Applications of Circulation Control Technologies
Edited by Ronald D. Joslin Office of Naval Research Arlington, Virgini...
5 downloads
349 Views
41MB Size
Report
This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form
Applications of Circulation Control Technologies
Edited by Ronald D. Joslin Office of Naval Research Arlington, Virginia Gregory S. Jones NASA Langley Research Center Hampton, Virginia
Volume 214 PROGRESS IN ASTRONAUTICS AND AERONAUTICS Frank K. Lu, Editor-in-Chief University of Texas at Arlington Arlington, Texas
Published by the American Institute of Aeronautics and Astronautics, Inc. 1801 Alexander Bell Drive, Reston, Virginia 20191-4344
American Institute of Aeronautics and Astronautics, Inc., Reston, Virginia 1 2 3 4 5 Copyright 0 2006 by the American Institute of Aeronautics and Astronautics, Inc. Printed in the United States of America. All rights reserved. Reproduction or translation of any part of this work beyond that permitted by Sections 107 and 108 of the U.S. Copyright Law without the permission of the copyright owner is unlawful. The code following this statement indicates the copyright owner’s consent that copies of articles in this volume may be made for personal or internal use, on condition that the copier pay the per-copy fee ($2.50) plus the per-page fee ($0.50) through the Copyright Clearance Center. Inc., 222 Rosewood Drive, Danvers, Massachusetts 01923. This consent does not extend to other kinds of copying, for which permission requests should be addressed to the publisher. Users should employ the following code when reporting copying from the volume to the Copyright Clearance Center: 1-56347-789-0/06 $2.50
+SO
Data and information appearing in this book are for informational purposes only. AIAA is not responsible for any injury or damage resulting from use or reliance, nor does AIAA warrant that use or reliance will be free from privately owned rights.
ISBN 1-56347-789-0
Progress in Astronautics and Aeronautics Editor-in-Chief Frank K. Lu University of Texas at Arlington
Editorial Board David A. Bearden The Aerospace Corporation
Abdollah Khodadoust The Boeing Company
John D. Binder viaSolutions
Richard C. Lind University of Florida
Steven A. Brandt U.S. Air Force Academy
Richard M. Lloyd Raytheon Electronics Company
Fred R. DeJamette North Carolina State University
Frank Pai University of Missouri-Columbia
Gail Klein Jet Propulsion Laboratory
Ning Qin University of Shefield
George Eitalbery German-Dutch Wind Tunnels
US.Naval Postgraduate School
Sanjay Garg NASA Glenn Research Center
Ben T. Zinn Georgia Institute of Technology
Eswar Josyula
Peter H. Zipfel U.S. Air Force Research Laboratory
US.Air Force Research Laboratory
Oleg Yakimenko
Foreword
T
HIS collection of papers represents a compilation of the state-of-the-art in circulation-control technologies by two of the foremost experts in the field. The volume is conveniently organized to enable experts and beginners alike to quickly obtain a thorough historical overview and then be brought up to speed on the latest research. The final chapter delves into new areas and draws attention to exciting new ideas in circulation control. A wide range of advanced experimental and numerical methods are discussed by a panel of international experts. The text will prove to be of great value to workers in this field.
Frank K. Lu Editor-in-Chief Progress in Astronautics and Aeronautics
Preface
T
HE GENESIS of this volume originated during the planning of the NASA/ ONR Circulation Control Workshop, which was held March 2004 in Hampton, Virginia. Over two full days, 30 papers and 4 posters were presented, with 110 scientists, engineers, and program managers in attendance. This book was conceived to distribute this rich body of technical information on circulation control to a broader audience and to provide historical documentation to support future circulation control applications. Since that workshop, the papers have been updated and peer-reviewed to arrive at a compilation of the state of the art in circulation-control technologies. The goals of this book are 1) to summarize the history and the state of the art in circulation control technology, 2) to provide a single up-to-date knowledgebase for circulation control design, analysis, and experimental testing, and 3) to highlight prediction tools for circulation control. Goals 1 and 2 are clearly achieved in the chapters by the diverse applications and significant breadth of insights offered by the experts in this field. Goal 3 is most notably achieved by the use and discussion of the diverse range of computational fluid dynamics (CFD) tools for circulation control. Results showing the successful prediction of performance and inadequacies of some predictions are presented for completeness. The book is divided into four sections. The first major section presents a historical overview of circulation control. Because the overview papers are very thorough, many of the remaining chapters present brief introductions. The second major section covers experiments and applications. Section I1 is divided into A. fundamental flow physics, B. aerospace applications, and C. nonaerospace applications. The third major section covers CFD-based prediction tools and some validation with experiments (most of which are detailed in Section 11). Section I11 is subdivided by the different predictive applications. Finally, the last section consists of a single chapter, which introduces a vision for the use of circulation control in a broad spectrum of nonvehicle applications. Although less rigorous than most chapters, this final chapter exposes the reader to some new insights into applications of circulation-control technologies.
Ronald D. Joslin Gregory S. Jones December 2005
xix
Table of Contents Preface
.............................................. I
.
.
xix
Overview
Chapter 1 Advantages of Combining BLC Suction with Circulation Control High-Lift Generation
............................
John L . Loth. West Virginia University. Morgantown. West Virginia
3
Nomenclature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Designing a CC Technology Demonstrator STOL Aircraft . . . . . . . . . . . . . . . . . . 5 1974 Flight Testing of the WVU CC Technology Demonstrator . . . . . . . . . . . . . . 12 1979 CC Flight Tests with a Grumman Aerospace A-6A . . . . . . . . . . . . . . . . . . 16 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
.
Chapter 2 Overview of Circulation Control Pneumatic Aerodynamics: Blown Force and Moment Augmentation and Modification as Applied Primarily to Fixed-Wing Aircraft Robert J. Englar. Georgia Institute of Technology. Atlanta. Georgia
.......
Nomenclature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Coanda Effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Applications of Circulation Control. Past and Present . . . . . . . . . . . . . . . . . . . . Powered Lift and Engine Thrust Deflection . . . . . . . . . . . . . . . . . . . . . . . . . . . Other Aircraft Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Nonflying Applications of Circulation Control . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
.
Chapter 3 Exploratory Investigations of Circulation Control Technology: Overview for Period 1987-2003 at NSWCCD
.......
