some experiments) Heat transfers For conduction a thermal resistance is calculated for each face of the structures meshes. This resistance is a function of the geometry as well as of the average thermal conductivity in the direction associated with the face. This latter value depends on the thermal conductivity of the materials in the mesh and on their geometrical arrangement (stratified or mixed). For convective heat, the transfer between wall and fluid is expressed with the general usual correlations (Nusselt formulation with laminar, turbulent and free convection regimes). For radiative heat transfer, a two dimensionnal model has been developed based on the Hottel’s method (21) (22) and taking into account the non isotropic aspects of the reflected flux linked with convex geometry of fuel rods and shrouds (23). Each enclosure (numerical cell) corresponds to an axial region composed of several structures and fluid meshes. The fluid is considered as an homogeneous, non scattering, emitting and absorbing medium. The transmissivity of the gas between walls is expressed in each spectral absorbing band (5 for steam) using the statistical model of Goody-Mayer (21). Structure surfaces are supposed to be gray, diffuse emitters, absorbers and reflectors. Only a fraction of the reflected flux by a surface is considered as isotropic and the remaining fraction is sent back to the emitting surface independently of the view factor. Geometric mean beam lengths are estimated with an analytical formula (23) while view factors are computed using the Hottel’s crossed string method (21). Precalculated view factors can be entered (for example for known experiments). Automatic calculation is possible for any bundle by entering in the data deck a simple description of the relative positions of the rods. The energetic balance of radiative power in the enclosure gives the total interchange areas between structures and fluid. These quantities are reactualized at given times during the transient and are used in the energy balance equation of the structure meshes.

SCADRE AND ICARE CODE SYSTEMS

193

Materials relocation models The materials movements can take place either radially in an axial level (“decanting” model) or axially from an axial level to one below (“candling” model). Only a mixture (molten or dislocated material) can flow downwards and only if it is not any more enclosed by compact materials. In the decanting model, the radial movement is done with an infinite velocity and allows the calculation of the falldown of a mixture containing solid debris or A.I.C. interaction with guide tubes. The candling model is applied, in a time step, after the decanting model. It assumes a constant candling velocity. The mass transport of mixtures, when occurence criteria are reached, is calculated towards the level below. Possible mass freezing can occur and is deduced from the energy equation where phase change is taken into account. Possible flow limitations of the mixture are predicted by checking the total available volume (structure +fluid). The mass and chemical composition of the structures are changed when candling and decanting occur. The associated changes of conductive heat transfer and of enthalpy are taken into account in the energy equation. Chemical reactions (oxidation) Oxidation is the result of the interaction of two phenomena: the oxygen mass flux which can be provided by the fluid, the oxygen diffusion within the metallurgical layer. In unlimited steam conditions, oxidation of the metal requires a certain amount of oxygen. If the oxygen mass flux provided by the fluid is higher, the reaction can occur without limitation. In the opposite case the reaction must be limited to the oxygen amount available (starvation and blanketing effects). For zircaloy, in case of unlimited steam conditions the diffusion process of oxygen which corresponds to oxidation is represented by parabolic laws giving zirconia layer growth and total oxygen mass consumption. Cathcart, Urbanic and Prater correlations (24) (25) (26) are available in the code. Assuming that zirconia is stoechiometric, the oxygen consumed in the alpha-Zr(O) layer is obtained by difference between the total mass of oxygen and the oxygen used for zirconia layer growth. The growth of alpha-zircaloy layer can then be derived. In case of limited steam conditions (due to fluid limitations), it is assumed that available oxygen supplies preferentially the alpha-Zircaloy layer for its growth. For stainless steel, similar modelling is used. Spinel (FeCr2O4), wustite (Fe0.947O) chromic oxide (Cr2O3) and nickel oxide (NiO) are produced perfectly mixed during oxidation. The oxidation of wustite in magnetite (Fe3O4) can take place when all the stainless steel is oxidised in a mesh. Oxygen mass uptake is given by parabolic laws which are specific for different kinds of stainless steel. Chemical reactions (interactions) Interaction between UO2 and solid Zircaloy can occur if the contact is good enough (for example in case of high pressure transient) and if the temperature is sufficiently high (greater than 1473°K). The UO2 reacts with the beta zircaloy to form an eutectic U-Zr located between two layers of alpha zircaloy. The reduction of the fuel by zircaloy is assumed to stop when the prior beta zircaloy is consumed. The growth of the 3 layers are calculated with parabolic correlations developed by Hofmann (27). UO2 and ZrO2 dissolution by molten Zircaloy are treated simultaneously in the code. For zirconia dissolution, the process is assumed to be diffusion controlled and is modeled by parabolic laws (Hofmann studies (28)). For UO2 dissolution, three different physical approaches are available in the code. The first two (28) (29) are time dependent,diffusion controlled for Hofmann or convection controlled for Kim and Olander in case of vertical fuel surface. For Hofmann model, a parabolic correlation is used after a short incubation during which mass fraction of UO2 reaches 0.358 in the U-Zr-O mixture. For Kim and Olander, a transport equation is giving the UO2 concentration in the UO2-Zr mixture. In the last model, UO2 equilibrium concentration is assumed to be reached instantaneously. The solubility limit curve in function of temperature is taken from the pseudo-binary diagram from Politis (30) or given as input data. Mechanical model No stress nor ballooning calculations are presently incorporated in the code. A simple loss of integrity model has been introduced with three possible physical states for a material (compact, cracked, dislocated) and with users defined criteria for changes of state. “No change” option can be chosen and the material will stay as initially until it melts. On the reverse, criteria can be defined using physical parameters such as temperature, layer thickness, temperature gradient with time or space, or any combination of parameters.

194

THE PHEBUS FISSION PRODUCT PROJECT

3.2.2. TRAPF TRAPF (31) describes the transport of fission products in the primary circuit after their release from the core. The release itself is calculated by ICARE2 either using CORSOR-M model which have been incorporated to ICARE or using EMIS model which is being validated on HEVA. TRAPF has been derived from the TRAPMELT code. The improvements which have been added are: - improved numerics. - Zimproved couplability to thermalhydraulic codes by introduction in the SIGAL structure. These two last points have required a complete rewriting of the code. - improved physics by the addition of new physical models: difrusiophoresis, bend impaction. 3.2.3. CATHARE CATHARE (32) (33) is one of the major advanced thermalhydraulic code. Use of this code in severe accident situation has been chosen for its very well adaptated capabilities: - The level of its two phase flow model is necessary for describing several situations in the primary circuit such as steam generator behavior (condensation, leak, actions related to accident management), plug formation and expulsion, blowdown, emergency injection…which are usual situations in severe accident scenario. - The fluid model includes besides steam, two non condensible gases which leads to the capability of handling situations with large contents of hydrogen or even fission product vapor. - Due to its model flexibility, it can describe the whole sequence and avoid the use of a different code for the beginning of the accident. - Its flexibility of topology enables any calculation: NPP or small scale experiment. - It benefits of a very large, extensive and continuous assessment program. - It benefits of a continuing development program which enables addition of new capabilities. Some of these future capabilities will be of great interest for severe accident prediction such as the increase of the number of distinct noncondensibles (up to 4) and such as the complete 3D two phase flow model. 3.3 Assessment and Application Most efforts have been focused till now on ICARE2 validation as CATHARE and TRAPF (or TRAPMELT) have their own independent assessment program. ICARE2 assessment is composed of qualification on separate effect tests. Validation of models for chemical reactions between materials are entering for example this category. ICARE2 validation is composed also of verification on more global experiments. The tests used for this validation have been chosen for their specifities which enable the verification of some sensitive models: tests with different bundle geometry to check the radiation model; tests with special conditions of core degradation (full steam starvation, with control rod or without,…); tests at different scales to check the model of melting and relocation. The following validation program has then been conducted: - PHEB US SFD C3: - PBF ST: - CORA5: - PBF SFD1.4: - TMI2:

analysis of interaction between UO2 and solid Zr analysis of oxidation of Zr by steam at high flowrates analysis of PWR absorber rod in severe condition (34) similar to CORA5 but in pile and different geometry whole core degradation on NPP with crucible formation (35)

An example of results obtained on TMI2 is given on Figure 4 (crucible formation) which illustrates the ICARE capabilities. The preceding list of developmental assesment, has to be completed with the verification of the code on all PHEBUS SFD tests which is performed during the tests interpretation, and especially by the participation to ISP28 (36). Application of ICARE2 has been done on a french PWR (analysis of full steam starved conditions) but remains limited because of the limited coupling with CATHARE code. On the reverse a full coupling of TRAPF and CATHARE has been realized which allows full applications for example for PHEBUS FP tests preparation.

SCADRE AND ICARE CODE SYSTEMS

195

Figure 4: TMI2 Core state at 174 mn as calculated by ICARE2

3.4. Future work The version 2. of ICARE2 was released at the beginning of 1991. For this version the efforts have been focused on documentation and verification with implementation of new modules. Additional models to ICARE have to be developed especially for the late degradation phase (debris bed, melting pool, fuel coolant interaction, vessel rupture). These models are being analysed. The next, actually ongoing step consists in developing a full coupling with the CATHARE2 code: ICARE2 will be an optionnal core module of CATHARE. By this coupling, it will be possible to take into account two phase flow situations in the core, to cope with multidimensional configurations in the reactor vessel (with the next version of CATHARE) and to compute simultaneously the core degradation and the thermalhydraulic behavior of the whole circuit (primary and secondary if needed). As CATHARE is already coupled with TRAPF a complete set of codes CATHARE/ICARE/TRAPF will allow a complete calculation of the primary circuit. ICARE2 code is being implemented in the European Source Term Evaluation and Research code (ESTER). 4. CONCLUSION Code development program in France is following a two level approach. In the first level, the ESCADRE system is providing a complete and useful tool for safety analysis. All transient phases are described: core degradation and corresponding thermalhydraulic primary circuit behavior with VULCAIN code, fission products transport in the circuit with SOPHIE (vapor) and AEROSOLS/CIRCUIT codes, containment with JERICHO code, aerosols in containment with AEROSOLS/B2, iodine with IODE and finally corium concrete interaction with WECHSL code. This complete set of codes is able to give predictions of severe accident transients on NPP’s which can be used for safety studies but which can also help in experiments definition by providing reference plant calculations. The ongoing work on ESCADRE will give a better user’s convenience by allowing an automatic sequential coupling if desired together with a stand-alone calculation capability. In the second level, most efforts have been focused till now on the primary circuit part. A mechanistic code for the core degradation is being developed: ICARE. Interesting results have been already obtained in the verification which show the

196

THE PHEBUS FISSION PRODUCT PROJECT

capabilities of the code in improving together with experimental programs the physical understanding of what occur in the reactor vessel. This improvement will be largely increased when the interactive transient coupling with the thermalhydraulic code CATHARE2 will be performed: this will enlarge the best estimate evaluation to the whole primary circuit and will add new indispensable capabilities such as 3D calculations (reactor vessel) and full two phase flow calculations (accident procedures, core quenching,…). The coupling of fission product code with CATHARE2 will finally provide a complete best estimate code for the primary circuit as it can be defined from the phenomenological analysis of severe accidents. It comes out that more development is needed for both code levels, on the late phase of the transient inside the reactor vessel, i.e. on lower head failure. Complete and extensive plans of code assessment are also required for both kind of codes. In this direction, a study has been started in order to establish a validation matrix which can be applied generally to severe accident codes. PHEBUS programs (SFD and FP) will give a significant contribution to this matrix. Finally, all the ongoing code development program is performed in close connection with the European Community Program: experience gained in the informatic structure SIGAL has been transmitted in the elaboration of the RS YGAL structure of ESTER code; codes from ESCADRE system are implemented in ESTER; they are JERICHO, AEROSOLS and IODE; ICARE2 code is also being implemented in ESTER. This cooperation will certainly insure reaching the final objective of better physically based analytical tools for NPP severe accident prediction. REFERENCES: 1 2 3 4 5 6 7 8 9 10 11 12

13 14 15 16 17 18

19 20 21 22

() J.DUFRESNE et al. Presentation of the ESCADRE system, together with a practical application. International Symposium on Severe Accidents in Nuclear Power Plants. Sorrento, Italy, March 1988. () A.PORRACCHIA et al. Review of main computer codes. Chapter of the OECD State Of The Art Report on Core Degradation. To be published. () C.GERBAUX, J.M.DUMAS. Système ESCADRE-Code VULCAIN Version 3.05 Modélisation de la degradation du coeur et de la thermohydrtaulique u circuit primaire lors d’un accident grave sur un REP. Note SETh/leml/89–169. () J.M.DUMAS, C.GERBAUX. Système ESCADRE-Code VULCAIN. Note d’information sur les modalités d’utilisation de la Version 3.6. () LE MAROIS et al. Source term experiment for the assessment of FP release and transport: the HEVA () C.LEUTHROT. Transport et dépot des produits de fission sous forme vapeur dans le circuit primaire d’un REP lors d’un accident severe, code SOPHIE Version 2.0. Note technique SCOS/LCC/89/041. () J.GAUVAIN, G.LHIAUBET. AEROSOLS/B2-version2. Code aérosols à spectre multicomposant discrétisé avec condensation en pluie de vapeur d’eau.Présentation du formalisme. Rapport Technique SASC/87/54. () J.GAUVAIN, J.P.L’HERITEAU, D.HOUCQUE. Système ESCADRE. code JERICHO Version 2. Thermohydraulique de l’enceinte d’un réacteur lors d’un accident grave. Rapport Technique SASC/88/63. () M.REIMANN, S.STIEFEL. The WECHSL-Mod2 Code: A computer program for the interaction of a core melt with concrete including the long term behavior. Model description and user’s manual. Report KfK 4477. () B.ADROGUER, G.CENERINO. Heat transfer in reactor cavity during core- concrete interaction. ICHMT International Seminar on Fission Product Transport Proceeses in Reactor Accidents. Dubrovnik, Yugoslavia, 1989. () J.GAUVAIN, P.PEPIN, M.FILIPPI. Système ESCADRE. code IODE Version 2. Transferts d’iode dans l’enceinte d’un réacteur lors d’un accident grave. Rapport Technique SASC/89/66. () J.M.DUMAS, G.LHIAUBET, G.LE MAROIS, G.DUCROS. Fuel behaviour and fission product release under realistic hydrogen conditions. Comparisons between HEVA06 test results and VULCAIN computation. ICHMT International Seminar on Fission Product Transport Processes in Reactor Accidents, Dubrovnik, Yugoslavia, 1989. () G.LHIAUBET. Résultats comparés du code AEROSOLS/B1 aux expériences EMIS et au code AEROSOLS/A2. Rapport Technique DEMT/84/174. () J.GAUVAIN, G.LHIAUBET. Marviken V-test 2b- Comparison between test and calculation with AEROSOL/B1 code. Presentation au 5° TAC Marviken DAS/SAER/84/682. () J.GAUVAIN. AEROSOLS/B2 Calculations for the CCE-LA4 aerosols code comparison exercise. Note SASC/87/853. () G.CENERINO, J.P.L’HERITEAU. CSNIPWG2 Ex-vessel task group. Standard problem SP3: SURC4 test calculation. Rapport Technique SASC/88/60. () J.GAUVAIN. Post test calculation of thermalhydraulic behavior in DEMONA experiment B3 with variouscomputer codes used in EC member states. Rapport EUR 12197 EN. () C.RENAULT, P.DUMAZ, C.HUEBER, J.P.L’HERITEAU. ESCADRE an adaptative system of stand alone or coupled codes for the analysis of water-cooled reactor severe accidents. Application to the modeling of containment behavior. ANS meeting Portland 1991, to be published. () C.RENAULT. Proposal of validation matrix for severe accident codes. Private communication. () P.CHATELARD, R.GONZALEZ, F.JACQ, A.PORRACCHIA. General description of the second version ICARE2. IAEA Technical committee. Workshop on computer aided safety analysis. BERLIN, April 1989. () H.C.HOTTEL, A.F.SAROFIM. Radiative heat transfer. Mac Graw Hill, 1967. () MANOHAR S.SOHAL. A radiation heat transfer model for the SCDAP code. Nuclear Technology -vol. 75—November 1986.

SCADRE AND ICARE CODE SYSTEMS

23 24 25 26 27 28 29 30 31 32 33 34 35 36

197

() A.MANDELL. Geometric view factors for radiative transfer within boiling water reactor fuel bundles. Nuclear Technology, 52 , 1981. () J.V.CATHCART, R.E.PAWEL. The kinetics of oxidation of zircaloy-4 in steam at high temperatures. Electrochemical Science and Technology. July 1979. () V.F.URBANIC, T.R.HEIDRICK. High temperature oxidation of zircaloy-2 and zircaloy-4 in steam. Journal of Nuclear Materials, 1978. () J.T.PRATER, E.L.COURTRIGHT. Properties of reactor fuel rodmaterials at high temperatures. NUREG/CR-4891. July 1987. () P.HOFMANN. UO2/Zr4-Chemical interactions and reactions kinetics from 1000°C to 1700°C. KfK 3552, November 1983. () P.HOFMAN, S.HAGEN, G.SCHANZ, A.SKOKAN. Chemical interactions of reactor core materials up to very high temperatures. KfK 4485. Januar 1989. () K.T.KIM, D.R.OLANDER. Dissolution of UO2 by molten zircaloy (convection and diffusion controlled reactions). Journal of Nuclear Materials, 1988. () POLITIS. Untersuchungen in Dreistoffsystem Uran-Zirkon sauerstoff. KfK 2167, 1979. () V.D.LAYLY, J.C.LATCHE. TRAPF a french stand alone module for the transport of fission products in a circuit (CEA tool derived from TRAPMELT). Private communication, Janvier 1991. () J.C.MICAELLI. Cathare best-estimate themalhydraulic code for reactor safety studies. Int. Conf. Thermal Reactor Safety, Avignon, France, October 1988. () A.FORGE et al. Comparison of thermalhydraulic safety codes for PWR systems. Graham and Trotman/CEC. () R.GONZALEZ, M.C.VICENTE,S.HAGEN. Results of ICARE2 validation on the severe fuel damage test CORA-5. To be published. () P.CHATELARD, R.GONZALEZ, A.PORRACCHIA. Analysis of the TMI2 core degradation using ICARE2 code. 27th ASME/ AICHE/ANS National Heat Transfer Conference MINNEAPOLIS USA, July 1991. To be published. () S.BOURDON, P.VILLALIBRE, B.ADROGUER, G.GEOFFROY. Analysis of the severe fuel damage test PHEBUS B9+ using ICARE code. 27th ASME/AICHE/ANS National Heat Transfer Conference MINNEAPOLIS USA, July 1991. To be published.