23 23 24 25 28 48 53 57 63 64
69
Robin Imber. Naval Air Systems Command. Patuxent River. Maryland; Ernest Rogers and Jane Abramson. Naval Surface Warfare Center-Carderock Division. West Bethesda. Maryland Nomenclature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
ix
69 70
X
Dual-Slotted Cambered Airfoil (LSB) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Self-Driven Rotary Thruster (TIPJET) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Annular Wing (CC-Duct) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Circular Wing (CC-Disc) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Miniature Oscillatory Valve (CC-Valve) for Unsteady Wing Load Reduction . . . . . Dual-Slotted Low Aspect Ratio Wing (CC Hydrofoil) . . . . . . . . . . . . . . . . . . . . Status of Design Capability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
70 73 79 85 91 93 99 100 101
1I.A. Experiments and Applications: Fundamental Flow Physics
.
Chapter 4 Measurement and Analysis of Circulation Control Airfoils
.....................................
105
F. Kevin Owen. Complere Inc., Paczjic Grove. California; Andrew K. Owen. University of Oxford. Oxford. England. United Kingdom
Nomenclature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Experimental Details . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sample Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
.
Chapter 5
Some Circulation and Separation Control Experiments
105 106 107 107 112 112
..
113
Dino Cerchie. Eran Halfon. Andreas Hammerich. Gengxin Han. Lutz Taubert. Lucie.Trouve. Priyank Varghese. and Israel Wygnanski. University of Arizona. Tucson.Arizona Nomenclature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Discussion of Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
.
Chapter 6
Noise Reduction Through Circulation Control
Scott E. Munro. Krishan K . Ahuja. and Robert J. Englar. Georgia Institute of Technology. Atlanta. Georgia
113 114 118 162 164 164
........ 167
Nomenclature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Facilities and Instrumentation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Technical Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
167 168 169 171 173 174
xi Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
184 186 186
1I.B. Experiments and Applications: Aerospace
.
Chapter 7 Pneumatic Flap Performance for a Two-Dimensional Circulation Control Airfoil
..............................
191
Gregory S. Jones. NASA Langley Research Center. Hampton. Virginia Nomenclature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . NASA CC Requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Theoretical Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . GACC Airfoil Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Experimental Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Airfoil Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
.
Chapter 8 Trailing Edge Circulation Control of an Airfoil at Transonic Mach Numbers
..............................
191 192 193 195 202 207 216 236 237 241
245
Michael G. Alexander. Scott G . Anders. and Stuart K . Johnson. NASA Lungley Research Center. Hampton. Virginia Nomenclature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Model Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Instrumentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Facli ty . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Test Procedures and Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Test Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Discussion of Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
.
245 246 247 251 252 253 254 254 263 275 275
Chapter 9 Experimental and Computational Investigation into the Use of the Coanda Effect on the Bell A821201 Airfoil
........... 277
Gerald Angle 11. Brian O’Hara. Wade Huebsch. and James Smith. West Virginia University. Morgantown. West Virginia Nomenclature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Experimental Apparatus and Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . .
277 278 279
xii Computational Model and Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Experimental Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Computational Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
.
282 285 286 290 291
Chapter 10 Novel Flow Control Method for Airfoil Performance Enhancement Using Co-Flow Jet
................ 293
Ge-Cheng Zha and Craig D. Paxton. University of Miami. Coral Gables. Florida Nomenclature. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgments. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
.
293 294 296 311 312 312
Chapter 11 Experimental Development and Evaluation of Pneumatic Powered-Lift Super-STOL Aircraft
................ 315
Robert J. Englar. Georgia Institute of Technology. Atlanta. Georgia; Bryan A . Campbell. NASA Langley Research Center. Hampton. Virginia
Nomenclature. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Experimental Apparatus and Test Techniques. . . . . . . . . . . . . . . . . . . . . . . . . Wind-Tunnel Evaluations and Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Comparison of Measurements and Predictions . . . . . . . . . . . . . . . . . . . . . . . . Potential Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgments. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
.
Chapter 12
Use of Circulation Control for Flight Control
315 316 320 321 331 333 333 335 335
........ 337
Steven l? Frith and Norman J. Wood, University of Manchester. Manchester. England. United Kingdom
Nomenclature. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Half-Span Cropped-Delta Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Full-Span UAV Configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgments. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
337 338 339 345 352 353 35 3
xiii
1I.C. Experiments and Applications: Nonaerospace
.
Chapter 13 Pneumatic Aerodynamic Technology to Improve Performance and Control of Automotive Vehicles
.............. 357
Robert J. Englar. Georgia Institute of Technology. Atlanta. Georgia
Nomenclature. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Basics of Pneumatic Circulation Control Aerodynamics . . . . . . . . . . . . . . . . . . DOE Pneumatic Heavy Vehicle Model Test Results . . . . . . . . . . . . . . . . . . . . Pneumatic HV Fuel Economy Testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Updated Wind Tunnel Evaluations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Pneumatic Sport Utility Vehicles (PSUVs) . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgments. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
.
357 357 358 360 367 371 374 379 380 381 381
Chapter 14 Aerodynamic Heat Exchanger: A Novel Approach to Radiator Design Using Circulation Control Richard J. Gaeta. Robert J. Englar. and Graham Blaylock. Georgia Institute of Technology. Atlanta. Georgia
................ 383
Nomenclature. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Technical Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgments. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
383 383 386 389 395 397 397
1II.A. Tools for Predicting Circulation Control Performance: NCCR 1510 Airfoil Test Case
.
Chapter 15 Investigation of Turbulent Coanda Wall Jets Using DNSandRANS
.....................................
401
Hermann F. Fasel. Andreas Gross. and Stefan W e n . University of Arizona. Tucson. Arizona
Nomenclature. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Investigated Configurations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Numerical Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Turbulent Wall Jet on a Circular Cylinder . . . . . . . . . . . . . . . . . . . . . . . . . . . Circulation Control Airfoil . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
401 402 403 404 405 415
xiv Conclusions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
.
418 419 419
Chapter 16 RANS and Detached-Eddy Simulation of the NCCR Airfoil
421
Nomenclature. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Geometry. Conditions. and Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Computational Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Grid Generation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Initial and Boundary Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
421 422 424 425 427 429 430 441 442 442
.......................................
Eric G . Paterson and Warren J. Baker. Pennsylvania State University. University Park. Pennsylvania
.
Chapter 17 Full Reynolds-Stress Modeling of Circulation Control Airfoils
445
Nomenclature. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mathematical Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
445 446 448 453 465 465 465
.....................................
Peter A . Chang 111. Joseph Slomski. Thomas Marino. Michael P. Ebert. and Jane Abramson. Naval Surface Warfare Center-Carderock Division. West Bethesda. Maryland
1II.B. Tools for Predicting Circulation Control Performance: NCCR 103RE Airfoil Test Case
.
Chapter 18 Aspects of Numerical Simulation of Circulation Control Airfoils
469
Nomenclature. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . GeometryandGrid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
469 470 472
.....................................
R. Charles Swanson. Christopher L . Rumsey. and Scott G. Anders. NASA Langley Research Center. Hampton. Virginia
xv Numerical Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Boundary and Initial Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Turbulence Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Jet Momentum Coefficient . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Numerical Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix: Coordinates of 103RE Airfoil . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
.
Chapter 19 Role of Turbulence Modeling in Flow Prediction of Circulation Control Airfoils
.............................
475 476 476 478 478 495 497 497 497
499
Gregory McGowan. Ashok Gopalarathnam. Xudong Xiao. and Hassan Hassan. North Carolina State University. Raleigh. North Carolina Nomenclature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Formulation of the Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
499 500 501 502 510 510 510
1II.C. Tools for Predicting Circulation Control Performance: GACC Airfoil Test Case
.
Chapter 20 Simulation of Steady Circulation Control for the General Aviation Circulation Control (GACC) Wing
........... 513
Warren J. Baker and Eric G. Paterson. Pennsylvania State University. University Park. Pennsylvania
Nomenclature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Geometry. Conditions. and Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Computational Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Grid Generation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Initial and Boundary Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Computational Resources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
513 514 515 516 518 521 523 523 536 537 537
xvi
.
Chapter 21 Computational Study of a Circulation Control Airfoil Using FLUENT
................................
539
Gregory McGowan and Ashok Gopalarathnam. North Carolina State University. Raleigh. North Carolina Nomenclature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Configurations and Experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Numerical Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
539 540 541 542 545 552 553 553
1II.D. Tools for Predicting Circulation Control Performance: Additional CFD Applications
.
Chapter 22 Computational Evaluation of Steady and Pulsed Jet Effects on a Circulation Control Airfoil
.................. 557
Yi Liu. Lakshmi N . Sankar. Robert J. Englar. Krishan K . Ahuja. and Richard Gaeta. Georgia Institute of Technology. Atlanta. Georgia
Nomenclature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mathematical and Numerical Formulation . . . . . . . . . . . . . . . . . . . . . . . . . . . Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
.
557 558 559 561 575 575 575
Chapter 23 Time-Accurate Simulations of Synthetic Jet-Based Flow Control for a Spinning Projectile
.............. 579
Jubaraj Sahu. U S. Army Research Laboratory. Aberdeen Proving Ground. Maryland
Nomenclature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Computational Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Projectile Geometry and Computational Grid . . . . . . . . . . . . . . . . . . . . . . . . . Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
579 580 581 584 586 594 595
xvii
. Exploring a Visionary Use of Circulation Control Chapter 24. Coanda Effect and Circulation Control for Nonaeronautical Applications ............................
599
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
599 600 612 612 612
...............................................
615
Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
623
Supporting Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
625
IV
Terence R . Day. Vortex Dynamics Pty Ltd. Mount Tamborine. Queensland. Australia
Index
I. Overview
Chapter 1
Advantages of Combining BLC Suction with Circulation Control High-Lift Generation John L. Loth* West Virginia University, Morgantown, West Virginia
Nomenclature CB = circulation control blowing efficiency factor CL = lift coefficient CLopt= optimum lift coefficient where aircraft L / D is maximum C, = blowing coefficient Di = induced drag D,, = parasite drag mcc = circulation control blowing mass flow rate pt = total pressure in the compressor bleed air supply duct q, = dynamic pressure r = circulation control rounded trailing edge radius S, = wing area t,, = non-dimensional circulation control blowing slot height tn = non-dimensional ejector nozzle slot height ts = non-dimensional suction slot height V,, = circulation control blowing velocity V, = equivalent airspeed, corrected for position error Vi = indicated airspeed V , = free stream velocity p = air density Subscripts a = angle of attack c = chord co = free stream conditions
*F’rofessor, Mechanical and Aerospace Engineering. Associate Fellow AIAA. Copyright 0 2005 by the American Institute of Aeronautics and Astronautics, Inc. All rights reserved.
3
4
J. L. LOTH
I. Introduction HE PURPOSE of this paper is to present the advantages of combining boundary layer control (BLC) by suction with circulation control (CC) by blowing for aircraft high lift generation. In the short take-off and landing (STOL) mode, the sharp trailing edge of the wing must be converted into a rounded Coanda surface for CC blowing. Jet engine hot, high-pressure compressor bleed air is the most commonly used source for the blowing air. Ducting this hot, highpressure air to the CC blowing slot involves problems arising from factors such as duct size, weight, pressure loss, required insulation and thermal expansion joints, and jet engine take-off thrust loss. It is shown here how adding an ejector for BLC suction just upstream of the CC blowing slot can diminish the impact of the aforementioned problems. It can reduce the amount of compressor bleed air required, and thus duct size, by more than 50%, provide structural cooling, and improve the CC blowing to free stream velocity ratio, bringing it closer to four, where the theoretical lift augmentation ratio reaches a maximum. Flight test results using such a configuration are provided, together with solutions for in-flight transition from the CC rounded trailing edge to a sharp trailing edge for low-drag cruise. Data were collected in 1974 during flight testing of the first CC Technology Demonstrator Aircraft, at West Virginia University. The use of blowing air to augment airfoil lift had already been proposed',2 in the 1920s. D a ~ i d s o n ,in~ his 1960 British patent application, referred to the concept of blowing over a circular cylinder as circulation control (CC). To improve the lift-to-drag ratio, Kind and Maull? at Cambridge University, experimented with CC on elliptical airfoils. Kind is also credited with developing the first boundary layer theory for CC blowing to correlate his experimental results. At zero angle of attack, elliptic airfoils produce two nearly identical suction peaks at their leading and trailing edges; this results in an aft shift of the center of pressure and thus nose-down pitching moment. Typical streamlines for such an airfoil, computed by S h r e ~ s b u r yare , ~ shown in Fig. 1. A schematic of a CC blowing slot is shown in Fig. 2. Blowing air must be supplied uniformly to the blowing slot. By Coanda turning, the jet generates a high suction force on the rounded surface. The angular position of the lower surface stagnation point, where the Coanda jet separates when meeting the flow from below the airfoil, determines the circulation and lift produced. Even today, most disagreements between computational and experimental results are
T
Fig. 1 Computed streamlines for an elliptic airfoil?