ESTER—a European Source Term Evaluation System A.V.Jones, I.Shepherd. (CEC Joint Research Centre, Safety Technology Institute)

Abstract The CEC sponsors considerable model development and validation in the area of LWR source term, and naturally wishes to see the results used as widely as possible. It also has a role in fostering collaboration between European teams involved in source term analysis, for which purpose Phebus-FP is acting as a focal point. To further both aims the JRC decided in 1989 to sponsor the development of the best-estimate code ESTER, which is both a software environment and a set of coupled source term modules which when completed should offer potentialities not currently available within Europe. This paper describes first the overall architecture of ESTER, then the component parts: the tools and services, the user interface, and the modules which perform the physics and chemistry calculations, emphasizing the design choices which have been made. The quality assurance system for the whole system is also re viewed. Contributions from the model developers both underway and expected are then surveyed in the context of the overall development of ESTER, and the planning of the creation and extension of ESTER is given. The paper closes with some proposals for sharing ESTER within Europe and for ensuring its maintenance and continued rational development. 1. Introduction The Phase B shared-cost action (SCAs) exercise for the dimensioning of Phebus-FP /1, 2/ made it clear that while a wide variety of source term codes are in use in Europe, intercomparison of their predictions is rendered difficult by the different environments in which they run, meaning different computers, different operating systems, different input and output formats etc. The codes and modules nevertheless represent the state of the art in source term predictions, and much could be learned from easier intercomparison. Furthermore, considerable effort is going into the development of source term modelling, but by different teams for different codes, leading both to some duplication and to difficulties in transferring improvements from one code to another. A third element in this picture is the CEC’s own contribution to source term model development and validation through its SCAs. The resulting developments, although worthwhile, have tended not to be used outside the team which originated them, partly because of the effort needed to transfer the models to a different code. The solution to all these problems proposed by the Commission was to develop a unique framework allowing the easy integration of new modules, and to urge European model developers to couple the best of existing codes or modules into it, and then make any further model improvements only in the context of the framework. In view of the traditional attachment of organisations and particularly developers to their own products this proposal might have seemed Utopian, but a combination of funding restrictions across the entire nuclear scene and continued interest in severe accidents resulted in the idea of a code framework being welcomed, and in 1989 the JRC proposed a general architecture to a group of national experts, together with a name, ESTER (European Source TERm Evaluation System). The architecture is presented in Section 2. Here we just emphasise that it is intended to eliminate direct communication between the various modules, and that systems with similar architecture are widely used in real-time control systems and other nonnuclear applications. The group approved both the name and the architecture, and stressed the importance of developing the framework with the utmost attention to QA (quality assurance). Thus encouraged the JRC launched a call for tender for an SCA to develop the ESTER framework. The contracts subsequently signed by CIS I Ingenerie (principal developer and manager) and by IKE Stuttgart (developer) undertook to develop the framework, including two trial modules, by the end of 1991, and included a tight specification setting out the tasks to be achieved and the timetable. The ESTER development project was thus launched at the end of 1989.

ESTER—A EUROPEAN SOURCE TERM EVALUATION SYSTEM

199

Fig. 1 Architecture of ESTER

2. What is ESTER? ESTER is simultaneously a framework, a set of tools and services, a group of coupled modules, a controller and a user interface, and a series of documents,all governed by QA procedures. ESTER is firstly a framework with a clear architecture, as shown in Fig. 1 Probably the main obstacle to coupling together existing modules or codes is the difficulty in transmitting data between them through t he mechanisms of FORTRAN such as argument lists or COMMON blocks. Experience shows that there is a strong risk that variables will be used before they have been defined, or that data will be “overwritten” i.e. accidentally replaced with other data by one of the modules of the system. Another difficulty with the traditional approach is encountered when due to a modification one routine needs to access certain data calculated in another routine; numerous other routines need to be modified in consequence just to transmit the required data. The architecture of ESTER is designed to overcome these problems by restricting communication between modules, structuring the data and controlling access to the data. Various modules are assumed to be available which can calculate aspects of a severe accident or an in-pile experiment e.g. fluid flow through the core, or clad relocation and oxidation, so that a complete calculation can be performed by invoking them in sequence, perhaps several times. In an initial phase input data are prepared and stored on the central database by the controller. The controller then invokes the modules as required, adjusting the overall timestep to achieve stability. Each module naturally has an interface to the controller, to enable it to receive the message to begin work, and to pass back the message that work is complete and/or error diagnostics. Modules require data, which they obtain from the database via another interface. Results from the calculations of the module are returned to the database through the same interface. When required the controller can use the contents of the database to produce graphics, printout or messages to the user informing him of the progress of the calculation. Notice that for the introduction of a new module one need only supply two interfaces, and make some adjustments to the logic of the controller. ESTER is thus a framework. It is also a set of tools and services. These include graphics, menu management for the user interface, a data checker, systems for adding to and modifying the database etc. Section 5 contains a brief survey of the tools and services. From a third, practical point of view ESTER is primarily a set of calculational capabilities, contained in the modules. The philosophy of ESTER in this regard is that existing modules represent the state of the art, but that communication between them is often too poor (or even non-existent) to represent the physical coupling of phenomena and events which characterise a severe accident. ESTER is designed to “mix and match” existing modules so as to obtain previously unavailable calculational abilities e.g. circuit thermal-hydraulics coupled with fission product chemistry. The modules are under continual development in many national laboratories, and the results can be used to improve ESTER at low additional cost. Only as a last resort is it intended to develop new modules for ESTER from scratch. Fourthly, the user of ESTER sees not the individual modules but the user interface and through it the controller plus any on-line graphics, error

200

THE PHEBUS FISSION PRODUCT PROJECT

Fig. 2 System Developer's View of ESTER

messages, and other outputs channeled through the controller to the user interface. The interface includes facilities for preparing input data and guiding the course of the calculation. Hence the addition of a new module does not change the “look and feel” of ESTER, and no retraining is needed to use the system in its new and extended form. Lastly ESTER is a set of documents which specifies the design choices, the quality assurance system, the potentialities of the tools and services, the user interface and the interfaces to modules. Also in preparation is a series of documents which will explain to code developers how to proceed when integrating an existing module into ESTER or writing a new module so that it can be integrated immediately into ESTER. There will also of course be a user manual, listings etc. for the complete system. The sections which follow look at some key ESTER documents as a way to illustrate the path taken by the development and to give a stronger feel for the scope of ESTER. 3. ESTER Specification A first specification of ESTER appeared in the contract, and it has been amplified and extended in successive iterations of the Specification Report. Key choices, apart from that of the architecture described above, were that where possible the same models should be used to describe the same phenomena wherever they might occur e.g. aerosol deposition in the RCS or the containment, and that modules should be incorporated from different sources, and might be in different languages. In practice all source term codes and modules of significance are written in FORTRAN77, and it was decided to use this language for ESTER and its tools also, because of its portability and ubiquity, unless there were special reasons to use another language e.g. in menu management. For the same reasons that Fortran was chosen as the programming language the operating system was chosen to be UNIX. It was required that ESTER should be proven to be portable across a wide range of machines with minimal effort, and provisions for parallel processing and networking should be included from the outset. Attention was drawn in the specifications to the importance of the database structure, and to the development and implementation of a Quality Plan (see next section). The specification report also sets out the development and testing plan, which is given in Section 10 of this paper. In a nutshell the plan is that a preliminary version is to be delivered to the JRC (which may invite selected users to assist in portability and other tests), which after optimisation will form the basis of the final version to be delivered together with documentation at the end of the CISI-IKE contract. A different view of the architecture of ESTER is shown in Fig.2, which omits the user interface, controller and modules to place more emphasis on the tools and database. The database is chosen for efficiency and portability reasons to reside not on disk as in the RSYST system of IKE/3/ but in the memory of the principal machine running ESTER. This choice has

ESTER—A EUROPEAN SOURCE TERM EVALUATION SYSTEM

201

Fig. 3 Tree-structured Database

some consequences for parallelisation, as will be seen later. The database is created and manipulated by a set of tools and is provided with input and output services, tools and services together having the general name of RSYGAL, and developed by CISI from the smaller SIGAL set used e.g. in the CEA core degradation code ICARE-2.When running ESTER the user chooses the particular configuration desired (i.e. choice of modules, integration scheme etc.), as well as the input files to read in, from menus on his workstation. All the checking operations, reading and writing to the database and storage on disk of restart files are then performed automatically by the controller, using the tools described in Section 5. The structure of the database is obviously a key question for the success of ESTER The basic mechanism of the database is quite simple: RSYGAL takes a large section of the machine memory as a Fortran COMMON, and then divides it into segments (continuous areas), each of which contains a “complex data object”. The segments are identified by pointers, and memory management modules manipulate only the pointers, not the segments. The programmer has access to the pointer value via the variable name. Complex data objects are databases or sub-databases and tables. Tables may be vectors of reals, integers or names/text strings, matrices of reals or integers or “couples”, which are name-addressed reals or integers. All objects in databases are associated with semantic “attributes” (single-valued quantities). Each attribute is characterised by a name, a type, a version number, and its value. For simple data objects (scalars) this value is just the value of the scalar. For attributes which represent complex data objects the value is the pointer to the segment containing the object. Fig.3 shows how this arrangement can be used to produce a hierarchical (tree) structure of the data. RSYGAL tools allow one to create, delete and copy databases. Tables can also be created, deleted, modified, retrieved etc. The ability to manipulate databases or tables as a whole gives ESTER certain characteristics of an object-oriented database. However features such as encapsulation and inheritance are absent, largely because they are not present in the design philosophy of typical source term modules. Section 6 looks at the question of database design specifically for source term problems. Notice that apart from the particular database structure chosen the ESTER framework could be used for any time-dependent problem solved by interlinked computational modules. Other major components of ESTER are the user interface and the driver. The user interface offers the user a succession of menus whereby he can (a) select the combination of codes/modules he wishes to use (b) choose to build up or modify data, run the job, or postprocess results, and (c) (depending on the choice made): construct or edit an input file; specify run parameters or select the file to be postprocessed and the tools to apply. The driver is responsible for memory initialisation, data reading and checking, sequential calling of the modules, management of the overall timestep, saving of databases for restarts or post-processing, and the production of job-dependent printouts. A general technique for the determination of overall tirhesteps will be included in ESTER, using whatever integration technique is provided within each module, an explicit time integration between modules, and a quadratic scheme for error control which has been applied with success in RSYST. More sophisticated overall timestep control may be implemented later. As well as the components already mentioned ESTER will be accompanied by maintenance tools to allow the upkeep of libraries of module or data versions, and of a database recording the relationships between codes, modules, versions and the files which actually contain the libraries. The tools will e.g. be able to identify all the codes, files and directories affected by the modification of a given module. Note that the specifications were not fully defined at the outset. ESTER to some extent breaks new ground, and its specification is therefore a continuous learning process for all parties. Section 8 describes some of the changes made since the first draft specification in late 1989.

202

THE PHEBUS FISSION PRODUCT PROJECT

Fig. 4 ESTER by the End of 1992

4. The Quality Plan One requirement of the contract was that the work should follow an approved quality plan, to be drawn up by CISI. The latest version was produced in November 1990. It begins by stating the contributions which each organisation will make to the development task, including the JRC and CEA (which contributes 40% of the costs of CISI). The Plan lists the staff, the codes and other tools they will employ, and the hardware upon which they will run. In essence CISI develops the RSYGAL tools and the UNIX operating environment, while IKE is responsible for developing integration techniques, integrating ICARE and FPRATE, defining a common database, and developing time integration techniques. Each party is responsible for checking the work produced by the other, and for testing the portability of software by running it on their own machines under its own operating system. The two contractors will also train specified users in the running of ESTER. The JRC (with CEA) is responsible for the overall supervision of the development work, for acceptance of reports and software, and for testing portability on their own machines. The JRC may also execute a QA audit if so desired. The work to be done is divided into a number of topics or “actions”, and then each action is further split into a design phase, a development phase, and a testing phase. There is an acceptance procedure for each phase of each action, and a procedure to deal with any nonconformities. Report titles, layouts and groupings (for contractual purposes) are specified, as are test procedures for the contractors. Such tests include elementary tests (the routine must perform its function on a series of test cases); global tests (ditto, in the context of an overall code, as if it was in the hands of a final user); non-regression tests (a global test to check that performance is not degraded after a modification), and transfer tests (to be sure that the routine works on a new machine). The contractors routinely exchange documents and routines by electronic mail, and make use of standard forms for signalling errors and logging corrections. There is also the question of QA for the data used by ESTER modules, such as material properties and thermodynamic data. The quality of such data is the responsibility of their originators but also of the JRC, which will remain responsible for choosing which data to incorporate in ESTER. The SCAs of recent years reviewing and extending databases, particularly for chemical species, have been of great assistance here. In summary QA requires considerable time and care but is recognised by all concerned as essential if ESTER is to be reliable and wellqualified. 5. ESTER Tools and Services ESTER is provided with RSYGAL tools for memory management and data management, with more general tools like the reader, analyzer and checker, and with services such as a graphics package, a menu management tool, and a user interface. These go to make up the ESTER “environment” within which all further development will take place. The tools for memory management and data management have been reviewed briefly in Section 2. Here we examine the data reader, checker and analyzer, referring to Fig. 2, before going on to look at the services. The user interface and controller give the signal to read input data, which is done using the READER. The data reader is designed to read in the data of any module included in ESTER. The data are divided into blocks indicated by keywords; the reader recognises the keywords and constructs the appropriate databases to store the input data in the central memory. It also generates standard error messages if syntax errors are discovered in the data. The data CHECKER is a small expert system which checks the conformity of the input data against rules specified by the user and stored in a separate rule file. The rules may concern the presence of obligatory or optional keywords, the use of the correct data type, the sizes of arrays, that values lie within specified ranges, and that a keyword introducing a particular set of data is followed by those data. The CHECKER thus replaces the long series of error traps to be found in the input routines of many codes. The ANALYZER is a restricted language interpreter which receives and executes user instructions to operate on the database e.g. extract information for post-processing, or add to or modify the database e.g.

ESTER—A EUROPEAN SOURCE TERM EVALUATION SYSTEM

203

in preparation for a restart. The ANALYZER can also perform more complex data checks than the CHECKER. The database may produce graphical output using the graphics package, referred to in Fig.2 as TIC. Modules (not shown) accessing or writing to the database use a set of tools called the database management library, while at system level the computer memory is handled by the memory management library. The tools described above are all written in Fortran and are fully portable. Portability considerations become more significant when one turns to the services provided with ESTER, particularly the graphics system and the menu management/ user interface tools. This situation is not the fault of the ESTER developers; it merely reflects the current absence of standardisation in the area of graphics and on-screen displays. The more powerful graphics systems are either not portable or are discouragingly expensive or both, and for ESTER the choice was made to develop a series of higher-level facilities on top of the GKS graphics primitives, on the assumption that most laboratories have GKS in one version or another, so that little change should be needed to implement ESTER graphics at a new site or on a new machine. The facilities offered include the ability to plot (in colour) a series of vectors e.g. in the familiar x-y plots, a matrix (2-d plots, contour plots), FE or FD meshes, and results on such meshes (iso-lines). More graphics facilities may be added later, as time permits. The menu management system is designed to operate under X-windows, now to be found on all UNIX workstations and also available on PCs. The option of using MOTIF has been discussed, but this package has not been generally adopted, so that something simpler and in the public domain was chosen. Lastly the user interface is intended to be displayed on a workstation or PC; it is designed to be as “friendly” as possible without spending too much development time on fancy displays, and will offer choices to the user through a sequence of menus handled by the menu management system. Run-time messages and graphics will be returned in separate windows on the same screen. This is in accord with the general trend in interfaces to large codes. ESTER is not intended to be run interactively, but there is nothing in the overall architecture to prevent this way of working. It is more a question of how the individual physics and chemistry modules were designed (usually for batch operation), and the rather long running time of certain source term modules on current computers. 6. Data Structure The importance of the data structure for the success of ESTER has been stressed earlier. The choice of database design essentially depends on the codes (modules) which will be incorporated. The preliminary version of ESTER will include the CEA core degradation code ICARE-2 v2 /4/ and the fission product release module FPRATE from the IKE core degradation system KESS/5/. ICARE and KESS have quite different data structures, although both discretise the system into volumes. In ICARE all the data for a given volume are stored together, while KESS stores variables of like kind together e.g. all the pressures are stored in a single array. There are many other differences in data structure. ICARE being larger than FPRATE and constructed using SIGAL which is similar to RSYGAL, it was decided to use the data structure of ICARE for the core region in ESTER (it is sufficiently general to handle reactor cores as well as bundles), and to write an interface for FPRATE which converts that module’s data into the same form as those of ICARE. The interface will collect data of like kind from all the volumes, and return them to the volumes after they have been updated. The interfacing system will be tested during the integration of FPRATE from KESS. The data structure of ICARE is worth a word of explanation, since it illustrates some possibilities of the RSYGAL database organisation. Physical objects (rods etc.) are associated with databases of axial meshes, each of which may contain several components e.g. layers of cladding. The components in turn have their own sub-databases corresponding to the different physical models e.g. geometric data, material, material properties, internal heat source and connections to other components. Knowing the overall data structure it is a simple matter to obtain e.g. the mass of the oxide layer on the cladding of a particular rod within a particular mesh, by proceeding down the tree structure to find the required elements. The choice of a volume-oriented data-structure is in some sense natural in a severe accident code, firstly because much of the physical interest is in phenomena such as aerosol agglomeration or chemical equilibrium which tend to be calculated one cell at a time, and secondly because in the core region at least the geometry changes with time so that control volumes can appear and disappear as the calculation progresses. While the choice of data structure appears clear for the core or bundle there is more difficulty over the circuit. The obvious choice for the thermal-hydraulics of the circuit would be a two-phase “system code” such as CATHARE or ATHLET. In meetings between the developers of these codes, the JRC and the ESTER contractors some rather uncomfortable facts emerged however. Firstly, both codes are much larger than all the remaining possible component modules of ESTER put together. The data structure of the system thermal-hydraulics code would thus tend to dominate ESTER as a whole. Secondly, although they both chose finite volume treatments of the circuit CATHARE and ATHLET thereafter made such different choices and so many of them that adopting a common data structure would require re-writing one or both codes. This would be far too demanding a task, especially since thermal-hydraulics is not the main focus of ESTER. Thirdly, the specification and discretisation of components such as vessels and pumps in the two codes is so different that it is difficult to imagine even a common input data format and reader. Both codes are thus sufficiently large and sufficiently set in their ways that their

204

THE PHEBUS FISSION PRODUCT PROJECT

adaptation to a newcomer like ESTER would require a great deal of work. Of course if a joint European system code is developed in the future it could be written to ESTER standards and conventions from the outset and the problem would disappear. It is also possible that one or both of the codes will be offered for integration in ESTER in the current round of SCAs (see Section 10); how ESTER develops partly depends on what is offered by European organisations! For the calculation of Phebus FP at least most tests require only single-phase fluid dynamics in the circuit. The DEIMOS code developed for the JRC by Matec/6/ could calculate such problems quite well, and may be integrated in the near future. Also associated with the circuit are fission product and vapour physics and chemistry. A discretisation into volumes as in the bundle is natural in the circuit, and matches the data structure of codes such as VICTORIA. A volume-oriented data structure as in ICARE thus seems suitable for the circuit FP transport as well as the thermal-hydraulics. For the containment the natural discretisation is into interconnected subcompartments (unless a fully three-dimensional treatment is envisaged), as a discussion between experts on CONTAIN and JERICHO emphasised. Each volume may then have sub-databases describing the walls, internal structures, sump, and atmosphere. This last will have its own database including thermal-hydraulic variables plus an aerosols database and so on. The details will become clearer as code integration proceeds, but the overall philosophy is clear. The description for all systems (core, circuit, containment) is to be in terms of volumes, with a well-defined and natural tree structure being imposed on the data describing each volume. 7. Code Integration An essential element in the ESTER documentation is the Integration Report. This will not only describe how ICARE and FPRATE were integrated into ESTER but provide guidelines as to how to integrate existing modules, and how to write new modules so as to facilitate integration. For existing modules or codes the procedure is well-defined. The user manual or other input description should first be used to write the rule file for the data checker. The code should then be split into three parts: preparation (reading in user data and/or a restart file, initialisation, opening scratch and output files etc.), the calculation itself, and post-processing, including output printing and graphics. The preprocessing part should then be replaced by calls to the data reader and data checker. Interfaces should be provided between the code and the database and controller, and all postprocessing operations should be left to the controller, calling on ESTER services as required. Codes brought into ESTER in this way are not integrated on a fine scale. For instance, a code may contain a routine for calculating certain material properties which are also required by another ESTER module. At this level of integration the desired properties could only be obtained by executing the entire code. If it is desired to break down a code or module into smaller modules individual interface routines must be created for them and the controller must be modified to access the module whenever its services are required, whether by the original code or by another module. The smaller modules thus become “functional modules”, fulfilling a specified function for many other modules. How finely one should modularise in ESTER will be decided pragmatically, as a balance between redundancy of coding and run-time efficiency. To avoid confusion certain conventions must be adhered to; the subroutine names must start with two significant letters characterising the module, as should the common blocks. Because the type of a Fortran function is determined by its first letter this letter is fixed, and the second and third letters of the function name then characterise the module. ESTER tools are under development which allow the necessary name changes to be made automatically. 8. Review of Specifications At least four meetings per year have been arranged with the developers, and during the discussions on designing the data structure meetings were more frequent still. Such meetings reviewed the progress made but also looked again at the feasibility or advisability of some of the choices made in earlier drafts of the specification report. One example concerns “external” software. ESTER must rely to some extent on software not created by the developers; compilers for instance are assumed to be available on the users’ systems. Graphics and menu management software is more problematical.The decision not to use MOTIF has been mentioned earlier. A preference was also expressed for the screen handling system PHIGS in the early days; unfortunately the situation turned out to be similar to that for MOTIF. Different vendors have produced different versions of PHIGS, and all of them are rather costly. To favour the portability and acceptance of ESTER cheaper and more readily available alternatives such as X-windows and GKS were therefore selected, even if these are not the most advanced available. There was also considerable discussion on data structure before the philosophy outlined in the previous section was adopted. The most knotty problem of specification however proved to be that caused by the reference in the original specification to networking and parallel processing. One paradigm of a general computing environment which is becoming more and more popular is that of distributed processing. In distributed processing the various modules (or processes) involved in the computation of a particular problem run on more