COMBINING BLC SUCTION AND CC HIGH-LIFT GENERATION
5
High -pressure fluid flowing from
Fig. 2 Schematic of a CC trailing edge.
a result of the sensitive relationship between circulation and location of this lower surface stagnation point. In the late 1960s, Robert Williams,6 then at NSRDC (Naval Ship Research and Development Center), started to experiment with CC airfoils developed by Kind. Williams697investigated the feasibility of a heavy-lift helicopter with dual plenum elliptic rotor blades and valves to control CC blowing rate to allow high forward speed. In 1968, the Office of Naval Research (ONR) contracted, with West Virginia University (WVU), research on CC airfoils, including testing at high Reynolds number and away from wind tunnel wall interference. Loth and Fanucci considered the possibility of protruding a CC blown airfoil from the roof of one of the WVU flight test aircraft. However, this would not be safe, because the roll moment produced by such a CC airfoil would exceed the available aircraft aileron control. To satisfy the contract requirement of flight testing CC technology, they decided it would be safer to fly a fixed-wing aircraft with CC blown wings. In the case of blower failure, an elliptic airfoil would not be flyable; therefore, new CC wings were designed at WVU, which were in-flight convertible from a high-speed, low-drag, conventional sharp trailing edge to a rounded trailing edge with CC blowing during slow-speed flight testing. In the following five years, several such CC airfoils were tested at WVU in the wind tunnel there. A comparison of blowing air requirements and lift capability for various high-lift systems was completed in 1973, as shown in Fig. 3.' This indicates that CC high-lift generation is more conservative in blowing air requirement than other methods.
11. Designing a CC Technology Demonstrator STOL Aircraft A Bede-4 homebuilt kit was found to provide an economical and suitable frame for test flying a CC wing. The simplest arrangement for in-flight conversion from a sharp trailing edge to a rounded CC blown trailing edge was first investigated. This is a forward folding flap, which exposes its semicircular hinge to provide a rounded trailing edge for CC blowing, as shown in Fig. 4. Dr. Norio Inumaru, visiting WVU from NRL in Tokyo, Japan, designed its drooped leading edge to prevent leading edge stall at high lift. The test model
J. L. LOTH
6
Fig. 3 Performance comparison between powered high-lift systems.'
was fabricated by riveting a sheet metal cuff to the wing leading edge and filling its cavity with foam. In 1970, Model A wing was tested in CC mode in the twodimensional 8 x 10 ft NSRDC wind tunnel, both in the sharp and round trailing edge configurations (Fig. 5 ) . The test data for CL,a, and C , are shown in Fig. 6 . Below stall, in the angle of attack range - 2 < a < 8 deg, they could be
L.E. DROOP DESIGN
Fig. 4 WVU Model A CC wing, wind tunnel tested at NSRDC in 1970.
COMBINING BLC SUCTION AND CC HIGH-LIFT GENERATION
7
Fig. 5 WVU Model A wing: left, in cruise; right, in CC mode.
curve-fitted by a linear equation:
In CC mode, the Model A chord length was reduced to 88% relative to cruise mode. All test data shown are referenced to cruise chord length, which effectively lowers CLm,, for the Model A in CC mode. The sharp trailing edge wing had a , = 0.09. Thus, for curve fitting test data in CC two-dimensional value of C mode, ,C , was reduced to 0.09 x 88% = 0.08. Excellent curve fitting was obtained by replacing SCL/SC, with cB/&, where CB is constant and named the blowing efficiency factor. Model A wing test data with CB = 4.3 provided the best curve fit, as shown in Fig. 7 using:
The disappointing performance of the Model A wing prompted a new design called the Model B wing. Instead of folding the flap inward for CC high lift generation, its flap was folded out. The 20% longer chord length, as shown in Figs. 8 and 9, was expected to increase the CC blowing efficiency factor CB from 4.3 to 4.3 x (120%/ 88%) = 5.9. In the Model B wing, great care was taken to achieve blowing
0 -I
0.2
0.4
0.6
0.8
1.0
CP,,
Fig. 6 WVU Model A CC wing, 1970 wind tunnel test results.
a
J. L. LOTH
5 4
1
0
0
0.4
0.2
0.6
0.8
CU
Fig. 7 WVU Model A CC wing empirical curve fit with CB = 4.3.
slot uniformity. This was accomplished by machining and bolting segmented aluminum nozzle blocks to an aluminum 3-in.-diam air supply duct, which also served as the rounded CC trailing edge. This provided a uniform 0.012in.-wide primary blowing slot (Fig. 9). The WVU wind tunnel model tests on a two-dimensional version of the Model B wing are described in Refs. 9 and 10. When applied to the CC Technology Demonstrator aircraft, the source of blowing air had to be selected. Boasson, in his dissertation, proved theoretically that the lift augmentation ratio CBreaches a maximum when Vcc/Vm = 4.'' Such low CC blowing velocity requires a high mass flow rate. Then Ap, the duct friction loss inside the air supplying 3-in.-diameter CC rounded trailing edge,
L.E. DROOP DESIGN
Fig. 8 WVU Model B CC wing, wind tunnel tested at WVU and flown on the CC Technology Demonstrator.
COMBINING BLC SUCTION AND CC HIGH-LIFT GENERATION
9
SUCTloN BLC
Fig. 9 WVU Model B CC wing, wind tunnel tested at WVU and flown on the CC Technology Demonstrator.
would far exceed the total pressure required at the CC blowing slot! This prompted the selection of a high-pressure air source, an auxiliary power GTC 85-72 gas turbine. However, its bleed air temperature of about 300"F, created new problems to be solved. The 100-in.-long aluminum CC rounded trailing edges of the wings had to be mounted on sliding bearings to allow for up to 0.5 in. of thermal expansion. Cooling had to be provided to shield the fiberglass wing filled with fuel from the hot CC rounded trailing edge. Incorporating an ejector with suction slot just upstream of the CC blowing slot as shown in Fig. 9 solved this problem. This also reduced the required compressor bleed air mass flow by more than 50% and reduced the blowing velocity ratio V,,/Vm closer to its optimum value. The 3 in. air supply duct contained a bell crank to transfer the torque needed to stow the rounded CC trailing edge within the wing for low-drag, high-speed cruise. The following derivation is included to illustrate the reduction in required compressor bleed air required and increase in blowing momentum obtainable by combining an internal ejector with CC blowing. For simplicity, assume one-dimensional, incompressible flow and constant area ejector with negligible wall shear loss. The dimensionless area ratios used are equal to those in the model B wing. Defining ejector exit CC slot height by tcc= 1, and the dimensionless ejector nozzle slot height by dividing by the CC exit slot height as t, = The remaining area for the suction slot height is divided by exit slot height to give ts = The suction velocity V, is made dimensionless by dividing by the CC blowing velocity V,, = 1. Likewise, the nozzle velocity becomes V,. The subsonic incompressible flow exit boundary condition applies: p n = p s . The Bernoulli equation gives suction gage pressure as ps = -qs = -0.5pV;V:,. We next apply the following equations. Continuity equation:
a.