ESTER—A EUROPEAN SOURCE TERM EVALUATION SYSTEM

205

than one machine, which may be of different types and at more than one location. Data needed during the computations may be obtained from still other machines, or possibly from large databases maintained by an outside organisation. With such an arrangement the processes could run in parallel most of the time, only being synchronised and exchanging messages and data at certain time points. The machines running the more computationally intensive processes could of course be themselves highly parallel internally. The AECL system of INTARES/7/ is designed to run on a distributed processing system of the sort described, motivated by the consideration that if to solve a problem one needs codes X, Y and Z then the right versions of the codes to use are those resident in the home systems of the organisations which develop and maintain them. The INTARES approach is certainly more futuristic than that of ESTER, but has certain disadvantages in the current state of technology. Communication over networks can be slow and unreliable, strict standards for data and message interchange must be imposed on all the codes and machines in the distibuted system, and there is a general difficulty in coordinating all the processes so as to ensure both numerical stability and efficiency. The choice made in ESTER of a central database in the memory of a single computer restricts its immediate adaptation to distributed systems to those with a shared memory architecture. If the memory is distributed parts of the central database will need to be copied to the processes and machines which need them, and the updated versions copied back when the process is complete. ESTER has facilities for copying segments of the database, but other control facilities would also be required to make ESTER work on a distributed memory system. Design decisions which prevent this route being taken in the future have been carefully avoided, however. Networking of computers is an easier problem. One machine could provide preprocessing e.g. in the user’s office, another perhaps more powerful one could perform the calculations, and a third with better display facilities could postprocess the results. This kind of arrangement is now routine under UNIX with machines of the same type, using remote login and file transfer facilities. Between different machines one may still encounter surprises, but standardisation is rapidly eliminating them. For the more general configuration in which the data for a problem are distributed over several machines (and not merely copied from one machine to another) techniques such as remote procedure calls are necessary to allow the driver to invoke procedures on other machines and to pass them the necessary data. ESTER can be extended to exploit such advanced techniques when they become generally available. 9. New Contributions to ESTER The shared-cost action programme for 1989 included contracts for the development of ESTER, as described above. It also included the CEC’s participation in the development of VICTORIA, a code from the USNRC which was originally part of ME LPROG. VICTORIA calculates fission product release from the fuel and control rods, and its transport in dry conditions within the circuit as aerosols or vapours. The code models the aerosol transport and deposition mechanisms to be found e.g. in RAFT, but places much more emphasis on the chemistry. Two hundred chemical species or more can be considered in the equilibrium calculations, and the aerosols are treated as multicomponent i.e. aerosols of different chemical composition may have different size distributions. VICTORIA development for the CEC is undertaken by the UKAEA, and is part of a wider programme of development by laboratories in the US and Canada. All the developments will lead to a much improved version of the code which will be incorporated in ESTER. ESTER will then be able to calculate release from the bundle, thermalhydraulics in the bundle and circuit, and fission product transport and chemistry in the circuit all in the one code system, a capability not previously available in Europe. Adaptation and improvement of codes and modules for ESTER was also included in the 1990 SCA programme. Two successful contractors were IKE, which will adapt significant parts of the KESS core degradation package to ESTER, and the UKAEA, which will further develop the modelling of vapour-aerosol interaction using results from the FALCON and ACE experimental programmes, and incorporate the models in VICTORIA and in the containment iodine chemistry code INSPECT. INSPECT will then be brought into the ESTER framework. A third contractor now at work on adapting codes to ESTER is CEA, which is modernising and converting to ESTER standards the containment thermal-hydraulics code JERICHO, the code AEROSOLS-B2 which calculates aerosol physics in the containment, and the iodine chemistry code IODE. IODE is simpler and more pragmatic in spirit than INSPECT, so there is no duplication in having them both in ESTER. Indeed, ESTER should make it easier to compare the predictions of the two approaches for specific problems such as Phebus tests. A last SCA contract related to ESTER now in progress is that with Vincotte, the Belgian firm of consultants. Vincotte is responsible for developing tools and, more importantly, a knowledge base for the on-line monitoring of the contents of the central ESTER database. This task is a recognition of the fact that the modules which will be incorporated in ESTER can never be perfect. There always exists a risk that in a given problem some module when called upon to operate will pick up from the database data deposited by a previous module which are outside its “area of competence” i.e. require it to calculate situations for which it is not expected to be reliable. A simple example would be data which cause a correlation in a module to be used outside its range of validity. Vincotte will need to acquire a thorough knowledge both of the ESTER system and of the modules which go into it, and is thus well placed to perform its secondary task, which is to provide quality assurance

206

THE PHEBUS FISSION PRODUCT PROJECT

consultancy on the ESTER system as a whole. The SCA programme includes other model developments not tied to particular codes. The JRC in collaboration with the relevant contractors intends to integrate the most appropriate improvements into ESTER modules directly. These could include chemical data, chemical ki-netics models, thermal resuspension models, and condensation and nucleation models. The SCA programme for 1991 (the last) again includes sections on model development and code incorporation into ESTER, and it is hoped that the response will be at least as extensive as last year. 10. ESTER Development and Validation Plans The ESTER development plan in the short term is as follows: in July the contractors should deliver to the JRC a preliminary version of the code consisting of the framework, tools and services, controller, and integrated versions of the modules ICARE-2 v2 and FPRATE. The contractors will undertake optimisation studies to minimise storage and timing overheads, improve the integration strategy and of course eliminate bugs. For its part the JRC, assisted by selected European laboratories, will test the portability of ESTER and report user reactions and any errors discovered. A final version of ESTER v0 will be released by the contractors together with all documentation at the end of 1991. Shortly afterwards DEIMOS for the circuit thermalhydraulics, the aerosols/chemistry code VICTORIA, and the CEA containment codes JERICHO, AEROSOLS-B2 and IODE will be incorporated, as will the VINCOTTE data consistency checker. The JRC may insert models for chemical kinetics and for nucleation developed under SCA, and possibly a special-purpose model for the difficult zone just above the bundle in Phebus. A certain period of consolidation will then be necessary to eliminate errors, improve the user interface and the controller etc., and ESTER v1 containing all the codes and models mentioned above should become available at the end of 1992. Fig. 4 shows the expected configuration at this time. Beyond that date certain further extensions are clear, while others must await the outcome of existing or planned SCAs. A number of KESS modules will become available, which should supplement the models of ICARE. The INSPECT iodine chemistry module is expected at this time also. Since the main application of ESTER in its early stages is to Phebus the code will need a pool scrubbing module; plans are in hand to supply this, probably the BUSCA code now being improved under an SCA. It is hoped that the ESTER chapter of the current call for SCAs will attract offers of a systems thermal-hydraulics code such as CATHARE or ATHLET (which will not be fully integrated unless considerable effort becomes available), of an advanced containment code such as FIPLOC, and of a core-concrete interaction code such as WECHSL. The selection of suitable codes or modules for incorporation in ESTER is essential for the quality of the system as a whole; it must take account of the standard of coding and documentation, the effort being put into continued maintenance and development, and the validation status of the code. Such selection will continue in the future. An additional “Darwinian” factor tending to improve quality is that if more than one module with similar functions is included in ESTER it becomes simple for users to compare them and see which performs best on given problems. Unreliable or over-simplistic modules will tend to be rejected by users and so become “extinct”. The continued maintenance and development of ESTER cannot be performed by contractors. The JRC is committed to establishing ESTER as the natural framework for code integration and for future code development in Europe, and will devote effort on its own behalf to the maintenance of ESTER, eliminating defects and extending facilities. ESTER validation is clearly essential; the individual modules are assumed to be validated by their developers or in international programmes such as LACE or ACE and Phebus, but the possibility of unforeseen interaction between them cannot be ruled out, and ESTER must be put through a series of qualification tests whenever a new module is integrated. This kind of work too demands a commitment from the JRC. A third type of work which needs JRC effort is the incorporation of improvements to individual modules in the adapted versions which will be present in ESTER. It is hoped that in the long term developers will naturally tend to prefer the ESTER version for further improvements because of the additional facilities to which it has access, and hence that this problem of updating will not grow unmanageable as the number of modules incorporated continues to increase. Validation of ESTER against Phebus tests is naturally included in the JRC source term R&D plan for 1992–94. There is a possibility that the further development and application of ESTER by the CEC will not have to rely entirely on the JRC’s own resources once the SCA programme is terminated. A new programme of “reinforced concerted actions” (RCAs) is now being launched by DG XII in Brussels with the object of favouring collaboration in severe accident studies related to new reactor designs. ESTER is highly suitable as a tool for facilitating collaboration on severe accident code development, and hence some support from the RCAs may be possible. Development beyond 1994 or so becomes harder to map out. It seems likely that as the Phebus programme matures ESTER will become more oriented to reactor problems. The ability to use directly the databases of geometrical and other specifications for real plants built up for other codes e.g. system codes, would be a real plus, since these often represent a considerable investment on the part of their compilers. ESTER should also become able to handle accident management

ESTER—A EUROPEAN SOURCE TERM EVALUATION SYSTEM

207

operations, more engineered safety features, and possibly phenomena associated with new reactor designs or with reactors of older design which may be operating in an extended Europe. 11. The Availability of ESTER The Commission of the European Communities has the right to distribute ESTER within the EC. The JRC intends to make the system freely available on condition that recipients join a Users’ Club. Membership of the club will carry with it an obligation to inform the JRC of any errors discovered and of any improvements made, and to report the results of any calculations which could help validate the system. In return members will receive not only the system but also updates and corrections or extensions to the documentation. Property rights to the individual modules of course remain vested in their developers, but a condition of acceptance for the incorporation of a module in ESTER is that is should be available for distribution by the CEC within ESTER to members of the Users’ Club, and this condition applies even to VICTORIA, which was not developed in Europe. This arrangement has a certain advantage for the code developer, as well as for the CEC and for Europe at large. A module inserted in ESTER is likely to be tested on a wider range of problems and by a wider clientele than the same module standing alone. The developer will naturally have access to the resulting error reports and other feedback regarding his module. If these lead him to improve the module and so improve ESTER the virtuous cycle will be complete. Although ESTER will be free in Europe, its use outside the EC will require a licensing agreement reflecting the effort which is being put into its development. Some preliminary enquiries about such agreements have already been received. In conclusion, ESTER has good prospects of fulfilling the CEC’s hopes of creating a framework which will foster collaboration and synergy between groups developing and validating codes and models in the source term area. If the present spirit of cooperation is maintained it should rapidly grow into the natural environment for future development efforts, to the benefit of severe accident studies in Europe as a whole Acknowledgements The authors would like to acknowledge critical comments and suggestions from F.Jacq of CISI, F.Schmidt of IKE, and P.Fasoli-Stella of the JRC, which did much to improve the paper. References 1 2

3 4

5 6 7

// Markovina, A., Fasoli-Stella,P., Mailliat,A. Analytical Assessment of the Capability of a Scaled-down In-pile Facility to Simulate PWR Phenomena under Severe Accident Conditions. AAR ‘89, 8th Annual Meeting, Reno, NV (USA). //Jones, A.V., Bonanni, E., Markovina, A. Principal Results of the “Phase B” Verification Studies in Support of the Phebus-FP Project. CSNI Workshop on Aerosol Behaviour and Thermal-hydraulics in the Containment. Fontenay-aux-Roses (F), November 1990. // Ruehle, R. RSYST as a Software Framework for a Unitary System for Computer-aided Aerospace Analysis and Design, (in German); Rechenzentrum der Universitaet Stuttgart und Lehrstuhl fuer Anwendungen der Informatik im Maschinenwesen. (1988) // Chatelard, P., Gonzalez, R., Jacq, F., Porracchia, A. ICARE: A Computer Code for Severe Fuel damage Analysis. General Description of the Second Version ICARE2. IAEA Technical Committee Workshop on Computer-aided Safety Analysis. Berlin, April 1989. // Hocke, K-D., Bisanz, R., Schmidt, F., Unger, H. Fission Product Release and Transport Modeling in KESS-2. KfK 3800/3. 5th Int Mtg. on Thermal Nuclear Reactor Safety. Karlsruhe, Sept. 1984. // Biasi, L.DEIMOS—A Fast-Running Circuit Thermal-hydraulics Model. Private Communication (January 1991) // McDonald, B.H., Wallace, D.J., Hanna, B.N., Dormuth, D.W. INTARES—A Code for Integrated Thermal-hydraulics and Aerosol Safety Analysis. CSNI Workshop on Aerosol Behaviour and Thermal-hydraulics in the Containment. Fontenay-aux-Roses (F), November 1990.

FPT0 TEST PRECALCULATIONS A.MAILLIAT*, A.JONES** and I.SHEPHERD** *Institut de Protection et de Sécurité Nucléaire, DRS/SEMAR, France **Joint Research Center, ISPRA, Italy

1. TEST PREPARATION ORGANISATION 1.1 Background to the First Test In order to understand better the phenomenology of severe accidents the CEA/Safety Institute and the Commission of European Communities have sponsored study of reactor severe accident sequence calculations in Europe through the CEC shared cost actions programme. Each organisation from five national teams was requested to assess the accident sequences which present the highest interest from the point of view of the risk to the population and of the relevance of the phenomena involved. From the results a synthesis was made [1] which identified the most important aspects of such accidents; these phenomena being the targets that the PHEBUS FP facility has to reach. The behaviour during core degradation (temperature heat up rates, gas mass flow rates…); the associated fission product and aerosol release rates; the reactor components where the fission product trapping is the largest; the main depletion mechanisms and their driving parameters. Having defined the phenomenology to investigate, the verification of the experimental circuits devoted to reproducing such a phenomenology was started. It was done by the teams previously involved in the reactor sequences calculations using preliminary lay-outs of the experimental circuits derived from dimensional analyses and boundary conditions deduced from the synthesis of the reactor results. From this second step it was possible to derive important information on various dimensioning problems and to adjust the geometrical dimensions of some parts of the circuit [2, 3]. As soon as the severe accident phenomenology was known, the European experts were asked to indicate the most interesting tests that should be performed. From this advice a provisional test matrix was drawn up both by CEA/IPSN and JRC/ISPRA and submitted for expert agreement. This test matrix is being periodically revised in order to integrate new knowledge. 1.2 Preparation of the First Test The tests of the PHEBUS programme are prepared by the Scientific Analysis Working Group (SAWG). Two groups of analysts, at CEA/Cadarache and at JRC/ISPRA propose for the SAWG approval an experimental protocol for each test. These proposals are based on joint calculations of tentative test protocols both by CEA and JRC teams and PHEBUS programme partners. The test preparation starts three years before the test itself and is initiated by the Experiment Objective Specifications (EOS) issued by SAWG. This document describes the test in terms of a fission product flow path and the reactor components to be simulated. During a one year period both the CEA and JRC analysts and PHEBUS partners must define the geometrical characteristics and the operating specifications of the experimental components. The results of this work, after SAWG agreement, is included in an Experiment Geometry Specifications (EGS) which is released 15 months before the test. This document includes a summary of the dimensioning work with identification of the scaling processes and the limits of representativeness inherent to scaling or to the technological constraints. It includes also a discussion and a finalization of the test objectives according to the results of the dimensioning work. A second period of nine months is devoted to boundary and initial condition adjustments of each component of the experimental circuit in order to reach the objectives assigned to those components. At the end of this so called exploratory

FPT0 TEST PRECALCULATIONS

209

calculation period, a set of provisional test conditions based on independent behaviour of the various components is made available and released as the Provisional Test Protocol (PTS). Finally a third period of 12 months devoted to sensitivity calculations of the experimental circuit with coupled components including a three month synthesis is required to finalize the test protocol which is released as the Final Test Protocol (FTP). This self-standing document is issued three months before the test itself and must contain all the information needed to carry out the experiment. It is known that modelling severe accidents is presently a rather imprecise art. It is therefore important to have as many calculation as possible with as many different codes and different users as possible and have a critical analysis of the results. By this way different opinions on the anticipated course of the transient are collected and a better appreciation of the uncertainties obtained. 1.3 Objectives of the First Test The first test FPT0 is devoted to study of the phenomenology of severe accident sequences for which the fission product flow path involves the primary side of the steam generator and the reactor containment building. The objectives of FPT0 may be stated separately for the bundle, the circuit and the containment. For the bundle, the first priority is to maximize the fission product release and the fuel rod degradation. A large degradation will be obtained by reaching the melting temperature of the fuel so enhancing the interaction between the fuel pellets and the molten cladding. It is required that the fission product emission takes place in a low pressure oxidizing gas steam with a fission product concentration inside the carrier gas representative of the reactor core. There should be boric acid addition to the carrier gas. Since the test will also provide information on rod bursting, the control rod internal pressure will be similar to that of in the reactor and the fuel rod pressure adjusted to be representative of an irradiated BR3 rod, which will be used in subsequent tests, under the same conditions. The preliminary boundary condition set for the bundle includes an initial heat up with fuel rod burst and melting of the control rod. There follows a temperature escalation period from the melting temperature of the zircaloy up to the zirconia melting devoted to fission product extraction and fuel dissolution by molten zircaloy. This second period is followed by an increase of the bundle temperature up to the melting temperature of the fuel to produce a large degradation of the bundle. Finally a rapid cooldown period with steam flow ends the test since previous tests e.g in PBF have shown surprisingly high amounts of fission product and aerosol release during cooldown. Regarding the circuit, the main objective is to investigate fission product depletion inside a primary side of a steam generator at low pressure (around 0.2 MPa) in hot conditions. Secondary objectives are to explore phenomena in the section just above the bundle exit where sharp changes of the carrier gas temperature are expected and to provide data on fission product chemistry including interactions of these fission product with pipe wall, under low pressure and high temperature conditions. Preliminary boundary conditions involves a constant temperature of 700°C for the pipe internal liner from the bundle down to the steam generator entrance to minimize condensation onto the pipe wall of the main fission product species like CsOH, CsI and Te. At the beginning of the test, the steam generator secondary side is maintained at 150°C to avoid steam condensation, fission product vapours condense both onto the steam generator pipe wall and the particles. Aerosol depletion through thermophoresis will be explored. For the containment the objective is to study the fission product chemistry and especially iodine radiochemistry in the sump water and atmosphere and the effect of paints in the ‘dirty’ chemical conditions of a reactor accident. Tentative boundary conditions for the containment involve two successive phases. During the first one the containment has an atmosphere temperature around 80°C, a humidity ratio near 100% and steam condensation onto the condenser surfaces. Incoming aerosols enter the sump water through settling and diffusiophoresis. At the beginning of the containment second period the bundle and circuit transient are ended and the containment is disconnected from these components. The second period duration is three days with an atmosphere temperature increase up to 150°C and superheated conditions. This period is devoted to iodine radiochemistry. This FPT0 test being the first test of the PHEBUS programme there are naturally some technological objectives also: to check out the instrumentation and its performance, to gain experience in operating the PHEBUS facility with a specified test procedure (including the pre-irradiation and the post test operations like e.g. decontamination).

210

THE PHEBUS FISSION PRODUCT PROJECT

2. BUNDLE STUDIES The first point to note about the Phebus-FP bundle is that it has many features in common with the Phebus-SFD bundle. There have therefore been many calculations performed on it that have been checked against experiment. An international standard problem is currently underway based on one of these experiments. The main differences are the power profile which is sharper for Phebus-fission product and the inner liner of the shroud which is made of zirconia in Phebus-FP not zircaloy. There is a major difference in the objectives of the two sets of experiments. The SFD experiments aimed to investigate the core degradation main processes for a given set of thermalhydraulic parameters; mostly flow rate and composition and temperatures. Phebus-fission product, on the other hand, aims to observe the fission product release during core degradation with representative conditions of severe accidents. Phebus FP is a harder problem to solve for the codes. The highest temperatures reached in the SFD series was in the order of 2700K whereas the target in FPT0 is above 3000K. The aim is to have up to 20% of the uranium dioxide melted. An added difficulty is that the oxidation reaction will be allowed, for the first test anyway, to proceed in excess steam. During the SFD series it was often limited by injecting helium. It is thus clear that the calculations for the bundle will not be able to predict the precise degree of core degradation. We hope to be able to show with the calculations that, even with this uncertainty, we can develop a test procedure that stands a good chance of achieving the objectives. In addition to the base case calculations we can obtain an estimate for the uncertainty by performing the same calculation with many different codes and different users. The results are analyzed critically. In this section we will describe the principal findings of these calculations. 2.1 Information from Shared Cost Action 2/B This was called the ‘Phase B’ exercise because it followed the ‘Phase A’ exercise. Phase A was a set of reactor sequence calculations. Phase B was a set of scaled down sequences in the various Phebus geometries with the object of simulating as closely as possible the Phase A phenomena. The assumed scaling factor at this time was 1:2000, so the flow rates were scaled to this factor to preserve chemical concentrations and a power specified that would achieve the desired temperatures. This meant a flow rate of 5 grams per second, dropping to 0.5 grams per second and a system pressure of 0.2 or 3.5 MPa according to the simulated sequence-[2, 3]. Nine organizations from six countries took part in the exercise using the codes ICARE-1, ATHLET-SA, KESS, MARCH3, BUTRAN and RELAP-SCDAP. The results showed reasonable agreement up to the time of rapid oxidation but not afterwards. Because the clad oxidation took place towards the beginning of the transient when the inlet flow rate was high it was never steam starved. About half the codes were capable of calculating core degradation and those that did found considerable movement of molten material and recommended that a core catcher be used in the actual experiment. The sharp axial power profile and heat losses to the shroud meant that the top and bottom of the bundle were significantly colder than the middle section. One participant warned that this could lead to a condensation of some fission products that would not be expected in the reactor case. Sensitivity studies showed that the system pressure was a relatively unimportant parameter for the bundle. The velocities of course changed but the phenomena in the bundle are more closely related to the flow rate, which was unchanged, than the velocity. 2.2 Dimensioning Verifications The objective of this first step of the test preparation is to obtain quantified estimations of the various processes which take place during the test and, from these, to check the geometrical characteristics of the test components. This set of calculations was the first attempt to perform a complete bundle to containment calculation. However, for the bundle, the feedbacks from the rest of the circuit are not very important. The rising pressure in the containment would cause a similar rise in the pressure in the bundle but nobody modelled this effect and, as we have seen, the overall sensitivity to the system pressure is low. The flow rate was lower than in the Phase B case because the fission product inventory was 5000 times less than in the reactor and not 2000 as had been supposed before. The flow started at 2 grams per second and fell to 0.2 at the end of the transient. Some hydrogen was injected at the end of the transient in order to have a representative composition of the carrier gas for chemistry [4].

FPT0 TEST PRECALCULATIONS

211

Fig. 1. Inlet flow rates (kgmoles/s)

Fig. 2. Bundle power (w)

These flow rates are shown in Figure 1 in kgmoles per second. The nuclear power supplied is shown in Figure 2. There were six participants to the exercise using six different codes (Table 1). The codes ICARE-1, BUTRAN and ATHLET-SA do not calculate the movement of material whereas the others do. The principal difference between the two KESS calculations was that IKEbase assumed that the αzirconium melts at 2250K and collapses whereas in IKEmelt it was assumed that the zirconium melts at the higher temperature of 2670K and immediately forms a eutectic with the uranium dioxide, the proportion of uranium dioxide to zirconium being a constant. The oxidation laws used were also somewhat different. Table 1 ORGANIZATION

CODE

CODEWORD FOR GRAPHS

Centro de Investigaciones Energiticas, Madioambienatales y Tecnologicas (CIEMAT), SPAIN Gesellschaft für Reaktorisiccherheit (GRS), München, GERMANY Electric Power Research Institute, Palo Alto (USA) Joint Research Centre, Ispra, ITALY Joint Research Centre, Ispra, ITALY

ICARE-1

CIEMAT

ATHLET-SA

GRS

CORMLT

EPRI

ICARE-2 BUTRAN

JRC BUTRAN

212

THE PHEBUS FISSION PRODUCT PROJECT

ORGANIZATION

CODE

CODEWORD FOR GRAPHS

University of Stuttgart, GERMANY University of Stuttegart, GERMANY CEA, CADARACHE, FRANCE

KESS KESS ICARE-2

IKEbase IKEmelt CEA

Fig. 3. Peak fuel temperature (K)

Fig. 4. Outlet vapour temperature (K)

Most codes modelled four representative pins and one control rod although KESS and ATHLET-SA modelled only one pin and no control rod. Only EPRI modelled more than one fuel channel and all codes used 10 or 11 axial nodes. Probably one fuel pin and one control rod would have been sufficient. The temperature was fairly uniform across the bundle. A subsequent sensitivity study with ICARE-2 showed that it was not necessary to model more than one fuel channel although, when there is a large blockage, the multi channel model has limitations. The peak fuel temperature is shown in Figure 3. The timing of the peak temperature during the runaway oxidation is rather uncertain but otherwise agreement is reasonable. The same can not be said for the outlet vapour temperature as shown in Figure 4. This is an important parameter because if it is too cold then too many gaseous fission products will condense and if it is too hot we risk burning out the heaters installed on the vertical line above the bundle. The mole fraction of hydrogen at the outlet is shown in Figure 5 and the percentage of cladding oxidized in Figure 6. These confirm that the coolant was never pure hydrogen and that the oxidation was extensive. The previous results, even if they present large variations in the detail of some parameters, give a coherent picture of the average thermal behaviour of the bundle and demonstrate the ability of this one to reach melting of UO2 with representative thermalhydraulic conditions. A more detailed examination of the results of this exercise will be found with [4].

FPT0 TEST PRECALCULATIONS

213

Fig. 5. Hydrogen mole fraction at outlet

Fig. 6. Percentage of cladding that has oxidized

2.3 Exploratory Calculations The second phase of the test preparation was devoted to boundary and initial condition adjustment of each component of the experimental circuit in order to define a preliminary test protocol. This test preparation phase involves what it is called exploratory calculations. It means a process of iterating calculations which must converge onto a satisfactory set of boundary conditions. A crude set of conditions deduced from the dimensioning phase are refined progressively in order to satisfy the test objective requests. If some of them are conflicting, then a part of those objectives would have to be redefined by SAWG. Such a situation was encountered for the bundle. Highlights about these calculations which are always under way at the time of this paper are given hereafter. In an attempt to achieve the objectives of the test a set of boundary conditions was proposed. In order to arrive at clad melting before the oxidation layer on the cladding was too thick, a faster power ramp was used. The power and flow boundary conditions are shown in Figure 7. This transient also featured a long steady state period with extensive clad degradation. This was because it had been realized that, with fresh fuel, many of the fission products would only be released after fuel melting. Details of the calculation results will be found in [5]. There were four participants with three different codes (Table 2). The main difference between the two ICARE versions is that version 1 does not allow oxidation of molten zirconium whereas version 2 does. The CIEMAT1 calculation allowed degradation of the control rod but no release. It is the other way round with CIEMAT2.

214

THE PHEBUS FISSION PRODUCT PROJECT

Fig. 7. Power and flow boundary conditions

Fig. 8. Fuel temperature half way up the bundle Table 2 ORGANIZATION

CODE

CODEWORD USED IN GRAPHS

Centre de Investigaciones Energeticas, Medioambientales y Tecnologicas (CIEMAT), SPAIN Centre de Investigaciones Energetica, Medioambientales y Tecnologicas, (CIEMAT) SPAIN Joint Research Centre, ISPRA, ITALY & CEA CADARACHE, FRANCE Joint Research Centre, ISPRA, ITALY & CEA CADARACHE, FRANCE University of Stuttgart, GERMANY

ICARE-2V1

CIEMAT 1

ICARE-2V1

CIEMAT 2

ICARE-2V1 ICARE-2V2 KESS

ICARE 2V1 ICARE 2V2 KESS

The fuel temperature halfway up the bundle in Figure 8 shows us immediately that KESS is the odd man out. In fact KESS predicts that the fuel melts at rather a low temperature, and falls out of the hot zone, lowering the total power. The percentage of cladding oxidized is shown in Figure 9 where the effect of ICARE-2V1’s neglect of the oxidation of molten zirconium can be clearly seen. The percentage of uranium dioxide that is melted or dissolved is shown in Figure 10. Despite the large differences in fuel temperatures, the amount of melted material is similar. This is partly as a result of the safety effect of the sharp power profile. It is difficult to melt the material at the top and bottom of the bundle. The flow area reduction inside the bundle according to material relocation is shown in Figure 11 at 5000 seconds as a function of axial position.

FPT0 TEST PRECALCULATIONS

215

Fig. 9. Percentage of cladding oxidized

Fig. 10. Percentage of liquid UO2

Although this proposed scenario succeeded in producing a large amount of melted or dissolved fuel and almost certainly a large fission product release which were the two main objectives, it was not judged successful because the condition that the coolant be always oxidizing was not satisfied. Originally the objectives for this test had included a low clad oxidation. This was shown to be impossible without steam starvation and so this objective was dropped. A new scenario was proposed and a new set of calculations is presently being performed. Preliminary results for the bundle are summarized below. Both the power driven to the bundle and the power generated by the zirconium oxidation are given in Figure 12. It can be observed that this second source of energy is rather small indicating that the thermal state of the bundle can be controlled by adjustment of the power driven to the fuel rods which is therefore one of the main boundary conditions. The power variations had been determined by imposing a predefined temperature history of the central rods (see Figure 13) and steam inlet mass flow rate; the code ICARE2 having the ability to compute the required power injection. The steam flow to the bundle was adjusted in order-to have oxidizing condition throughout the transient and a correct temperature for the gases exiting the bundle see Figure 14.

216

THE PHEBUS FISSION PRODUCT PROJECT

Fig. 11. Flow area as a function of height up the bundle

Fig. 12. Powers in the bundle

The preliminary calculations show a reasonable outlet temperature that was not so cold that fission product vapours condense or hot enough to damage the heating elements above the bundle. Figure 15 demonstrates that during the main part of Cs, I, Te releases the previous constraints are satisfied; the exit temperature ranging between 1300 and 1600K. The same Figure gives the CORSOR code estimation of the fraction of the bundle inventory which is released during the present transient. This estimation will have to be carefully considered as this code is known to be unrepresentative of low burn up fuel, which is the case for the FPT0 test, because of low interconnectivity between fuel grains. Regarding the degradation of the bundle, Figure 16 presents a summary of the UO2 during the transient. After 2000 seconds melting of zirconium is reached in the central part of the bundle. Only a small amount (2%) of fuel is dissolved by the molten zirconium according to the preferential steam oxidation of this material. After 5000 seconds fuel melting starts and, after 7000 seconds when the cooldown phase is initiated, approximately 11% of the total fuel mass in the bundle is dissolved or melted. A global picture of the geometrical state of the bundle degradation after 9000 seconds is given in Figure 18, the initial geometry being that depicted in Figure 17.

FPT0 TEST PRECALCULATIONS

217

Fig. 13. Temperature half way up

Fig. 14. Flow IN/OUT the bundle

2.4 Main Problems Encountered A precise knowledge of the shroud material conductivity is of an extreme importance for having a good prediction of the thermal behaviour of the bundle. In fact the fraction of the injected power which is carried outside of the bundle through convective heat transfer is rather small once the temperature exceeds 1500°C. Therefore the main part of the power is transferred to the test train coolant flow. First by radiating heat transfer from the fuel rods to the shroud then by conduction from the shroud to the test train coolant flow. Because the mass flow rate of this is high its temperature is practically constant and therefore the temperature level in the bundle is imposed by the injected power and the shroud conductivity. This conductivity needs to be better known for temperatures higher than 2000°C. Fission product and aerosol release rates during the heat up phase of the bundle are presently computed with the CORSOR code which is known as poorly verified for low burn up fuel.

218

THE PHEBUS FISSION PRODUCT PROJECT

Fig. 15. Outlet temperature and FP releases

Fig. 16. Fraction of liquefied or molten UO2

Furthermore, it is thought that the cooling phase will be the occasion of an enhanced release period for fission products and aerosols because of fracturing of the pellets. Up to now there are no tools available to estimate these release rates. Therefore the estimated amounts of deposited materials along the circuit or arriving in the PHEBUS containment have to be regarded as indicative rather than precise figures. In addition it is clear that the present status of codes does not allow us to predict the precise degree of bundle material relocation during the degradation process. 3. CIRCUIT STUDIES By circuit we mean from the bundle exit to the containment entrance and, so far, these calculations have been performed in two phases; first the thermalhydraulics then the fission product transport. In a reactor the decay heat of deposited aerosols would

FPT0 TEST PRECALCULATIONS

219

Fig. 17. Initial geometry

be expected to heat the pipe walls and thus feedback to the thermalhydraulics but this effect will be negligible in Phebus. It is therefore-reasonable to perform the calculations uncoupled. In Phebus, unlike in a reactor, most of the pipe walls are heated to a fixed temperature so the thermalhydraulic codes do not have to calculate them. The first exception is the part of the vertical line just above the bundle exit which is expected to be at such a high temperature that heater elements placed there would burn out. The second exception is the steam generator entrance where the fluid temperature falls from 700°C down to 150°C on a length which varies with its mass flow rate. Clearly having the wall temperatures defined makes the job easier for the thermalhydraulics codes. Furthermore, in FPT0, the coolant flow will always be single phase vapour. Originally, as we shall see, it had been proposed to allow condensation in the steam generator but this scenario has now been postponed to a later test. As with the bundle calculations it has been a policy to perform calculations with as many codes and as many users as possible and to have a critical analysis of the results. Some of them will be presented hereafter. 3.1 Information from Shared Cost Action 2/B In the Phase B exercise four different cases related to the conditions of; TMLB’, AB, S2D and Bypass sequences were calculated. Although none of these correspond precisely to the FPT0 test protocol which had not defined at that time, some of the conclusions are still valid. There were six participants, from four countries, to the exercise. For the thermalhydraulics TRAC, ATHLET-SA and RELAP5 were used. Everybody used TRAPMELT for the fission product transport except for UKAEA who used VICTORIA. Although VICTORIA has the capability of calculating chemical equilibria this option was not used. The disagreement in results in the thermalhydraulics was unexpected and probably caused by the difficulty of imposing Phebus boundary conditions on codes that were designed for reactors. The uncertainties in fission product transport modelling, on the other hand, are known to be large so differences in results were less surprising. Nonetheless it was possible to discern some trends.

220

THE PHEBUS FISSION PRODUCT PROJECT

Fig. 18 Bundle geometry after 9000s

The vertical line just above the bundle exit in Phebus FP is shorter and colder than would be expected in a reactor. The dominant retention mechanisms were therefore thermophoresis and wall condensation whereas gravitational settling would dominate in a reactor. Therefore it was decided to abandon the idea of representing the reactor upper plenum in PHEBUS. Certain pipes in the circuit do not represent any reactor component but only transport the coolant from one component to another with the objective to be as ‘neutral’ as possible at least with regard to aerosol depletion. The calculations showed that the proposed geometry should be changed because significant amounts of fission products were deposited on them. Regarding the aerosol retention in the steam generator, Phase B calculation results have demonstrated that trapping is mainly located along the first metres of this component and half height mock-up is sufficient in the PHEBUS facility. All the results of this exercise have been summarized in [6]. 3.2 Dimensioning Verifications The inlet conditions for these calculations were taken from the results for the bundle. As has been mentioned a scenario was proposed whereby 3000 seconds into the transient the steam generator secondary side was cooled down in order to provoke condensation of steam on the primary side. The participants to this exercise are summarized in Table 3. The difference between the two CEA calculations is that CEAbase used a standard model of TRAPMELT whereas CEAbase used a version with a preliminary version of a diffusiophoresis model. Generally primary circuit aerosol codes, unlike those that are used in the containment, do not model diffusiophoresis. During the ‘cold’ steam generator phase this is expected to be a significant and dominant mechanism. None of the other codes modelled it. CEA were also the only group to include the feedback effect of the rising pressure in the containment.