i.
tsvs
+ t,vn = 1
(3)
J. L. LOTH
10
Inserting values for slot height gives 3 4
-vs
1
+-v, 4
or
=1
v,
=4-3vs
Momentum equation: tsv,2+t,v,2+T= Ps
1
PVCC
(4)
Inserting values for slot height gives 3 1 -V,2+-V,2-0.5V,2= 4 4
1
Substituting V, from above gives the quadratic equation: 5V; - 12Vs
+6 = 0
Solving for Vs < V, gives Vs = 0.7 when inserted in the preceding relation, which gives V, = 1.9. This means that t,V, = x 1.9 = 47.5%; in other words, the nozzle needs to supply only 47.5% of the CC blowing air. The balance of the blowing air tsVs = x 1.9 = 52.5% is supplied by the BLC suction slot and need not be supplied through the CC rounded trailing edge duct. The CC jet exits at near ambient pressure with thrust Tcc = hccVCc.The required ejector nozzle thrust T, is only 0.83Tcc, This demonstrates that incorporating an ejector can 1) provide cooling by boundary layer suction, 2) increase CC blowing momentum by (1-0.83) or 17%, 3) lower the velocity ratio Vcc/VW to increase blowing efficiency factor CB.Furthermore, it reduces compressor bleed air mass flow rate required by 52.5% which lowered duct pressure loss with associated duct size and weight savings. In the WVU wind tunnel model tests, the availability of flap hinge suction also allowed the CC flap to be deflected up to 15 deg without stall for additional lift augmentation. Arrows in Fig. 11 highlight the special features of this aircraft. Arrows have been used to show the CC blowing slot on the top of the 3-in.diameter rounded trailing edge. Suction boundary layer control (BLC) is shown at the flap hinge. The pilot can dump the blowing air overboard by actuating an air dump valve to achieve Direct Lift Control, called (DLC) as indicated. A layout of the WVU CC Technology Demonstrator aircraft is shown in Fig. 10 with a GTC 85-72 gas turbine mounted in the rear passenger seat area. Note that the jet engine exhaust discharges upwards, to prevent igniting the blacktop on the parking area. In Fig. 12 are shown details of the flap retraction and extension mechanism by a two horsepower electric motor. It turns the CC air supply duct inside the
a
9
COMBINING BLC SUCTION AND CC HIGH-LIFT GENERATION
EMPTY WT-
11
1720 I@
W. BROSS 2400 Ib
Fig. 10 WVU CC Technology Demonstrator dimensions and layout.
cabin, which is welded by bell cranks to the two 3-in.-diameter CC rounded trailing edges. For increased roll control at low speed, the ailerons are drooped and blown with compressor bleed air supplied via small ejectors inside the cabin for cooling purposes. To increase aileron effectiveness, they are connected to a flow diverter valve, which alters the left and right wing blowing rate. The bolt shown in the air splitter tee serves as a hinge for the splitter valve inside this tee. Fences and structure used to strengthen the cavity at the bottom of the wing, into which the CC rounded trailing edge retracts.
BLC at flap hinge line
Fig. 11 WVU CC Technology Demonstrator location of CC, BLC, DLC, and space for flap folding.
12
J. L. LOTH
Fig. 12 Compressor bleed air enters into a worm-gear driven pipe, connected by bell cranks to the left and right CC rounded trailing edges.
111. 1974 Flight Testing of the WVU CC Technology Demonstrator Prior to 25 h of flight testing, which started on 10 April 1974, the CC slot was tested for blowing uniformity and its ejector for providing adequate cooling to the fiberglass wings. The flight tests, performed by test pilot Shawn Roberts, started with calibrating airspeed and position error, with the use of a Pitot tube mounted with a boom to the left wing tip (Fig. 13). This boom also contained angle of
Fig. 13 Large position error in cockpit speed indicator against equivalent airspeed based on boom-mounted pitot tube readings.
COMBINING BLC SUCTION AND CC HIGH-LIFT GENERATION
13
attack and yaw measuring instrumentation. The aircraft in flight is shown in Fig. 14, with the CC blowing flap deployed. A summary of the flight test data is shown in Fig. 15. Shown are three scales for the lift coefficient, all based on dynamic pressure q, calculated using equivalent airspeed and reduced to sealevel density. The left column indicates the trimmed-out aircraft CL. The middle column has the tail download added to the lift and is termed CL wing On the far right column is the maximum CL value, which occurred at the flap centerline. For example, the average wing lift coefficient increased from 2.0 without blowing, at C , = 0, to 4.3 with blowing at C, = 0.12. Near stall the difference in angle of attack was negligible, thus the blowing efficiency factor C,, from Eq. (2) can be solved from:
G
C, = ACLI C = (4.3 - 2 ) / m = 6.6
(5)
This is more than 10% better than could be expected by extrapolating the Model A test results for the increased chord length, or C, = 4.3 x (1.2/0.88) = 5.86. This improvement can be credited to the utilization of an ejector in the Model B wing. It is interesting to calculate the CC blowing air horsepower required if the blowing air were supplied at standard sea-level conditions. The blowing slot height of 0.050 in. along the two 100-in.-long CC flaps resulted in blowing slot area A,, = 10 in.’. Consider flight with the propeller at idle, with C, = 0.12 and CL = 3.8. From the definition of C, = T,,/(q,S,) and CL= L/(q,S,), calculate blowing momentum
Tcc = (0.1213.8) x (W = 2400 lb) = 76 lbf
(6)
Fig. 14 WVU CC Technology Demonstrator during 25 h of flight testing, starting 10 April 1974.
14
J. L. LOTH
r
w
3
0
f
L
J
J
U
0
t t
.OM
5.6
’
1.7
t.3
-
12
2.1
- LO
20
. !.I
I
1
I
1
q 5 30
Sp
473
-
Vi
KNOTS
Fig. 15 WVU CC Technology Demonstrator flight performance map with CC blowing efficiency factor C, = 6.6.