FPT0 TEST PRECALCULATIONS

221

Fig. 19. Temperature of gases at the bundle outlet Table 3 ORGANIZATION

CODE FORGRAPHICS THERMALHYDRAULICCODE DEPOSITIONCODE

Centre de Investigaciones energeticas, Mediaambientales Y Tecnologicas (CIEMAT), SAPIN Joint Research centre, ISPRA, ITALY CEA, CADARACHE, FRANCE CEA, CADARACHE, FRANDE Electric Power Research Institute, PALO ALTO, USA

CIEMAT

TRAC

TRAPMELT

JRC

MERGE (modified for PHEBUS boundary conditions) CATHARE 2 CATHARE 2 PSAAC

TRAPMELT

CEAbase CEAmodif EPRI

TRAPMELT TRAPMELT RAFT (version received from Argonne-January 1990)

Generally the thermalhydraulics codes managed to predict the fixed wall temperature parts of the circuit better this time. The two areas of difficulty were the unheated part of the vertical line and the ‘cold’ steam generator. The coolant temperature in the section above the bundle is shown in Figure 19. Part of the disagreement was due to differences in the outlet temperature of the bundle but the modelling of the heat transfer in this region played its part. It was found essential to nodalize this region rather finely and to take into account the radiative heat transfer. This is shown in Figure 20 which compares the wall temperature profile both with and without the radiative heat transfer modelled 2000 seconds into the transient. Table 4

Aerosol CIEMAT EPRI JRC Tellurium CIEMAT JRC Caesium Iodine CIEMAT JRC Casesium Hydroxide CIEMAT JRC B C L P

UP 1

UP 2

HL

SG 1

SG 2

CL

Th Th Th/L

Th Th/L

S Th S/L

Th/Tu Th Th/L

Tu Th

S/Tu S/Tu S/L

C C

C C

C C

Wc/C

C

Th

Th

R Wc/C

Wc

Wc/C

Wc/C

Th C

Th C

C

Turbulent deposition in bends Chemical absorption Laminar deposition Condensation on particles

S Th Tu Wc

Gravitational settling Thermopheresis Turbulent impaction Wall condensation

222

THE PHEBUS FISSION PRODUCT PROJECT

UP 1 R

UP 2

HL

SG 1

SG 2

CL

Revaporisation

Fig. 20. Wall temperature profile in first part of the vertical line

During this exercise it was realized that none of the major thermalhydraulic codes had adequate models for wall condensation in the presence of hydrogen. This was one factor in the postponement of a cold steam generator till a later experiment. It is thought that advantage will be taken of such a delay to develop adequate models and to check them on a PHEBUS scale 1 companion facility. The deposition of aerosols after 4850 seconds is shown in Figure 21. The retention is given as a percentage of the aerosols released by the bundle. It can be seen that most of them arrive in the containment if the cold phase of the steam generator is disregarded. The difference between the CEAmodif and CEAbase results indicates the importance of modelling diffusiophoresis in the cold steam generator. The main retention mechanisms are shown in Table 4. All the results for this dimensioning verification exercise are summarized in [4]. 3.3 Exploratory Calculations In the same way as for the bundle calculations, the analysts of CEA, JRC and PHEBUS partners are involved in the exploratory phase of the test preparation in order to define a preliminary test protocol for the circuit. Exploratory calculation exercise for the fission product behaviour is at present under way with circuit input conditions, i.e. carrier gas and FP and aerosol mass flow rates from the bundle calculated by CEA using ICARE2V2 with CORSOR, and temperatures in the circuit deduced by CEA using CATHARE2. VICTORIA, RAFT, TRAP-FRANCE and MACRES predictions have been received. Some of these calculate chemical equilibrium as well as aerosol transport. Because of the bundle objective adjustment during the exploratory phase and the delay imposed for recomputing the bundle transient, most of the previous contributions not yet analysed. Nonetheless some of the existing circuit calculation results will be presented here after. For the time being these results have to be regarded as tentative

FPT0 TEST PRECALCULATIONS

223

Fig. 21. Aerosol deposition at 4850 seconds

Fig.22. Wall temperature before entering the heated sections

test protocol conditions. The SAWG on the bases of all the future contributions will have to decide of a finalized test condition set. 3.3.1 Thermalhydraulics of the circuit As has been said previously, the main interests of the circuit thermalhydraulic are the thermal gradients at the bundle exit and at the steam generator entrance. Regarding the temperature of the pipe wall just before entering the heated pipe section, Figure 22, the calculations demonstrate that the tentative test protocol conditions are acceptable both for the technological constraints and scientific purpose of minimizing Cs, I, Te vapour condensation. The level of sleeve burn out (1100°C) is never reached during the test and the wall temperature along the no heated length is practically always above 600°C during the fission product emission period. Nonetheless this result has to be treated with caution as it is rather dependent on outlet gas temperature and axial conductivity both in the structures and the gas. Future bundle calculations will indicate uncertainty ranges. The second location where thermalhydraulics is important is the steam generator entrance. As aerosol depletion through thermophoresis and FP vapour condensations are driven by the temperature difference between the flow and the wall, a correct prediction of these processes requires a precise estimation of thermalhydraulic conditions. Figures 23 and 24 give an example of the temperature and velocity variations of the carrier gas inside the steam generator at 1100 seconds.

224

THE PHEBUS FISSION PRODUCT PROJECT

Fig. 23. Gas temperature along the circuit

Fig. 24. Velocities along the circuit

3.3.2 Fission Product Depletion in the Circuit In this section we examine results from the TRAPFRANCE code. Generally RAFT results were similar but VICTORIA predicted much more retention of all species. The thermalhydraulic conditions were from CATHARE2 and incoming fission product and aerosol mass flow rates deduced from the elemental release resulting from the ICARE2-CORSOR calculations. At the bundle exit iodine was considered as CsI and the remaining Cs as CsOH.For Tellurium two calculation options were assumed. The first one considers all the Te as having reacted with Tin from the zirconium alloy to form SnTe. The remaining tin is included in the aerosol particles. For such an option the chemical sorption of SnTe is not yet available and the results for this species concern its vapour condensation only. The second option assumes that molecular Te arrives in the circuit. Cadmium is considered under its vapour form at the bundle exit. Regarding the aerosols, their initial distribution at the bundle exit was supposed lognormal with a geometric mean radius of 0.1 micron and a geometrical standard deviation of 1.7. An aerosol density of 5.5 gr/cm3 since UO2 release is vast by comparison with all but Sn (and Cd vapour). Some sensitivity calculations with different values of these previous parameters have been performed. No large differences were observed. Figure 25 gives an illustration of the aerosol release from the bundle. It can be noticed that according to CORSOR the aerosol release rate decreases drastically as soon as the bundle temperature is reduced during the cooldown phase. It is thought that this behaviour is not representative of the emission during this phase. In fact it could be one of the important periods for the release during the test according to the fracturating of the fuel pellet due to thermal strain. Unfortunately there is no tools available modelling this specific emission of FPs and aerosols. Figures 26 to 29 give a global picture of the retention inside the 18 volumes used to represent the circuit. The results are given as the fraction of the material deposited in each volume versus the total retention in the circuit. It indicates that the main location for retention is the steam generator tube (volumes 9 to 13) and more specifically the upwards part of this tube. Typically 80 or 90% of CsI, CsOH, SnTe and Cd depletions in the circuit are located in the steam generator tube.

FPT0 TEST PRECALCULATIONS

225

Fig. 25. Aerosol concentration at the bundle exit

Fig. 26. CsI depletion along the circuit

Regarding species like CsI, CsOH and SnTe the main depletion mechanisms are the vapour condensations onto the wall in the first meters of the tube and thermophoretic deposition of aerosol particles enriched of the previous species according to non-homogeneous vapour nucleation. Figures for the total retention in the circuit, expressed as percentage of the emitted amount of each species are: 29% for CsI; 38% for CsOH; 19% for SnTe and 58% for Cd. The behaviour of Tellurium is totally different if molecular Te is considered. In such a case the main retention process is the chemical sorption on the pipes walls. Roughly all the Te emitted by the bundle is deposited in the circuit and 90% of the retention is located in the horizontal line before the steam generator. Note that for the present calculations the pipe wall was considered as stainless steel instead of inconel according to the present modelisation of the code. These results show the importance of chemical speciation in fission product transport, at least for tellurium.

226

THE PHEBUS FISSION PRODUCT PROJECT

Fig. 27. CsOH depletion along the circuit

Fig. 28. SnTe depletion along the circuit

Regarding the aerosol retention 18% of the emitted mass are located in the circuit. Figure 30 gives a picture of the locations of the retentions. The steam generator is once more the main component for aerosol deposition (approximately 40% of the circuit deposited mass) and the driving mechanism is the thermophoretic one. But it can be observed that aerosol retentions take place else where along the circuit especially in the no heated part just above the bundle through thermophoresis and in the bends located on the line before the steam generator. Agglomeration, according to TRAPF sensitivity calculations, plays a minor role in its effect on total retention.

FPT0 TEST PRECALCULATIONS

227

Fig. 29. Cadmium depletion along the circuit

Fig. 30. Aerosol depletion along the circuit

3.3.3 Chemistry inside the circuit Within the next few weeks it is planned to analyse results of VICTORIA and RAFT codes both from JRC/ISPRA and CEA/ SEMAR.

228

THE PHEBUS FISSION PRODUCT PROJECT

3.4 Main Problems Encountered As has been said previously, the thermalhydraulic calculations are not very complex. Nonetheless a precise determination of the thermal behaviour just above the bundle required a radiative heat transfer model including gas emissivity. The CATHARE2 model was change to better accommodate our physical conditions but, up to now it does not include the effect of the aerosol particles on the emission properties of the steam-hydrogen gas mixture. A special purpose code, UPPLE, is also available for this region. Regarding aerosol and fission product transport results, one of the main problems is to have a relevant estimation of the release rates in the cooldown phase for which CORSOR predicts practically, no release. Without such an estimation the present results have to be regarded as lower bounds of the deposited amounts in the circuit. 4. CONTAINMENT STUDIES 4.1 Information from Shared Cost Action 2/B There were six participants from four countries. Thermalhydraulic calculations used CONTAIN, COCMEL, RALOC, WAVCO, CONTEMPT4 and JERICHO. Aerosol calculations used CONTAIN, AEROSIM, NAUA, HAARM-DTM and AEROSOLS-B2. Some iodine chemistry studies used IMPAIR-2. The principal results were summarized in [8]. The main findings were: • There was a large scatter in the thermalhydraulic code results. This was due partly to the codes themselves. Some participants used more than one code and obtained different results. Another reason was that the boundary conditions were not always exactly the same. • It was realized that there was no set of target thermalhydraulic conditions in the containment to aim for. A flexible approach was suggested instead whereby each test aimed for certain phenomena. • The dose rate in the sump was found to be too small compared to a reactor. • The results of sensitivity studies to the initial aerosol size distribution were contradictory. Some participants found it affected the long term size distribution and others found it did not. • A study of venting showed that condensate evaporated from walls but that relative humidity was lowered. • The thermal inertia of the vessel wall was found to be important. 4.2 The Containment Task Force The realization that the fission product inventory scaling factor was 5000 rather than 2000 led to the decision to reduce the containment volume from 25m3 to 10m3 in order to preserve chemical concentrations in the atmosphere. In addition the controversial results obtained from the shared cost action led CEA and JRC to set up a task force to look into this and other matters because the ordering of the vessel from the manufacturers was imminent. Many of the arguments used to define the geometry were either from technological constraints or from simple scaling laws but a large number of calculations were performed to check that they would work in practice. Some are reported in [9]. Apart from approving the volume change, other major changes from the previous design were the sump design, the introduction of an internal condenser and wall heaters in order to have a better representation of the surface versus volume ratio and the correct amount of dry and wet surfaces. Condensation was then supposed to occur on the inner condenser rather than the outer walls which should remain dry easing the decontamination process. 4.3 Dimensioning Verifications In order to produce a containment calculation it is necessary to have the results from the bundle and the circuit so inevitably there were less participants to this part of dimensioning verification calculations. In fact there were three. JERICHO and CONT were used for thermalhydraulics; NAUA and AEROSOLS-B2 for the aerosols.

FPT0 TEST PRECALCULATIONS

229

Fig. 31. Humidity in containment

The containment transient was divided into two parts. The first of them with an high humidity ratio and condensation of steam on the condenser structure. This phase represents the reactor containment thermalhydraulics before initiation of the molten core concrete interaction (MCCI). The second containment phase represents the long term period in a reactor. The conditions are expected to be superheated. Its duration is approximately three days. The radiochemistry of iodine in the atmosphere and the sump water will be studied throughout the transient. The main conclusions were: • The containment objectives were not clear. It was suggested that the target should be a particular relative humidity rather than to simulate a certain type of reactor. • The thermalhydraulic results depended very much on the wall heat transfer correlation chosen. The humidity in the short term phase is shown in Figure 31. It was suggested that this influence could be reduced by specifying a power extraction boundary condition rather than an atmosphere-condenser temperature difference. • The differences between the aerosol calculations were mostly consistent with the differences in thermalhydraulics. The case with the highest wall condensation showed most diffusiophoresis. The case that reached a relative humidity of near unity predicted bulk condensation onto the aerosols, bigger aerosols and more settling. • The inlet temperature is the temperature of the cold leg pipe and so it can be controlled easily in the experiment. • The thermalhydraulics of the atmosphere can also be affected by the sump temperature and a strategy should be defined for its control.

4.4 Exploratory Calculations The containment thermalhydraulic behaviour is divided into two main phases. During the first one the containment has an atmosphere temperature near saturation condition, an humidity ratio near 100% and steam condensation takes place onto the condenser surfaces. The incoming aerosols enrich the sump water through settling and diffusiophoresis. After the bundle and circuit transient have ended, the containment is disconnected from these components and the second phase of the experiment begins. The second period duration is three days with an atmosphere temperature increasing up to 150°C and superheated conditions. During this second phase the containment objective is to study the fission product chemistry and especially Iodine radiochemistry in the sump water, the atmosphere and the effect of paints in the ‘dirty’ chemical conditions of a reactor accident.

230

THE PHEBUS FISSION PRODUCT PROJECT

Fig. 32. Vessel boundary conditions

4.4.1 Containment thermalhydraulics Containment calculations are only performed after bundle and circuit predictions. The delay in fixing a reference scenario for these means that few containment calculations have been performed so far. The CEA calculations were performed with a part of the ESCADRE code system; namely JERICHO code for the containment thermalhydraulics, AEROSOL B2 for aerosol deposition and IODE for radiochemistry of iodine. Comparable calculations have been performed from JRC/ISPRA with CONT, CONTAIN and IMPAIR codes. Figure 32 gives the boundary conditions of the containment for the first period. In a preconditioning phase, starting from an equilibrium initial state at 80 °C the temperature of the vessel wall is increased up to 100°C the sump temperature reduced to 60°C and the condenser one maintained to 80°C over 10,000 seconds. Then the 9000 seconds of the containment phase devoted to high humidity and condensation starts. Figure 33 gives a comparison on the vessel incoming steam flow, the condensation rates onto the condenser structure and the sump water surface. It can be observed that this condensation onto the water surface contributes about a third of the total condensation approximately. According to these JERICHO calculations condensation on the vessel wall has been successfully avoided. The humidity ratio behaviour is depicted in Figure 34. At the end of the preconditioning phase and before arrival of steam from the bundle and the circuit this ratio is around 50%. As soon as steam is injected humidity reaches 98% and is maintained around such a value by adjustment of the condenser cold temperature throughout the 9000 seconds. This adjustment is rather complex, even for the homogeneous and ideal descriptions of the containment behaviour given by the code, because of the variations of the incoming mass flow rate and the inertia of the thermal response of the condenser. After this first phase of the transient, the containment is isolated from the bundle and the circuit and behaves as a closed system. Its atmosphere is driven to superheated conditions by increasing the temperatures of the vessel wall and the condenser (which is now acting as an heater system) up to 150°C. During this phase the sump water temperature is also increased to suppress condensation onto its surface. Figures 35 and 36 give a summary of this second phase behaviour. An alternative and simple scenario for the containment thermalhydraulics was also calculated by JRC with the CONT and CONTAIN codes. The initial state is with all structures and the sump water at 80°C. Relative humidity is 100%. The bundle transient begins without a vessel preconditioning phase and the flow of steam, hydrogen and fission products enters the vessel over 9000 seconds. To avoid condensation on the wall the heating circuits are used to produce the wall temperature transient. The portion of the vessel in contact with the sump is assumed temperature to be controlled to ensure zero heat exchange with the water. The resulting temperature transients of the atmosphere and the sump are shown in Figure 37. The atmosphere is throughout somewhat warmer than the vessel wall, which, in turn, is warmer than the sump water. To induce condensation and diffusiophoresis on the painted surface, the condenser is operated with an extracted power history which corresponds to a condensation rate of 0.6 gr/s at 3000 seconds decreasing linearly to zero at 7000 seconds. The

FPT0 TEST PRECALCULATIONS

231

Fig. 33. Condensation rates

Fig. 34. Relative humidity inside the containment

resulting relative humidity transient is given in Figure 38 and is quite similar to that of CEA calculations reported above with different boundary conditions. The aerosol deposition calculated by JRC with CONTAIN are similar to those in the CEA calculations. In particular the ratio of diffusiophoretic deposition to settling is approximately 1 to 10. 4.4.2 Aerosols inside the containment Figure 39 gives an overall picture of the main depletion processes for the containment transient. It will be noted no more that 20,000 seconds are required to extract from the atmosphere the main part of the suspended mass.

232

THE PHEBUS FISSION PRODUCT PROJECT

Fig. 35. Boundary conditions

Fig. 36. Pressure for long term period

4.4.3 Iodine chemistry inside the containment The CEA results regarding the iodine behaviour in the containment has been computed with IODE code. They integrate the thermalhydraulics and aerosol depletion rates previously described. The results take into account both the wet and dry surfaces of the condenser and the immersed painted surface of the sump. The dose rate evaluation has been improved and the sump water pH is deduced by a POTHY code estimate. For the present calculations the oxidation of iodine by O3 and silver were not considered. The main trends of the iodine are given with Figures 40 and 41. It can be observed that the FPT0 inventory for this species and the approximately neutral initial pH lead to a low iodine airborne concentrations and therefore to a high partition coefficient.

FPT0 TEST PRECALCULATIONS

233

Fig. 37 Containment temperatures

One important point to note is the fast evolution of the concentrations during the first 50,000 seconds then an approximately stabilized state of the species in the atmosphere and the sump water. If such behaviour is confirmed by future calculations it might help to shorten the long term transient of the containment. 4.5 Main Problems Encountered Regarding the containment, the main problems are related to the thermalhydraulics and especially that of the period devoted to high humidity ratio. The control of this parameter is known to be very complex. Furthermore this quantity is highly dependent of local non-homogeneity in the vessel atmosphere or ‘cold points’ on the vessel wall which might be associated with instrumentation penetrations. The present status of the codes does not allow to represent the containment behaviour with such a level of detail. Therefore we cannot be totally confident of our ability to obtain a correct control of the condensation rates and the humidity ratio. A preliminary test of the containment and its condenser device thermalhydraulics is highly desirable both for appreciation of the technological constraints of these components and a correct appreciation of the code ability to compute their behaviours. In the absence of such a test a rather simpler operating scenario may be adopted for the first PHEBUS FP test. Condensation might be induced on the painted surfaces of the condenser structure, but without high relative humidity ratio because of the risk of condensation on the vessel wall. Such condensations must be avoided both to respect the volume/area ratio characteristics of PWRs and for decontamination ease of the vessel. 5. CONCLUSION This paper has described a long programme of pre-test calculations involving many different organisations. All proposed sequences have been calculated by at least two different groups and two different computer codes. Mostly more than two have been used and this provides a quality check of the results. In the event very few input errors have been detected and

234

THE PHEBUS FISSION PRODUCT PROJECT

Fig. 38. Relative humidity vs time

Fig. 39. Balance of aerosol depletion processes in containment

differences in results have been mostly attributed to alternative ways that complex physical and chemical phenomena can be modelled. The first set of calculations, the ‘Phase A’ exercise, was for full-sized reactors and the results provided target phenomena that should be reproduced in Phebus. A test-circuit was then designed based on a scaling factor of 2000 and certain technical constraints. A set of scooping calculations on this geometry, the ‘Phase B’ exercise, checked this design for a number of simulated reactor sequences at high and low pressure. The conclusion was that Phebus could simulate most phenomena reasonably well but a number of design modifications would improve representativeness. The most important of these was

FPT0 TEST PRECALCULATIONS

235

Fig. 40. I2 concentration in the atmosphere

Fig. 41. I2 concentration in water

that the circuit should be scaled by (1:5000 if similar aerosol and fission product concentrations to a reactor were to be reached because this is the ratio of the Phebus fission product inventory to the real case). High retention in certain pipes that were supposed to be neutral and non-representative behaviour in the containment led to separate studies of these two components. Results from a horizontal line benchmark exercise recommended a larger diameter pipe than had been envisaged before and the containment task force looked into all aspects of the containment. They performed many calculations and examined closely the technical constraints before suggesting a smaller vessel, a sump that did not extend across the whole cross section of the vessel, an outer wall hot enough to discourage condensation and an inner structure, the ‘condenser’, that would condense and collect water from the atmosphere.

236

THE PHEBUS FISSION PRODUCT PROJECT

The next set of calculations, the dimensioning verifications (‘Benchmark exercise’) concentrated on the first test FPT-0 and calculations were sought for the whole sequence, rom bundle to containment. This exercise was enthusiastically carried out by partners from many organisations. The circuit results indicated that objectives could be met apart from a phase when the secondary side of the steam generator was below the saturation temperature of the steam. This phase of the transient has now been dropped so the boundary conditions can be considered adequate. The bundle and the containment still posed problems. The trouble with the bundle was that it was found impossible to achieve an objective of low clad oxidation together with a low fraction of hydrogen in the coolant. Subsequent calculations confirmed the impossibility of this scenario in Phebus and the low clad oxidation requirement has now been dropped. The containment calculations showed that a power extraction boundary condition for the condenser rather than a fixed temperature should be used in the test. Armed with this knowledge a new set of boundary conditions was proposed which offered more hope of success. Results using these showed that, although the test objectives were achievable, uncertainties in shroud properties would necessitate some feedback loop to control the bundle power. High humidities in the containment were shown to be difficult to achieve without extensive thermalhydraulic testing of the containment vessel so it was decided to aim for a lower humidity in this test. For the first time chemical equilibrium calculations were made in the primary circuit and, although results are available, they require more study before we can really assess the importance of chemistry in the test. Codes are being modified to predict the test better. Some of these modifications address problems that are not found in reactors such as high radial heat fluxes in the bundle and the internal condenser in the containment; other modifications will result in better reactor calculations as well. Results from the calculations are being fed to the experimental team who need to have an idea of the range of values that one might expect in the measurements. Advice is also given to the experimental team as to the best position and operating conditions for instruments in order that they should not disturb the experiment too much. Currently a major sensitivity analysis is underway. This will involve many codes and many organisations. It will examine sensitivity to boundary conditions such as flow rates, sensitivity to physical properties such as shroud conductivity, sensitivity to different model for such phenomena as fission product release and core degradation, and sensitivity to numerical models, timestep and spatial nodalisation. The result of all these calculations will be used to define a full Test Protocol for the first Phebus-FP test. ACKNOWLEDGEMENTS The authors gratefully acknowledge numerous contributions to the work presented in this paper. In particular thanks are due to Miss Isabelle Drosik for her contributions to the CEA ICARE bundle results, Mr Klaus Hocke for the IKE KESS calculations, Mrs Carmen Vicente for the CIEMAT ICARE results, Mr Martin Kissane for the CEA CATHARE calculations of the circuit thermalhydraulics and the TRAP-F aerosol transport results, Mr Joaquim Capitao for the JRC RAFT circuit aerosol results in section 3.3.3, Mr Luigi Biasi for the JRC containment thermalhydraulics calculations in section 4.4.1, Mr Philippe Marsault and Mr Claude Hueber for thermalhydraulic, aerosol and chemistry results inside the containment. REFERENCES [1]

[2]

[3] [4]

[5] [6]

[7] [8]

A.MARKOVINA, P.FASOLI-STELLA and A.MAILLIAT. Review of the major predicted phenomena during FP transport and deposition in the RCS and containment building under severe accident conditions. International seminar on fission product transport processes in reactor accident. Dubrovnik, May 22–26, 1989. Phase B.A.MARKOVINA, P.FASOLI-STELLA and A.MAILLIAT. ‘Analytical assessment of the capability of a scaled down in pile for facility to simulate PWR phenomena under severe accident conditions’. AAAR ‘89, Eighth annual meeting, Reno, Nevada, Oct. 10–13, 1989. A.MARKOVINA and A.MAILLIAT. ‘Scoping calculations in support of the PHEBUS FP experimental programme’. Seminar on the Commission contribution to the reactor safety research. Varese, Italy, Nov. 20–24, 1989. I.SHEPHERD, I.DROSIK, P.DUMAZ, B.FABRE, A.MAILLIAT, M.PIGNARD, A. BALL, K.TRAMBAUER, F.BARBERO, F.OLIVAR DOMINGUEZ, L.HERRANZ, L.BIASI, J, FERMANDJIAN and K.HOCKE. PHEBUS-FPT0 Benchmark Calculations. SAWG 90/006/3. I.SHEPHERD, I.DROSIK, M.C.VICENTE and K.HOCKE. PHEBUS-FPT0. Pre-test calculations for the bundle. SAWG 91/019/0. A.MARKOVINA and I.SHEPHERD. Summary of the dimensioning verification studies of the Phebus FP experiments (Phase B of the share cost actions: scoping calculations in support of the Phebus FP project) Volume 2: The reactor coolant circuit. To appear as EUR report. A.V.JONES, E.BONANNI and A.MARKOVINA. Principal results of the ‘Phase B’ verification studies in support of the Phebus FP project. CSNI aerosol workshop, Fontenay-aux-Roses, September 1990. L.BIASI. Implementation, development and application of LWR fission product transport models and codes contract number 386889– 12 ED ISP I, Final report.

FPT0 TEST PRECALCULATIONS

237

Discussion following the presentations of SESSION IV Summary of the chairman Mr. A.Tattegrain

The first presentation clearly pointed out the state of the art in the field of source term codes. Questions to Dr. Hayns concerning the practical way in providing the best precalculations for a test were answered by his co-author Dr. Kinnersly as follows: - For some phenomena there is a need for precise models. For example, for processes occurring in long pipes, which are not typical for reactors detailed mechanistic codes are required in order to eliminate uncertainties from this source. - For some features related to core degradation and melt progression, it is appropriate to tune your code to the characteristics of the Phebus reactor. By that way you produce a design code for your experiments. The Phebus CSD experiments could provide a data base for this purpose. You should, however, avoid integrating your data base in the code but rather keep it separate for easy access and revision. It should be kept in mind that fission product (FP) calculations need a reliable thermohydraulic base for the primary circuit and the containment. There is little use in improving FP modelling as long as the basic thermohydraulics are limited. These remarks were completed by the following comments from the audience: - Usually people have too high expectations from a so-called detailed mechanistic code. There is always a lapse of time between code design and its application, which implies that the code is never up to date. To analyse specific phenomena, you should, for that reason, rely upon the best available phenomenological models and make the necessary adjustments to the code. You should not forget that codes essentially are integration tools. - A code should never be separated from the knowledge of the experimentalists who, alone, are able to evaluate the uncertainties. - The decision concerning “at what stage code development should be stopped”, has actually to come from the code user by specifying the application case. The second part of the session was devoted to the code systems ESCADRE, ICARE and ESTER. Concerning the use of the ESTER code package, Dr. Jones pointed out that the Commission intends to apply it, first to Phebus-FP. Most of the contractors involved so far in the development of ESTER are interested in using ESTER for their future needs. The discussion of the third part of the Session was related to the precalculations of the first Phebus-FP test. In reply to a question regarding the relocation blockage of the bundle, Mr. Mailliat pointed out that with the calculations performed so far a complete co-planar blockage has never been observed. However, this possibility has been accounted for in the bundle design by installing an additional carrier gas line above the bundle. In case of bundle blockage, flow injection will be triggered through this line. Refering to the decrease of the inlet flow rate during core degradation, Mr. Mailliat said that according to the reactor calculations performed in the frame of the phase A scoping calculations, a decrease of the steam outflow from the core was observed starting from about 10 kg/s and decreasing down to zero. During the same period the H2 mass flowrate increases accordingly. The intention is to reproduce this tendency for representativity purposes in FPT0. The chairman concluded the Session in saying that the topic of this session was rather ambitious. The principles stated by Dr. Kinnersly could certainly have been discussed for hours. The ESCADRE code suite is undergoing continuous improvement. ESTER, the interesting and ambitious approach of JRC-Ispra into which, amongst other, the ICARE code is integrated, seems to be progressing well. The present state of test precalculations was demonstrated for the case of the first Phebus-FP test. The chairman expressed his hopes that the international contributions to the project will continue to increase in order to assure a wide-spread expertise and the largest possible benefit.

PANEL DISCUSSION VALUE OF THE PHEBUS-FP AND RELATED SOURCE TERM STUDIES FOR THE SAFETY OF LWR’S

Chairman: H.F.Holtbecker (CEC) Members: M.Akiyama (JAP) , R.E.van Geen (B) , M.Pezzilli (I) , C.Lecomte (F) , M.R.Hayns (GB) , T.P.Speis (USA) , M.Banaschik (D) Closing remarks List of participants Index of authors

PANEL DISCUSSION Value of the PHEBUS-FP and related source term studies for the safety of LWRs

Chairman H.F.Holtbecker, Director of the Safety Technology Institute, Joint Research Centre Ispra. Panel Members M.Akiyama Professor, Department of Nuclear Engineering, Faculty of Engineering, University of Tokyo R.E.Van Geen Professor, Chairman of National Council of Science Policy and Vice-Chancellor of the Free University of Brussels M.Pezzilli Project Manager for Safety Research, Energia Nucleare ed Energie Alternative, Centre di Ricerca Casaccia, Roma C.Lecomte Head of Service for Accident Studies, Commissariat a l’Energie Atomique, Institut de Protection et de Sûreté Nucléaire, Fontenay-aux-Roses M.R.Hayns Head of Advanced System Division, AEA Technology Harwell T.P.Speis Deputy Director for Research, Office of Nuclear Regulatory Research, Nuclear Regulatory Commission, Washington D.C. M.Banaschik Division Head Research Management, Gesellschaft für Reaktorsicherheit, Köln Secretary W.Krischer

Safety Technology Institute, Joint Research Centre Ispra, Commission of the European Communities Introduction

The chairman opened the panel session and presented the panel members. In his introduction he summarised the objectives and the status of the PHEBUS-FP project. The problem addressed by the project is to quantitatively describe phenomena occurring during various core melt-down scenarios. The intention is to solve the problem of source term quantification for the improvement of accident management and post-accident recovery procedures. There is a need for continuous updating of project objectives and definitions of accuracies to be achieved in the results. There is a demand in Europe for a common source term evaluation procedure which can be used by the licensing authorities, the reactor builders and the utilities. Major expectations from the PHEBUS-FP programme relate amongst others to: -

FP release during early and late stages of core melt; refractory nuclides released during UO2 melting; chemical reactions during transport and deposition of aerosols; effect of control rod and structural material on aerosol formation; aerosol deposition in bends and flow discontinuities; interaction of FP with boric acid.

We shall perform and/or evaluate separate effect tests supplementary to the integral PHEBUS-FP tests. Experimental activities are tightly linked to model and code development and their validation. To obtain an integrated view on separate effects, a common data base will be required. To harmonise the scope in model and code development, the EC develops an

PANEL DISCUSSION

241

informatic structure, called ESTER. In addition, a close collaboration between code developers and experimentalists is being promoted. Recommendations made during the Seminar so far, related to difficulties which might arise during the performance and interpretation of the experiments and suggestions how they might be overcome will be duly considered. There will be a continuous effort in updating the PHEBUS-FP test matrix, taking account of lessons learnt from separate effect tests and from previous in-pile tests. Each in-pile experiment will be evaluated with regard to the various phenomena involved. We have to know how to pre-calculate the fuel behaviour, as a function of the power input and the fission product release, considering morphological and chemical aspects of the fuel. We have to address unresolved problems in the modelling of e.g. mixed aerosols (soluble and insoluble), thermal-hydraulics, FP chemistry, etc. The chemistry aspects, covering a rather large and widely unknown area, are thoroughly taken care of within the PHEBUSFP project. This relates to the FP inventory of the high burn-up fuel, the control rod releases, the role of boric acid and structural components. To enable a proper understanding of all important aspects involved in FP chemistry, the experimental conditions have to be well under control. This implies on-line instrumentation and post-test analysis, like temperature measurements above the core and in the high temperature region of the primary circuit as well as sampling of aerosols throughout the circuit. In this field we have to aim for an optimal approach between space and financial constraints and between reliable, conventional and advanced measuring methods. The project will show whether existing models and codes are valid for PHEBUS-FP and for extrapolations to the reactor case. To achieve this objective, we have to make sure that the experimental boundary conditions are within the frame of realistic reactor conditions. The application range of a code must be well defined in advance, to enable an appropriate validation. The chairman invited the panel members to state their position on the PHEBUS-FP programme orientation, including supporting activities, on the expected progress in the field of source term studies and asked for suggestions for possible improvements. Mr. M.Akiyama First of all, I would like to congratulate the project staff for keeping the PHEBUS-FP programme in steady progress and for the so far excellent achievements. I am aware that in the field of severe accident research, the PHEBUS-FP programme plays a significant role in providing a large amount of valuable information with a series of integral tests to be carried out in a test reactor under carefully elaborated conditions. Let me start with two basic considerations concerning the execution of the programme. First of all, high priority should be assigned to topics which could only be addressed appropriately by integrated experiments. Furthermore, experimental and analytical efforts should be well coordinated and directed versus the intended objectives. In general, a strategic approach is needed prior to the detailed planning of experiments, including the design, construction and operation of the test facility and the arrangement of the test matrix. In case of the PHEBUS-FP programme, this kind of approach was adopted successfully in a combined effort of extensive analytical works, including PSA’s and by investigation of the application range of PHEBUS-FP test results. In order to cope with future discussions and criticisms on nuclear reactor safety, research in this field should address both practical and scientific problems. When defining the test matrix, high priority should be given to those aspects which are closely related to the dominant candidate contributors to the results of PSA level-1 and level-2. It should, however, be kept in mind that it is not always reasonable to rely upon PSA results which perhaps have only recently been obtained and which need upgrading by continuous research. These reservations apply to the Japanese PSA results presented at the foregoing PHEBUS-SFD seminar, which have yet to be cleared by the government and the industry. The PSA results once evaluated in conjunction with associated analytical work, could give some guidance in setting up a strategic plan for important programmes like PHEBUS-FP. In this respect, the containment by-pass sequence should duly be considered in the strategic plan. A close collaboration between analysts and experimentalists on the one hand, and the scientific and technological experts in related areas on the other, should be aimed at. The complexity of the phenomena investigated can be handled appropriately only in such a setting. The complexity involved is expressed in the new discipline called “mechanics of continua”, which deals with the combination of physico-chemical-thermal-fluid aspects. With present computer capacities a detailed simulation of a severe accident would possibly take something like ten years. For this reason, the modelling of the phenomena and the setting up of a data base are of indispensible value. In addition, the test matrix should be complemented with out-of-pile tests to study some of the phenomena less well known. Let me finally make some comments with regard to the shroud material. In one of the PHEBUS-SFD reports, the uncertainties in the shroud heat conduction characteristics have been pointed out. Difficulties might be encountered also

242

THE PHEBUS FISSION PRODUCT PROJECT

during manufacturing. With this in mind, it might be advisable to consider besides the composite of a porous and a high density Zirconium-oxide an alternative material with better known properties. Mr. R.E.Van Geen Let me make an attempt, in my quality of Chairman of the Belgian Science and Technology Policy Council and as an external observer of the PHEBUS-FP programme but potential end-user, to make in the five minutes I have been granted some general comments. PHEBUS-FP is evidently a very valuable programme, not only in nuclear technology but also as an exercise and configuration of a new European science policy, involving the development of a real network. I regret that Belgium and the Nuclear Research Centre SCK Mol, which remains one of the large research centres on reactor and fuel safety, have grossly underestimated the importance of the PHEBUS-FP programme. One of the reasons for this is the strong involvement in beyond design accident research for fast breeders. However, we intend to join the project and respective negotiations have already been taken up. The reason to join this venture is that Belgium considers the results of PHEBUS-FP as part of a European reference solution to the source term issue. One of our major problems is clearly public acceptance. In my country, as well as in others, there is a general tendency for the life-extension of the actual nuclear power plants (i.e. PWRs). The difficult part will be the relicensing before the end of the century. We already experienced which kind of difficulties we have to face with our licensing authorities, if we fail to provide an appropriate knowledge of the mechanics behind fission product release and if we cannot demonstrate that, in case of a severe accident, we have the FP-release under control. The PHEBUS-FP results will support in terms of a European reference the relicensing of existing NPPs. A still better solution would, however, be to have European rules for licensing. Without going into the details of the experiments and the related codes, I would like to make some remarks concerning the foreseeable weaknesses in the methodology: - We need a better understanding of the chemical reactions and mass transport between the gas phase and the aerosols under realistic conditions in view of temperature, pressure, flow and turbulence. In other words, we have to move from thermohydraulics to thermodynamics. - We face rather complex systems involving nearly half of Mendeleev’s periodic table. Thus, the analysis of the physical and chemical systems has to be limited to the basic phenomena and to simplified chemistry, with exemption of their mathematical treatment. The modelling of the phenomena involved plays an important role with regard to our future purposes, i.e. design, licensing and relicensing, since we have to demonstrate the validity of our models and codes for extrapolation. - Although transuranian products are believed not to be released, it still has to be demonstrated in order to satisfy licensing procedures. In any case, it is less evident, if we move from traditional to (15–20%) MOX fuel. Due to informatics, scientific research is undergoing today profound methodological changes. The time spent on the development of a theory and on its experimental validation has revolved. Supercomputer analysis, and in particular non-linear finite element analysis, in conjunction with computer assisted measurements offer the possibility for an interactive approach between experiment and modelling. This enables to take advantage of the actual latest state of knowledge at each time step. The combination of thermodynamics and chemistry should take advantage of this development. Having this in mind, I am convinced that the PHEBUS-FP programme can provide more than just interesting experimental results. The project could become a model for collaboration in the European Scientific Community by establishing links which could be extended, at a later stage, to a coordination network. Mr. M.Pezzilli As you certainly know, the public apprehension for the nuclear risk forced the Italian Government to close operating plants and stop the construction of new ones, with a five-year moratorium. At the same time, a mandate was assigned to the national operators (utility, industry, R&D organization and licensing authority) to revise general safety principles and to re-orientate research and development programmes towards new concepts offering more persuation for public acceptance. In view of this new approach to safety, now under discussion, I would like to comment briefly the importance of the PHEBUS-FP programme for Italy. The major safety goal of this new approach is to limit the environmental impact and off-site radiological consequences in such a way that neither any evacuation plan will be needed, nor that significant long-term land contamination will occur. Consequences shall be ensured by deterministic analysis for all conceivable events, even extremely rare ones like severe accidents with advanced core degradation and large radioactive releases. With such an approach, of a “defense in depth”

PANEL DISCUSSION

243

concept, the containment system plays a key role. Its reliability and the adequacy of all auxiliary systems have to be demonstrated for all accident circumstances, including severe events. The required study has to model phenomenologies not completely known and not easily reproducible under experimental conditions, as there are: - the core melt progression and the fission product release; - the transport and retention of contaminants in the reactor loops; - the fission product behaviour in the containment atmosphere, including the effects of compartment geometries and thermodynamic conditions; - the efficiency of pool scrubbing; - the potential of energy generation inside the containment, in case of: . high pressure melt ejection and direct containment heating, . steam explosion, . hydrogen deflagration; - the corium-concrete interaction effects; - the containment loading response (including material properties under dynamic loads, penetration behaviour, etc.). Part of these problems is addressed by the PHEBUS-FP programme. The main objective is to obtain integral test results for code validation, in a facility specially designed to achieve an optimal representativity of the source term and the environmental conditions encountered by fission products on their path during an accident. In view of this, the PHEBUS-FP programme covers all we need, but certainly constitutes an experimental basis for the “deterministic evidence” of some phenomenological aspects to be evaluated in severe accident analysis. To confirm the validity of this approach, we have to demonstrate that we are capable to interprete the experiments. This leads to questions, already discussed during the seminar, like: - are the phenomena occurring in a radionuclides releasing accident adequately known?; and - are the experimental technique and the instrumentation adequate to evidence and monitor these phenomena? I am convinced that a positive answer can be found, considering an integrated approach in which contributions are coming from: - analytical tests and separate effect experiments; and - integral tests of PHEBUS-FP. A considerable effort is on the way to analyse different phenomenological aspects and to integrate and harmonize the analytical and experimental qualifications of different teams. Two ladies spoke yesterday about programmes in support to PHEBUS-FP: Mme. Lecomte about the CEA activities and Mrs. Fasoli-Stella about the actions supported by the European Community through contracts with European and extra-European partners. I believe that, with such an approach, the PHEBUS-FP programme can fulfil the objectives of an integral test by assigning the main importnce to the various phenomena under representative reactor conditions and thus providing a valuable experimental reference for code validation. With this aim, ENEA and other Italian organizations are supporting the PHEBUS-FP programme through participation in the actions promoted by the Commission of the European Communities and through a collaboration starting to be taken up in the framework of the ENEA-CEA agreement for the research on future reactors. Mrs. C.Lecomte As I have been requested to speak here from a safety analysis point of view, I first want to stress the link of PHEBUS-FP to the French safety philosophy concerning intervention strategies with respect to potential radioactive releases to the environment. The counter-measures which would eventually be taken are indeed closely connected to the level, nature and kinetics of releases and are in accordance with international protection recommendations. As such, the first interest of the PHEBUS-FP programme is to provide a global validation of the tools used for source term evaluation: this validation has to address the most significant phenomena which contribute to the risk and give comprehensive information usable to elaborate prevention, mitigation, and optimization of intervention plans which are designed to cope with severe accidents of nuclear power plants.