At sea level density, CC velocity would be
0.00237776 x (10/144)
)’”=678 ft/s
(7)
and mass flow rate would be rit = pAccVcc= 0.002377 x (10/144) x 678 = 0.1 12 slug/s
(8)
Then the blowing power kinetic energy required is equal to V2
l i z s = 0.112 x 0.5 x 6782 = 25742 ft.lbf/s = 46.8 hp 2
(9)
COMBINING BLC SUCTION AND CC HIGH-LIFT GENERATION
15
To minimize blowing power required," the CC blowing velocity V,, should equal 4 times the flight velocity V,. At a flight speed of 39 kn, V,, should then be: 4 x 39 = 156 kn = 260 ft/s. For T,, = 76 lbf, the CC blowing mass flow rate should be 76/260 = 0.288 slug/s or kinetic power required would be as low as 0.5 x 0.288 x 2602 = 9734 ft.lbf/s = 17.7 hp. This reduction in CC blowing power required demonstrates the advantage of optimizing the ratio V C C / V ~ .
The pilot was quite satisfied with the handling qualities, and surprised how well the direct lift control DLC valve worked to make correction on the glide angle on approach to landing without inducing significant attitude changes. The CC flap deployment and stowing process worked well and required less than a 17 lb change in stick force, as shown in Fig. 16. To significantly reduce the stick force during flap deployment or retraction, Loth'* filed U.S. Patent 4,600,172, which allows converting a Fowler flap into a CC rounded trailing edge flap by only folding out a rounded trailing edge, which also supplies the blowing air (Fig. 17). The BLC suction is sufficiently strong to hold the Fowler flap against the CC pipe without the need for mechanical attachments. The ability to stow away the CC rounded trailing edge for high-speed low-drag cruise is an important aspect for operation with circulation control high lift systems. Slow flight was the most challenging aspect of the flight test program. With the propeller at 135 hp, the aircraft could be slowed to 23.5 kn indicated, which corresponds to 33.2 kn calibrated airspeed. This corresponded to a trim lift coefficient of 5.1 and wing average lift coefficient of 5.6 while blowing at 13 psig. Then there is little or no power to spare to prevent the onset of stall, which always started with a rapid roll and up to 1000 ft of altitude loss. Clearly this represents flying on the backside of the power curve, as
FLAP
FULL OUT
FLAP FOLDING ANGLE /3*
w,T:\yE RETRACTED
Fig. 16 WVU CC Technology Demonstrator shows acceptable trim force during flap folding.
16
J. L. LOTH
Fig. 17 U.S. Patent 4,600,172 allows flap stowing with greatly reduced actuator torque.'*
shown in Fig. 18. Note it takes only half as much power to cruise twice as fast at 70 kn.
IV. 1979 CC Flight Tests with a Grumman Aerospace A-6A The WVU successful demonstration of CC flight on a fixed-wing aircraft motivated the Navy to contract with Grumman Aerospace to convert an A-6A bomber to STOL operation with CC blowing. The challenge of converting an existing large aircraft to operate with CC blowing far exceeded that of building the small WVU CC Technology Demonstrator from scratch. Bob Englar, at NSRDC, began, in 1974, a careful CC wind tunnel test program to cover the
Fig. 18 WVU CC Technology Demonstrator performance safety is limited by the effect of high induced drag on power required.
COMBINING BLC SUCTION AND CC HIGH-LIFT GENERATION
17
entire range of operational aspects for the A-6A. His well-publicized test results are currently considered to be the most reliable available data and to which computational solutions are being compared. 13- l7 Of particular interest is Englar’s16 “STOL Potential of the Circulation Control Wing for High-Performance Aircraft.” This contains the performance map of a wind tunnel study of the A-6A wing without a tail, as shown in Fig. 19. When linearized using Eq. (l), the best fit constants are CL, = 0.09 (per degree) and CB = 6.3, as shown in Fig. 20. These results are similar to those found for the WVU CC Technology Demonstrator, although the A-6A has a greater percent of wing area equipped with CC blowing. The drawback of modifying an existing large aircraft is that the CC blowing system had to be an add-on-feature with no possibility for rounded trailing edge retraction to maintain its low-drag, high-speed cruise capability. The magnitude of the CC rounded trailing edge is clearly visible in a close-up photo; see Fig. 21 and in-flight Fig. 22. The STOL performance data for the A-6A were close to those predicted from wind tunnel tests, resulting in 1) 140% increase in usable CL;2) 30-35% reduction in take-off and approach speeds; 3) 60-65% reduction in take-off and landing ground roll; and 4) 75% increase in payload.
-1.4
L,-..
-4
o
4
a
IZ
16
20
24
zs
a in degrees Fig. 19 Wind tunnel test data for wing of Grumman A-6A.
J. L. LOTH
18
4.4
'0'
C+= a!*,'
,,#G= Q,2 3.6
'I
,,,'
,' f
)' h
>
2.8
'5 -I
,'
,#'
w v
-#'
,r'
(CI
C
,' ,'
,/'.
,'8'
,/'
2 <, , ,' #'
# ,'
2 ,4'
0
<, #'
,'
1.2
,*'
'8'
,/I
/' ,f
,i' 0'
,''I
,/ IS
, '
,' /'
0.4
,' *'
'I
',
-0.4 -4
4
-r
12 a in degrees
20
2E
Fig. 20 Empirical curve fit to A-6A wind tunnel data shows blowing efficiency factor Ce = 6.3.
V. Conclusions In 1974, the WVU CC Technology Demonstrator STOL was the first aircraft to demonstrate the high-lift capability of CC. Its wings incorporated an in-flightretractable CC rounded trailing edge to enable high-lift generation by CC blowing on an extended wing, and in-flight conversion to a reduced wing area with sharp trailing edge for low drag, high speed cruise. The use of a retractable CC rounded trailing edge required supplying the hot high-pressure CC blowing air through the rounded trailing edge. To minimize air pressure loss by friction, the duct cross-sectional area had to be at least twice that of its choked-flow area A*. To achieve that, an ejector was inserted inside the CC blowing plenum. The area of its choked flow nozzle was five times smaller than the flow area in its rounded trailing edge. Adding such an ejector provided several other advantages: 1) Its entrainment provided boundary layer suction just upstream of the CC blowing slot, which increased the blowing efficiency factor CB.
COMBINING BLC SUCTION AND CC HIGH-LIFT GENERATION
19
Fig. 21 Grumman A-6A CC conversion by an add-on fixed CC blowing duct improving STOL performance at the expense of cruise speed.