244

THE PHEBUS FISSION PRODUCT PROJECT

As a consequence, two specific features of the PHEBUS-FP programme are of special interest: - first, it is very useful that PHEBUS-FP has the major aim to quantify relevant phenomena involved in sequences selected for their importance from probabilistic and release level points of view, rather than to reproduce any specific accident sequence; - second, it is important to note that PHEBUS-FP is not a pure demonstration experiment. It is also designed as a means to validate scientific tools; this necessarily implies an adequate scaling effort, in order to be sure that the phenomenology investigated is the same as in the reactor case. We were also asked to suggest possible ameliorations to this programme. Speaking of an already very well developed programme, what follows are not really suggestions for improvements but rather remarks which tend to stress the necessity of a continued effort on specific points: - the above-cited objectives for source term evaluation imply kinetic measurements, as well as detailed investigations on fission product chemistry, including iodine chemistry which is an essential topic for present and future reactors; - generally speaking, it would be wise to give priority to the study in PHEBUS of phenomena that cannot be studied elsewhere, either because of the specific chemistry, or because of the interaction between phenomena; the results of complementary analytical tests will provide the necessary information to complete the status of knowledge; - on the other hand, it is essential to study in PHEBUS phenomena which may have a significant impact on source term, like retention in steam generators; the results may influence source term evaluation and optimization of Accident Management strategies. As a conclusion, PHEBUS-FP is an important programme for present reactors and for the optimization of efforts for source term reduction in future reactors. The effort which has been presented through this seminar must be pursued to provide qualified data for validation of source term codes. Mr. M.R.Hayns The progress reported by the PHEBUS project team members was very impressive and I would wish to express my congratulations to them on keeping to time and to cost what is a very complex multi-disciplinary and multinational project. This seminar had been a success in my view since its objectives of giving advice on the status of the project and providing a forum for discussion of interested parties had certainly been achieved. It was important to re-affirm the original rationale and objectives of the PHEBUS-FP project. Also to endorse the CEC decision to concentrate on severe accident research and on source term/core melt behaviour in particular. The international scene on severe accident R&D makes it even more important that a single international focus be provided for the large-scale irradiated experiment which is required as a validation for the complex suites of codes which have been developed over the past decade in this area. International activities have become the norm rather than the rarity in this area as people in all countries operating nuclear stations recognise both the importance of this topic and the need for sharing facilities and knowledge amongst each other. Within the project and from this seminar a number of technical issues arose. The most important amongst them was that of instrumentation. Instrumentation is absolutely vital to the success of the experiment and there is a continuing need to keep upto-date with techniques and requirements. However, because of the enormous complexity of potential measuring techniques it was vital to be realistic both in terms of cost and reliability in choosing the instrumentation to be implemented on the experimental train. It was also highlighted that pre-calculations were an exceptionally cost-effective way of ensuring that the experiments were as well devised as possible. In this context it was important to highlight the value of advice which should be given to the working group from countries interested in the successful outcome of the experiment. There was also a continuing requirement to remain aware of international developments both in code advances and in understanding of data and mechanisms. Harmonisation of views in the international community towards the topic of severe accidents was essential both from the point of view of sharing expensive and difficult data, but also in reassuring the public that world class experts all agreed on the nature of the science and technology involved and the routes to the solution of the engineering problems. It was also important to point out that the end user for these data and the customer in terms of paying the bills should always have the highest priority in the technical teams minds. The experiment is focused on providing validation for the major source term codes and is not an end to itself. The usefulness of the seminar had proved itself as a mechanism for providing a forum for discussion and feedback between interested technical specialists and those responsible for running the programme. However, it is also important to record that

PANEL DISCUSSION

245

the real technical feedback needs to be channelled via the working groups of the PHEBUS project and so it is very important that as much technical input and support as is possible is given to them. Mr. T.P.Speis I am very happy, and indeed honoured, to have been asked to participate in this very important seminar of the PHEBUS-FP project. I have found most of the papers and the ensuing discussions very enlightening and well focused on the subject of this seminar, which is of course the source term and the related experimental/analytical program of the PHEBUS-FP project. The cardinal question, of course, is what do we want to learn from the PHEBUS-FP program. There is no question that we are going to learn a number of things relating to the source term. Some of these were discussed during this seminar. But still we need to put this whole program in some perspective. What we need first of all is a STRATEGY, i.e. what use is to be made of this source term (including the additional knowledge to be gained from the program). We in the United States, when we talk about the source term, we mean the radioactivity which enters the containment during a severe accident and which includes the type and quantity of the various nuclides as well as their physical and chemical characteristics. This source term, in conjunction with containment performance requirements to severe accident challenges, will be utilized in the design of the containment and in a number of equipment/systems associated with it. This source term has to be chosen in such a way that changes to its characteristics (sensitivity studies) have no effect on the containment design; this source term can thus contribute to a “robust” containment design. In addition to using the source term to set the design requirements for the containment and a number of systems associated with it, the releases to the environment will also have to be assessed to ensure that they meet some preset goal (e.g. a numerical criterion for a large release or a specified limit on a particular nuclide). Thus, how the source term behaves inside the containment, including its interaction with the containment environment during the evolution of a severe accident (e.g. availability or not of containment sprays), is crucial to this assessment and here the PHEBUS data/insights will be very valuable, especially in providing confidence in the use of analytical tools to explore various types of behaviour. It is doubtful if the PHEBUS program uncovers any new phenomena associated with the source term. The theoreticians have thought a lot about the source term the last few years, and it remains for PHEBUS to provide some more quantitative information about phenomena/processes already known but under a broader spectrum of accident conditions. It is also possible that PHEBUS can provide data on the later stages of a core melt accident, i.e. fission product releases from rubble beds and possibly molten pools assuming, of course, that the experiments can be performed under more extreme conditions than presently planned. But again, I think that the main contribution of the PHEBUS-FP program will be to add to our confidence regarding our understanding of the source term and the severe accident issues associated with it in a more integral sense. During this seminar there were many new ideas/suggestions of how to improve the test matrix, how to extend the test conditions or to add more instrumentation, etc. Some of these ideas sounded good, but whether and to what extent they can substantially contribute to the program’s upgrade without “breaking the bank” remains to be seen. In any event, all of us need to think carefully about all these proposals and provide our conclusions to the project as soon as possible. It could also be worthwhile to have an ongoing review of the program as it proceeds and to take the new data/insights into account in the performance of the follow-on experiments if this is practical. We at NRC are committed to the PHEBUS-FP program and will participate in all the related programmatic activities to the extent needed to assure its success. Mr. M.Banaschik The presentations of the two days have shown that a thorough planning and realization of the PHEBUS-FP experiments is on the way. In the overall reactor safety research source term experiments and analyses are of high importance for the containment performance and fission product issues with consequences to the public. Therefore, the most important features for the performance of the PHEBUS-FP experiments seem to be: - realistic boundary conditions for the experiment emphasizing the separate effect features in an integral test facility; - completeness of phenomena governing fission product release; - reliable and tested instrumentation for the data acquisition system. With this in mind, we expect a consistent set of fission product release data which will fit well to the results of other in- and out-of-pile experiments. Apart from the experimental work on FP-release and chemistry it is necessary to proceed with code development. We are funding on the national basis code developments for fission product and aerosol behaviour under severe accident conditions. The coupling of thermohydraulic and fission product behaviour is of high importance for the appraisal of realistic source terms.

246

THE PHEBUS FISSION PRODUCT PROJECT

Accident management (AM) has gained high priority in the safety of reactor operation. It is a powerful tool to further enhance the operational safety and to prevent or mitigate accidents. The development of various accident management measures has revealed that additional safety margin is available to prevent and mitigate beyond design basis accidents by the flexible application of operating and safety devices. The sequences of accidents and the related phenomena are expected to be so complex that it seems opportune to reproduce the state of the plant under abnormal conditions as realistically as possible. Therefore, it is of major importance to gain results from the PHEBUS-FP experiments which can be assigned to well defined boundary conditions in order to provide a set of data for code validation. Let me conclude with complements to the organizers of the seminar of the PHEBUS-FP project and congratulate the Research Centre of Cadarache to the setting up of the experiments. Let me thank the Joint Research Centre Ispra for providing the opportunity to EC member states to participate in the PHEBUS-FP project. I appreciate that most, countries with, nuclear programmes are going to join the project which may serve as a forum for discussion of source term issues in severe accidents.

DISCUSSION

The chairman opened the discussion inviting the audience to participate. Mr. A.G.Markovina, CEC/JRC Ispra During the present seminar, attention was paid exclusively to source term problems associated with current reactor designs. Could the panel give some indication on the evolution of source term issues in view of the future generation of reactors, i.e. the so-called inherently safe reactors, and how this would effect the priorities of the PHEBUS-FP programme? Mr. M.R.Hayns I recommend to avoid the term “inherently safe reactors”, since if they really would be inherently safe, they never would produce core melt accidents. Understanding FP behaviour and transport will help to design future reactor systems. There was always a significant feedback of experience on the next generation of design. In this context PHEBUS-FP could make an important contribution in the understanding of the underlying phenomena, in code validation and setting up of a data base on FP/aerosol behaviour. Mr. M.Pezzilli Apart from passive preventive measures like improved FP retention capacity of the fuel, there should be not much difference for future reactor designs in terms of accident evolution and the respective source term issues. Independently of future design, there is a need for a common denominator with regard to release limits to the environment. Mr. G.Sandrelli, ENEL/CRTN Milano In principle, the phenomenology related to source term will be the same for new designs as for existing LWRs. But more stringent demands concerning external releases, in conjunction with larger amounts of water which are involved in new nuclear power plants (released into the containment at a late stage of a severe accident) make the knowledge about the chemical behaviour of fission products in water pools and sumps, in the presence of a high radiation field, one of the most crucial requirements. In this respect, accurate chemical measurements, carried out during the PHEBUS-FP experiments, will be particularly valuable for the development of new nuclear power plants. Mr. M.Akiyama The current PHEBUS-FP experimental programme was, as already stated, entirely devoted to the present type of reactor design. However, once the analytical tools are validated by the PHEBUS experiments, they can surely be applied to a certain extent also to future reactor designs. With this in mind, future reactor safety could be discussed within the scope of the PHEBUS-FP programme. Mr. B.R.Bowsher, AEA Technology Winfrith Experiments such as PHEBUS-FP always represent a balance between demonstration and the study of phenomena. In PHEBUS-FP it is proposed to vary a large number of parameters between each test (i.e. temperature, oxidation, pressure, sump pH, geometry, additives, etc.). This necessitates supplementary separate effect studies (both analytical and experimental). However, the interpretation of the results will remain a rather complex and, therefore, uncertain issue. I wonder whether we could not gain more from the experiments if we would alter the balance slightly in favour of the phenomenological issues, e.g. fixed geometry and varying injection of aerosols between the tests. Mr. H.F.Holtbecker This remark represents actually the major concern of the programme management. It has the rather delicate task to propose a scientifically valid test programme which covers all parameters affecting the major phenomena involved. Opinions in this respect vary from suggesting drastically simplified experiments up to reactor representative tests. To encounter this dilemma, the project management is trying to learn from each experiment and to take separate efect tests for interpretation judgements into consideration. Mr. A.Alonso-Santos, Universidad Politecnica de Madrid What actions are undertaken at the Community level to make sure that the knowledge gained from the PHEBUS-FP project will reach in an appropriate form the reactor operators, for accident management aspects, the reactor designers and engineering companies involved, for safety technology considerations and the educational institutions, training the next generation of nuclear scientists and technicians?

248

THE PHEBUS FISSION PRODUCT PROJECT

Mrs. C.Lecomte With regard to the French situation the transfer of knowledge to the utilities is assured by direct contacts. Mr. H.F.Holtbecker On the Community level an effort has to be made to convene potential customers of the results in order to assure an optimal approach and a scientifically valid programme. In a first step we should try to identify these potential customers inside and possibly outside the Community. We could jointly review the experimental programme in view of the resources available and the most urgent needs in the source term area. Such an approach would assure right from the beginning of the project an ongoing dialogue and the transfer of results to the customers of concern.

CLOSING REMARKS

Mr. Holtbecker closed the seminar with the following remarks: I think all persons involved in the project can feel encouraged by the results of this seminar. I would like to thank those who prepared or contributed to the seminar, i.e. the coordinators, authors and speakers, chairmen, panel members and organizers. We return home from this seminar motivated to continue our work and with a lot of new ideas yet to be digested. It is the intention to meet again sometime between the first and the second experiment, to discuss first results and review jointly the test matrix. In the meantime, the project should take the necessary steps to launch contacts with the potential customers of the results. I also would like to thank all participants joining the seminar and contributing to the fruitful discussions. I think we can congratulate those directly in charge for the excellent organization of the seminar. Since all of them are present, let us thank them collectively.

LIST OF PARTICIPANTS ABECASSIS MICHEL CEA, DRS/SEA CEN CADARACHE, DRS/SEA, BAT. 721 F—13108 ST-PAUL-LEZ-DURANCE ADROGUER BERNARD CEA CEN CADARACHE, DRS/SEMAR, BAT. 702 F—13108 ST-PAUL-LEZ-DURANCE AKIYAMA MAMORU UNIVERSITY TOKYO 7–3–1, KONGO, BUNKYO-KU J—TOKYO ALBIOL THIERRY CEA CEN CADARACHE, DRS/SREAS, BAT. 346 F—13108 ST-PAUL-LEZ-DURANCE ALEZA SANTIAGO UNIVERSIDAD POLITECNICA, ETS II CATE. TECHNO. NUCLEAR C/ JOSE GUTIERRES ABASCAL, 2. E—28006 MADRID ALLELEIN HANZ-JOSEF GRS-KOELN SCHWERTNERGASSE 1 D—5000 KOLN 1 ALONSO AGUSTIN UNIVERSIDAD POLITECNICA CATE. TECHNO. NUCLEAR ETS. INGENIEROS INDUSTRIALES C/JOSE GUITIERRES ABASCAL, 2 E—28006 MADRID ARBOUSSET ANNIE CEA CEN CADARACHE, DRS/SREAS, BAT. 721 F—13108 ST-PAUL-LEZ-DURANCE AREIA CAPITAO JOAQUIM CEC—JRC JOINT RESEARCH CENTER T.P. 250 I—21020 ISPRA (VA) ARNAUD ANDRE CEA CEN CADARACHE, DRS/SEMAR, BAT. 702 I—13108 ST-PAUL-LEZ-DURANCE ARTINGSTALL GRAHAM H. S. E. RLSD, BROAD LANE UK—S3 7HQ SHEFFIELD AUJOLLET JEAN MARIE CEA, DERS/SREAS CEN CADARACHE, DRS/SREAS, BAT. 721 F—13108 ST-PAUL-LEZ-DURANCE BAILLY JEAN CEA CEN CADARACHE, IPSN, BAT. 250 F—13108—ST-PAUL-LEZ-DURANCE BALOURDET MARCEL CEA CEN CADARACHE, DRS/SEMAR, BAT. 702 F—13108 ST-PAUL-LEZ-DURANCE BALZ WERNER CEC RUE DE LA LOI, 200 B—1049 BRUSSELS BANASCHIK MANFRED VIKTOR GRS SCHWERTNERGASSE 1 D—5000 KOELN 1 BANDINI GIACOMINO ENEA VIA MARTIRI DI MONTE SOLE, 4 I—40129 BOLOGNA BARTHELEMY GILLES CEA

LIST OF PARTICIPANTS

CEN CADARACHE DRS/SREAS, BAT. 721 F—13108 ST-PAUL-LEZ-DURANCE BASANSKI EVGENI K. I. A. E. MOSCOW KURCHATOV SQ URSS—123182 MOSCOW BELOVSKY LADISLAV NUCLEAR RESEARCH INSTITUE NRI RES, DEPARTEMENT 205 CS—25068 REZ PR BENSON CHRISTOPHER GORDON ISPRA—JRC TP060, CCR I—21020 ISPRA (VARESE) BERLIN CLAUDE CEA CEN CADARACHE, DEC/SECA, BAT. 238 F—13108 ST-PAUL-LEZ-DURANCE BERLIN MARGUERITE CEA CEN CADARACHE, DRS/SREAS, BAT. 346 F—13108 ST-PAUL-LEZ-DURANCE BEYLY DIDIER CEA CEN CADARACHE, DRS/SREAS, BAT. 721 F—13108 ST-PAUL-LEZ-DURANCE BOURDON SERGE CEA CEN CADARACHE, DRS/SEMAR, BAT. 702 F—13108 ST-PAUL-LEZ-DURANCE BOWSHER BRIAN AEA TECHNOLOGY WINFRITH TECHNOLOGY CENTRE UK—DORCHESTER, DORSET DT2 8DH CARLUCCI LIBERATO AECL-RESEARCH CHALK RIVER NUCLEAR LABORATORIES CANADA—KOJ 1JO CHALK RIVER, ONTARIO CARTA MARIO ENEA CRE CASACCIA VIA ANGUILLARESE 301 I—00060 S. MARIA DI GALERIA (ROMA) CHALOT ANDRE CEA CEN CADARACHE, DRS, BAT. 250 F—13108 ST-PAUL-LEZ-DURANCE CEDEX CODRON LUC CEA CEN CADARACHE, DRS/SEA, BAT. 721 F—13108 ST-PAUL-LEZ-DURANCE CONTZEN JEAN PIERRE JRC/CEC 200, RUE DE LA LOI B—1049 BRUXELLES COQUERELLE MICHEL CCE CCR INSTITUT TRANSURANIENS POSTFACH 23 D—7500 KARLSRUHE CRANGA MICHEL CEA CEN CADARACHE, DRS/SEMAR, BAT. 729 F—13108 ST-PAUL-LEZ-DURANCE CRESTIA JEAN-CLAUDE CEA CEN CADARACHE, DRS/SEMAR, BAT. 702 F—13108 ST-PAUL-LEZ-DURANCE DADILLON JAMES CEA CEN CADRACHE, DRS/SEA, BAT. 721 F—13108 ST-PAUL-LEZ-DURANCE DAGUZAN-LEMOINE FRANCETTE CEA

251

252

THE PHEBUS FISSION PRODUCT PROJECT

CEN CADARACHE, DRS/AAF, BAT. 702 F—13108 ST-PAUL-LEZ-DURANCE DE ACHA ANTONIO CONSEJO DE SEGURIDAD NUCLEAR CALLE JUSTO DORADO, 11 E—28040 MADRID DE BOECK BENOIT AIB VINCOTTE NUCL. 157 AV. DU ROI B—1060 BRUXELLES DE ROSA FELICE ENEA VIA MARTIRI DI MONTE SOLE, 4 I—40129 BOLOGNA DEL NEGRO ROLAND CEA CEN, CADARACHE, DRS, BAT. 721 F—13108 ST-PAUL-LEZ-DURANCE DELBRASSINE ALBERT CEN/SCK 200, BOERETANG B—2400 MOL DELCHAMBRE PHILIPPE CEA CEN CADARACHE, DRS/SEA, BAT. 721 F—13108 ST-PAUL-LEZ-DURANCE DELLA LOGGIA VINCENZO ENZO CEC—DG XII 200 RUE DE LA LOI B—1049 BRUXELLES DICKSON LAWRENCE AECL RESEARCH CHALK RIVER LABORATORIES CANADA—CHALK RIVER, ONTARIO DOUGNAC JEAN-CLAUDE CEA CEN CADARACHE, DRS/SEA, BAT. 721 F—13115 ST-PAUL-LEZ-DURANCE EDDI MICHEL MINISTERE RECHERCHE ET TECHNOLOGIE 1 RUE DESCARTES F—75005 PARIS EMAIN PHILIPPE EDF/SEPTN 12–14, AV.DUTRIEVOZ F—69628 VILLEURBANNE CEDEX FASOLI STELLA PAOLA CEC JRC ISPRA THERMODYNAMICS DIVISION JRC T.P. 421 I—21020 ISPRA (VA) FEHRENBACH PAUL AECL RESEARCH CHALK RIVER LABORATORIES CANADA—KOJ 150 CHALK RIVER, ONTARIO FINZI SERGIO C. C. E 200, RUE DE LA LOI B—1049 BRUXELLES FRUTTUOSO GIANCARLO DCMN UNIVERSI. PISA VIA DIOTISALVI2 I—56100 PISA FUJISHIRO TOSHIO JAERI NAKA-GUN, IBARAKI-KEN J—TOKAI-MURA FURRER MAX P.SCHERRER INSTITUTE CH—5232 VILLIGEN/PSI GAUVAIN JEAN CEA CENFAR, DPEI, BP. N°6 F—92265 FONTENAY-AUX-ROSES GOETZMANN ODO

LIST OF PARTICIPANTS

KERNFORSCHUNGSZENTRUM KARLSRUHE GMB POSTFACH 3640 D—7500 KARLSRUHE 1 GOMOLINSKI MAURICE CEA/IPSN BP n°6 F—92265 FONTENAY-AUX-ROSES GONNIER CHRISTIAN CEA CEN CADARACHE, DRS/SEA, BAT. 721 F—13108 ST-PAUL-LEZ-DURANCE CEDEX GONZALEZ JAVIER UNIVERSIDAD POLITECNICA GATE. TECHNO. NUCLEAR, ETS II C/JOSE GUTIERREZ ABASCAL, 2, E—28006 MADRID GONZALEZ EDUARDO CONSEJO DE SEGURIDAD NUCLEAR CALLE JUSTO DORADO, 11 E—28040 MADRID GUILLEMARD BERNARD CEA CEN FAR IPSN, BP. N°6 F—92265 FONTENAY-AUX-ROSES GUNTAY KAMIL SALIH GROUP LEADER P.SCHERRER INSTITUTE CH—5232 VILLIGEN PSI GYENES G.I. CENTRAL RESEARCH INSTITUTE H—BUDAPEST HAAPALEHTO TIMO LABORATORY MANAGER LTKK P.O. BOX20 SF—53851LPR HACHE GEORGES CEA CEN CADARACHE, DRS/SEMAR BAT. 702 F—13108 ST-PAUL-LEZ-DURANCE HAESSLER MAURICE CEA CEN CADARACHE, DRS/SEA, BAT. 721 F—13108 ST-PAUL-LEZ-DURANCE HAGEN SIEGFRIED KERNFORSCHUNGSZENTRUM KARLSRUHE GMB POSTFACH 3640 D—7500 KARLSRUHE HAYNS MICHAEL AEA TECHNOLOGY B.329 HARWELL LABORATORY, DIDCOT, UK—OXON OX11 ORA HERRANZ PUEBLA LUIS E. CIEMAT AVDA. COMPLUTENSE, 22 E—28040 MADRID HOFMANN PETER KERNFORSCHUNGSZENTRUM KARLSRUHE POSTFACH 3640 D—7500 KARLSRUHE 1 HOLMSTROM HEIKKI OECD/NEA 38, BOULEVARD SUCHET F—75016 PARIS HOLTBECKER HELMUT CCR—ISPRA INSTITUT DE TECHNOLOGIE DE LA SURETE I—21020 ISPRA (VA) HUEBER CLAUDE CEA CEN CADARACHE, DRS/SEMAR, BAT. 702

253

254

THE PHEBUS FISSION PRODUCT PROJECT

F—13108 ST-PAUL-LEZ-DURANCE, CEDEX HUET YVES FRAMATOME TOUR FIAT 1, PLACE DE LA COUPOLE F—92084 PARIS LA DEFENSE CEDEX 16 IIJIMA TOSHINORI NUPEC/JINS FUJITAKANKO-TOYANOMON BLDG. 7F, 3–17 J—TOKYO ISMUNTOYO ROBERTUS, P.H. NATIONAL ATOMIC ENERGY AGENCY PPTKR-BATAN SERPONG, KAWASAN PUSPIPT INDONIESIE—TANGERANG JAWA-BARAT. IVANOV VICTOR MINISTRY OF NU. INDU. AND POWER STAROMONETNYI PER. URSS—109180 MOSCOW JACQUIN MICHELE EDF COMITE DE RADIOPROTECTION, 3 F—75384—PARIS CEDEX 08 JAMOND CLAUDE CEA CEN CADARACHE, DRS/SEA, BAT. 721 F—13108 ST-PAUL-LEZ-DURANCE JIMENEZ JOSE CIEMAT AVDA. COMPLUTENSE, 22 E—28040 MADRID JONES ALAN V. CEC-JRC JOINT RESEARCH CENTER, A 650, I—21020 ISPRA (VA) JUDE PATRICE FRAMATOME TOUR FIAT F—92084 PARIS LA DEFENSE KANIJ JOHAN NV KEMA PO BOX 9035 NL-6800 ET ARNHEM KAYSER GASTON CCR CEN CADARACHE, DRS/SEMAR, BAT. 702 F—13108 ST-PAUL-LEZ-DURANCE KERBOUL CLAIRE CEA-DSE/DSP 31–33, RUE DE LA FEDERATION BP 510 F—75752 PARIS CEDEX 15 KINNERSLY STEPHEN AEA TECHNOLOGY WINFRITH, UK—DT2 8DH DORCHESTER, DORSET KISELEV WLADIMIR IBRAE INSTITUT DE SURETE DE L’ACADEMIE URSS—11319 MOSCOU KRISCHER WOLFGANG CEC—JRCISPRA I—21020 ISPRA (VA) KRONENBERG JURIS IKE, UNIVERSITAET STUTTGART PFAFFENWALDRING 31 D—7000 STUTTGART KTORZA CHANTAL CEA CEN CADARACHE, DRS/SEA, BAT. 721 F—13108 ST-PAUL-LEZ-DURANCE KUCZERA BERNHARD KERNFORSCHUNGSZENTRUM KARLSRUHE POSTFACH 3640

LIST OF PARTICIPANTS

D—7500 KARLSRUHE 1 KUJAL BOHUMIR NUCLEAR RESEARCH INSTITUTE REZ CS—25068 PRAGUE L’HOMME ALAIN CEA CEN FAR, IPSN/DPEI, BP. N° 6 F—92265 FONTENAY-AUX-ROSES LAYLY VICTOR CEA CEN CADARACHE, DRS/SEMAR, BAT. 729 F—13108 ST-PAUL-LEZ-DURANCE LE BORGNE EDMOND CEA CEN CADARACHE, DEC/SECA F—13108 ST-PAUL-LEZ-DURANCE LE ROUX GEORGES FRAMATOME TOUR FIAT, 1, PLACE DE LA COUPOLE F—92084 PARIS LA DEFENSE CEDEX 16 LECOMTE CATHERINE CEA CEN FAR, IPSN/DPEI/SEAC BP N°6 F—92265 FONTENAY-AUX-ROSES LHIAUBET GILLES CEA CEN FAR, IPSN/DPEI/SEAC BP. N° 6 F—92265 FONTENAY-AUX-ROSES LILJENZIN JAN-OLOV DEPARTMENT OF NUCLEAR CHEMISTRY, CT S—41296 GOTEBORG LIVOLANT MICHEL CEA CEN FAR, IPSN BP. N° 6 F—92265 FONTENAY-AUX-ROSES LOBODA SERGEJ KURCHATOV INS. ATOMIC ENERGY KURCHATOV SQ URSS—123182 MOSCOW MAC DONALD RODERICK D. AECL RESEARCH CHALK RIVER LABORATORIES CANADA—DN3378 CHALK RIVER, ONTARIO MAEGEY MICHEL CEA CEN CADARACHE, DRS/SREAS, BAT. 721 F—13108 ST-PAUL-LEZ-DURANCE MAILLIAT ALAIN CEA CEN CADARACHE, DRS/SEMAR, BAT. 702 F—13108 ST-PAUL-LEZ-DURANCE MARKOVINA ANTONIO CEC—JRCISPRA JRC ISPRA I—21020 ISPRA (VA) MARTIN ESPIGARES MANUEL CIETMAT AVDA. COMPLUTENSE, 22 E—28040 MADRID MATZKE HANSJOACHIM EUROPEAN INSTITUTE FOR TRANSUR POSTFACH 2340 D—7500 KARLSRUHE MEYER-HEINE ANTOINE CEA CEN CADARACHE, DRS, BAT. 250 F—13108 ST-PAUL-LEZ-DURANCE MORONI MAX CEA CEN CADARACHE, DRS/SREAS, BAT. 721

255

256

THE PHEBUS FISSION PRODUCT PROJECT

F—13108 ST-PAUL-LEZ-DURANCE MOTTE FRANCOIS CEN/SCK BOERETANG 200 B—2400 MOL NAKAMURA TAKEHIKO JAERI IBARAKI-KEN 310 J—TOKAI-MURA NERVI JEAN-CLAUDE CEA CEN CADARACHE, DRS/SREAS, BAT. 721 F—13108 ST-PAUL-LEZ-DURANCE NICHOLS ALAN LESLIE AEA TECHNOLOGY WINFRITH, UK—DT2 8DH DORCHESTER (DORSET) NICOLAY DENIS CCR BAT.J MONNET B4/077 L—2920—LUXEMBOURG NODA KOICHI M. I. T. I. 1–3–1, KASUMIGASEKI, CHIYODA-KU J—TOKYO NOVIKOV ALEXANDER MINISTRY OF NUCLEAR IND. POWER STAROMONETNYI PER. URSS -109180 MOSCOU ORIOLO FRANCESCO DCMN UNIVERSITY OF PISA VIA DIOTISALVI 2 I—56100 PISA PALLER ANDREAS IKE, UNIVERSITAET STUTTGART PFAFFENWALDRING 31 D—7000 STUTTGART PALOMO MORALES JOSE HIDROELECTRICA ESPANOLA HERMOSILLA, 3 E—MADRID PAZDERA F. NUCLEAR RESEARCH INSTITURE CS—250 68 REZ PR PERRET COURT GENEVIEVE CEA CEN CADARACHE, DRS/SEA, BAT. 721 F—13108 ST-PAUL-LES-DURANCE PEZZILLI MASSIMO ENEA CRE CASACCIA VIA ANGUILLARESE, 301 I—00060 S MARIA DI GALERIA (ROME) PORRACHIA ALAIN CEA CEN CADARACHE, DRS/SEMAR, BAT. 729 F—13108 ST-PAUL-LEZ-DURANCE PUGA JOSE UNESA C/FRANCISCO GERVAS, 3 E—28020 MADRID QUENIART DANIEL CEA CEN/FAR IPSN, B.P. N° 6 F—92265 FONTENAY-AUS-ROSES RASTOIN JEAN CEA CENFAR, IPSN, B.P. N°6 F—92265 FONTENAY-AUX-ROSES REOCREUX MICHEL CEA CEN CADARACHE, DRS/SEMAR, BAT. 729 F—13108 ST-PAUL-LEZ-DURANCE REPETTO GEORGES CEA CEN CADARACHE, DRS/SEA, BAT. 721

LIST OF PARTICIPANTS

F—13108 ST-PAUL-LEZ-DURANCE RONGIER CATHERINE CEA CEN CADARACHE, DRS/SEMAR, BAT. 702 F—13108 ST-PAUL-LEZ-DURANCE ROSSINI JEAN PIERRE CEA CEN CADARACHE, DRS/SREAS, BAT. 721 F—13108 ST-PAUL-LEZ-DURANCE ROYEN JACQUES OECD 38, BOULEVARD SUCHET F—75016 PARIS RUBINSTEIN MARIE-CLAIRE CEA CEN CADARACHE, DRS/SEMAR, BAT. 702 F—13108 ST-PAUL-LEZ-DURANCE SABATHIER FRANCIS CEA CEN CADARACHE, DRS/SREAS, BAT. 721 F—13108 ST-PAUL-LEZ-DURANCE SAIRANEN RISTO RESEARCH ENGINEER TECHNICAL RESEARCH CENTER OF FINLAN SF—00181 HELSINKI SANDERVAG ODDBYORN SWEDISH NUCLEAR POWER INSPECTO SEHLSTEDTGT 11 S—10252 STOCKHOLM SANDRELLI GIANCARLO ENEL/CRTN VIA MONFALCONE 15 I—20132 MILAN SANG-BAIK KIM KAERI PO BOX 7, DAEDUK-DANJI, COREE—305–306 TAEJON SASAJIMA HIDEO JAERI DFSR, NAKA-GUN, IBARAKI-KEN J—319–11TOKAI-MURA SCHIMETSCHKA EDGAR BATELLE EUROPE AM ROMERHOF 35 D—6000 FRANKFURT/MAIN SCHMITZ FRANZ EURATOM CEN CADARACHE, DRS/SEMAR, BAT. 702 F—13108 ST-PAUL-LEZ-DURANCE SCHOECK WERNER LABORA.AEROSOLSPHY.FILTERTECH. KERNFORSCHUNGSZENTRUM KARLSRUHE, POSTFACH 3640 D—7500 KARLSRUHE 1 SHUMSKI ANATOLI OKB “GIDROPRESS” URSS—PODOLSK SPEIS THEMIS US/NRC MAIL STOP NL/S-007 USA—20555 WASHINGTON DC STEMPNIEWICZ MAREK IEA INSTITUTE OF ATOMIC ENERGY PL—05400 SWIERK TAKUMI KENJI NUPEC 3–13, 4-CHOME, TORANOMON, MINATO-KU J—TOKYO TANI AKIRA NUPEC 3–13, 4-CHOME, TORANOMON MINATO-KU J—TOKYO TATTEGRAIN ALAIN

257

258

THE PHEBUS FISSION PRODUCT PROJECT

CEA CEN CADARACHE, DRS, BAT. 250 F—13108 ST-PAUL-LEZ-DURANCE TATTEGRAIN BETTY CEA CEN CADARACHE, DRS/SEA, BAT. 721 F—13108 ST-PAUL-LEZ-DURANCE TESCHENDORFF VICTOR H. GESELLSCHAFT REAKTORSICHERHEIT GRS, FORSCHUNGSGELANDE D—8046 GARCHING TIRINI SANDRO ENEA VIA MARTIRI MONTE SOLE,4 I—40129 BOLOGNA TONELLO DAVERIO MARIA ELENA CEC JRS ISPRA THERMODYNAMICS DIV. JRC T.P. 421 I—21020 ISPRA (VA) TOTH B. CENTRAL RESEARCH INSTITUT H—BUDAPEST TURVEY FRANK NUCLEAR ENERGY BOARD 3, CLONSKEAGH SQUARE, CLONSKEAGH RO IRL—14 DUBLIN VALACH MOJMIR NUCLEAR RESEARCH INSTITU. REZ NRI REZ DEPARTMENT 205, CS—25068 REZ PR VAN RIJ HENK M CEC—JRC ISPRA TP 60, CCR I—21020 ISPRA (VA) VANEL MICHEL CEA CEN CADARACHE, DRS, BAT. 250 F—13108 ST-PAUL-LEZ-DURANCE VARET JACQUES MINISTERE RECHERCHE ET TECHNOL 1, RUE DESCARTES F—75231 PARIS CEDEX 05 VICENTE CARMEN CIEMAT AVDA. COMPLUTENSE, 22 E—28040 MADRID VIGNESOULT NICOLE CEA CEN FAR, IPSN, BP. N° 6 F—92265 FONTENAY-AUX-ROSES VON DER HARDT PETER CEC—JRC ISPRA CEN CADARACHE, DRS, BAT. 721 F—13108 ST-PAUL-LEZ-DURANCE CEDEX WADSWORTH STEPHEN L. ONTARIO HYDRO 760 UNIVERSITY AVENUE, CANADA—TORONTO, ONTARIO WARLOP RAYMOND CEA CEN GRENOBLE BP 85 F—38041 GRENOBLE WILLIAMS DEWIARFON AEA TECHNOLOGY 211/A32, WINFRITH, UK—DT2 8DH DORCHESTER, (DORSET) WRIGHT ROBERT WILLIAM US/NRC USA—20555 WASHINGTON, DC ZAMMITE RENE CEA CEN FAR, IPSN, BP. N° 6 F—92265 FONTENAY-AUX-ROSES ZEYEN ROLAND

LIST OF PARTICIPANTS

CCR ISPRA CEN CADARACHE, DRS/SEA, BAT. 721 F—13108 ST-PAUL-LEZ-DURANCE

259

INDEX OF AUTHORS

ADROGUER, B., 119 AKIYAMA, M., 32, 319 ARNAUD, A., 144

SOFFER, L., 14 SPEIS, T.P., 14, 319 TAKUMI, K., 32 TATTEGRAIN, A., 216, 314

BANASCHIK, M., 319 CONTZEN, J.P., 3

VANGEEN, R.E., 319 VAN RIJ, H.M., 139 VILLALIBRE, P., 119 VON DER HARDT, P., 159, 171, 216

DELCHAMBRE, PH., 159 FASOLI-STELLA, P., 204, 222 FINZI, S., 45

WRIGHT, R.W., 49

GAUVAIN, J., 139, 243 GEOFFROY, G., 108 GONNIER, C., 108 HAGEN, S.J.L., 49 HAYNS, M.R., 227, 319 HOLTBECKER, H.F., 319 HUEBER, C., 85 JONES, A.V., 261, 276 KINNERSLY, S.R., 227 LECOMTE, C., 190, 319 LEE, R.Y., 14 LHIAUBET, G., 171, 190 LILJENZIN, J.O., 64 LIVOLANT, M., 1, 9 MAILLIAT, A., 276 MARKOVINA, A., 144, 204 MEYER, R.O., 14 MEYER-HEINE, A., 136 NICHOLS, A.L., 85 PEZZILLI, M., 319 REOCREUX, M., 243 REPETTO, C., 108 SCHÖCK, W., 64 SHEPHERD, I.M., 261, 276 SODA, K., 32

260