2) The entrained flow rate about equaled the nozzle flow rate, thereby doubling the CC blowing mass flow rate. This lowered the Vcc/Vm velocity ratio to increase the blowing efficiency factor CB. It also increased the CC blowing slot height by a factor of four. 3) Ejector entrainment provided wing structure cooling and reduced thermal expansion problems.
Fig. 22 Grumman A-6A during CC flight tests in 1979.
20
J. L. LOTH
4) Ejector entrainment reduced the amount of compressor bleed air required with its associated take-off thrust loss.’o,’’9’8 5 ) During take-off in a jet aircraft, thrust loss associated with compressor bleed, is at least two and one-half times the bleed air momentum 6) In 1979 the second CC Technology Demonstrator STOL aircraft, a converted Grumman A-6, demonstrated excellent STOL performance in terms of increased lift-off weight capability and reduction in required runway length. 7) More research is needed to reduce induced drag associated with flying at high-lift coefficients. Not having to fly on the backside of the power curve would greatly increase safety in STOL flight. Since 1974, numerous other applications for CC blowing over a rounded trailing edge have demonstrated the versatility of this technology. For example: 1) Wake drag reduction behind cars, trucks, torpedoes, etc. 2 ) Propeller downwash drag reduction on tilt rotors. 3) Improved performance of low drag horizontally mounted radiators in cars. 4) Lightweight, hot exhaust gas deflectors on helicopter engines in ground effect. 5 ) Providing an alternative to a helicopter tail rotor, to cancel rotor torque. 6) Noise reduction by wake dissipation on helicopter rotors. 7) Improved effectiveness and control with upper surface blowing (USB). 8) Providing pneumatic control on fixed flight control surfaces. These developments indicate that the future for new CC applications is bright.
References ‘“Wings with Nozzle Shaped Slots,” NACA Translation TM 521, July 1929 (from Berichte Der Aerodynamischen Veruchsenstalt in Wien, Vol. 1, No. 1, 1928). *“The Use of Slots for Increasing the Lift of Airplane Wings,” NACA Translation PW 635, Aug. 1931 (Proceedings L’Aeronautique, June 1931). 3Davidson, I. M., “Aerofoil Boundary Layer Control System,” British Patent No. 913,754, 1960. 4Kind, R. J., and Maull, D. J., “An Experimental Investigation of a Low-Speed Circulation Controlled Airfoil,” The Aeronautical Quarterly, Vol. XIX, May 10, 1968, pp. 170-182. ’Shrewsbury, G., “Numerical Evaluation of Circulation Control Airfoil Performance Using Navier Stokes Methods,” AIAA Paper 86-0286, Jan. 1986. 6Williams, R. M., “Some Research on Rotor Circulation Control,” Proceedings of the Third CALIAVLABS Symposium, Vol. 11, June 1969. ’Williams, R. M., and Howe, H. J., “Two-Dimensional Subsonic Wind Tunnel Tests on a 20% Thick, 5% Cambered Circulation Control Airfoil,” NSRDC TN AL-176, 1070, AD 877764. ‘Loth, J. L., “Some Aspects of STOL Aircraft Aerodynamics,” Business Aircraft Meeting, Wichita, KS, 3-6 April 1973, Paper No. 730328.
COMBINING BLC SUCTION AND CC HIGH-LIFT GENERATION
21
’Loth, J. L., Fanucci, J. D., and Roberts, S. C., “Flight Performance of a Circulation Control STOL Aircraft,” AIAA Paper 74-994, 6th Aircraft Design, Flight Test and Operations Meeting, Los Angeles, CA, Aug. 1974. “Loth, J. L., Fanucci, J. D., and Roberts, S. C., “Flight Performance of a Circulation Control STOL Aircraft,” Journal of Aircraft, Vol. 13, No. 3, 1976, pp. 169-173. “Loth, J. L., and Boasson, M., “Circulation Control STOL Optimization,” Journal of Aircraft, Vol. 21, No. 2, 1984, pp. 128-134. ‘’Loth, J. L., “Retractable Rounded Trailing Edge for Circulation Control Wing,” U.S. Patent No. 4,600,172, issued 15 July, 1986. 13 Englar, R. J., “Investigation into and Application of the High Velocity Circulation Control Wall Jet for High Lift and Drag Generation on STOL Aircraft,” AIAA Paper 74-502, 17-19 June 1974. 14 Englar, R. J., “Circulation Control for High Lift and Drag Generation on STOL Aircraft,” AIAA Journal of Aircraft, Vol. 12, No. 5, 1975, pp. 457-463. ”Englar, R. J., Trobaugh, L. A., and Hemmerly, R. A., “Development of the Circulation Control Wing to Provide STOL Potential for High Performance Aircraft,” AIAA Paper 77578,6-8 June 1977. 16Englar,R. J., Trobaugh, L. A., and Hemmerly, R. A., “STOL Potential of the Circulation Control Wing for High-Performance Aircraft,” Journal of Aircraft, Vol. 15, No. 3, 1978, pp. 175-181. ”Englar, R. J., Hemmerly, R. A., Moore, H., Seredinsky, V., Valckenaere, W. G.,and Jackson, J. A., “Design of the Circulation Control Wing STOL Demonstrator Aircraft,” AIAA Paper 79-1842, Aug. 1979; also published in Journal of Aircraft, Jan. 1981. 18Loth,J. L., “Circulation Control STOL Aircraft Design Aspects,” NASA Circulation Control Workshop, 19-21 Feb. 1986, NASA Ames Research Center, NASA Pub. CP2432, pp. 569-588. ”Loth, J. L., Funk, M. S., “Thrust Savings Limitations with Blown High Lift Wings,” AIAAIAHSIASEE Aircraft Design, Systems and Operations Meeting, St. Louis, MO, 14-16 Sept. 1987.
Chapter 2
Overview of Circulation Control Pneumatic Aerodynamics: Blown Force and Moment Augmentation and Modification as Applied Primarily to Fixed-Wing Aircraft Robert J. Englar* Georgia Institute of Technology, Atlanta, Georgia
Nomenclature Aj = blowing slot area b = wingspan c = chord length cf = flap chord length c d , cD = 2-D or 3-D drag coefficient CDE= equivalent drag coefficient C,, CL = 2-D or 3-D lift coefficient C,,,, = maximum lift coefficient CN = yawing moment coefficient CT = thrust coefficient C y = side force coefficient C , = 2-D or 3-D jet momentum coefficient (see Eq. 1) h, hj = blowing jet slot height L/D,, = equivalent lift/drag ratio m = measured jet mass flux P d , P D = duct total pressure q = freestream dynamic pressure ($pV2) r = circulation control R = universal gas constant Re = freestream Reynolds number, based on chord length c S = wing area *Principal Research Engineer, Aerospace, Transportation, and Advanced Systems Lab., Georgia Tech Research Institute. Associate Fellow AIAA. Copyright 0 2005 by Robert J. Englar. Published by the American Institute of Aeronautics and Astronautics, Inc., with permission.
23
R. J. ENGLAR
24
S, = ground roll Td = duct total temperature V = freestream velocity vj = blowing jet velocity, isentropic x/c = nondimensional chord-wise location xTO = takeoff distance a = angle of attack y = ratio of specific heats Sf= flap deflection angle Sj =jet deflection angle p = freestream density pj = blowing jet density Ic, = yaw angle (side wind angle)
I. Introduction HE USE of pneumatic devices in the form of blown jet airfoils has been employed or been under consideration in the field of aerodynamics as far back as the 1930s, and perhaps even earlier.’,* In most of these devices, which generally fall into the categories of jet flaps or blown flaps, a jet sheet exits from the trailing edge of the airfoil at a fixed angle or tangent to a flap with a sharp trailing edge. This augments aerodynamic forces by entraining and deflecting the airfoil flowfield pneumatically, rather than solely by deflecting a mechanical surface. These are “pneumatic flap” lift augmentors, which have been shown to be successful if a sufficient onboard source of compressed air is available. The aerodynamic concept now known as “circulation control” (Fig. 1) is a logical follow-on to these devices, with one very important difference, which has made a significant performance improvement. The tangential jet sheet exits over the curved trailing edge of the surface replacing the flap, and this curvature can turn through a full 180deg or more. The jet remains attached to that curved surface because of a balance between the subambient pressure in the jet sheet and the centrifugal force in the jet going around the curvature. Initially, at very low blowing values, the jet entrains the boundary layer to prevent aft flow separation, and is thus a very effective boundary layer control (BLC; see Fig. 1 lift plot). Eventually, as the jet continues to turn, a rise in the static pressure plus viscous shear stress and centrifugal force combine to separate the jet sheet, and a new stagnation point and stagnation streamline are formed on the lower surface. The large flow entrainment rate of this jet and the large deflection of the stagnation streamline produce a pneumatic camber, and thus pneumatic control of the airfoil’s circulation and lift. Although it is a very effective BLC device, the interest in this concept comes from its ability to further augment the circulation and lift, and thus giving rise to the name “circulation control (CC).” Several additional benefits became obvious from early experimental investigations of the concept as a means of lift augmentation:
T
1) Only very small flap size or even nonmoving control surfaces were required. 2) Lift augmentation could be achieved independent of airfoil angle of attack.
OVERVIEW OF CC PNEUMATIC AERODYNAMICS
25
TANQENTIAL BLOWING OVER ROUNDED TRAlLlNQ EDGE SURFACE
Fig. 1 Basics of circulation control aerodynamics.
3) Jet turning angle was no longer limited by physical jet exit angle or blown flap deflection angle. 4) Very high force augmentation was generated per unit of input blowing momentum. Roughly 70 years have passed since the very earliest revelation of this type of curved pneumatic device, and a very large variety of pneumatic configurations have been proposed and evaluated. The author has been actively involved with many of these since around 1967. To further expose this wide range of actual and potential applications, this paper will discuss a large number of these pneumatic devices with which the author is familiar from both past and current research, as well as provide an indication of where the use of CC aerodynamics may be heading. It is by no means a complete and exhaustive study of all known efforts, but rather contains representative cases from a wide variety of pneumatic force/moment augmenting and modifying devices. This paper concentrates primarily on fixed-wing aircraft, but CC is certainly not limited to that application alone. The following examples will confirm the multiple uses of CC devices as the following: 1) aerodynamic force and moment augmentors (Fig. 1 shows ACL/CIL= 80, or 8000% return on the invested momentum); 2) aerodynamic force and moment reduction if/when needed (drag in climb out and cruise); 3) aerodynamic moment control and stability augmentation; and 4) aerodynamic device simplifier (moving parts elimination, complexity and weight reduction). 11. Coanda Effect The CC concept is actually based on the now well-known Coanda E f f e ~ t , ~ - ~ named after the Romanian inventor Henri Coanda, who claimed to have discovered it in Paris prior to 1935. There is a Romanian postage stamp (and
26
R. J. ENGLAR
associated story) showing that Coanda had originally used the Coanda device for a totally different purpose: as a means to deflect the exhaust of a radial piston engine away from a wooden aircraft fuselage. During its first flight, these shielding plates actually entrained the hot exhaust flow inward, igniting and destroying the aircraft. Figure 2 shows the basic Coanda device as later formulated by him (after the fire exhaust incident) and its application to a fixed-wing aircraft (which in this case appears to be a form of BLC device). Note that in these (and in all other Coanda cases found), Coanda aligns acuteangle “steps” downstream of one side of a jet nozzle to deflect the jet to that side and entrain large masses of fluid from the opposite side. The distinctive steps and angles were intended to generate a separated vortex flow at each corner, and thus enhance mixing there. The concept was applied by Coanda to many other devices, including car engine exhaust scavengers, wind-tunnel turning vanes, thrust augmentors, water propulsion units, injection wind tunnels, deflection surfaces, and rotary pumps. However, efficiency questions arose because of added friction along all the steps and separated flow at each corner. Nevertheless, the concept forms the basis for the present CC aerodynamics; an infinite number of small-angled steps simply becomes a continuous curved surface with even greater entrainment capability and less energy loss due because its lack of discrete corners. The following discussion will present a number of favorable applications of CC aerodynamics, where the governing difference between CC and the jet flap/blown flap will be the continuously curving surface downstream of the tangentially blown jet, with force augmentation/modification being mainly a factor of jet blowing parameters, not the angles of the sharp flap trailing edge or the jet angle relative to the chord line. The main emphasis here will be on fixed-wing devices and applications. Application of CC to rotary-wing aircraft offers many additional benefits, as discussed in Refs. 6 to 9. These are based on
Original Coanda Device, Approx. 1935
Fig. 2 Coanda devices and high-lift, low-drag wing?
OVERVIEW OF CC PNEUMATIC AERODYNAMICS
27
high-lift generating capability independent of angle of attack, as Fig. 3 shows, thus eliminating the previously required cyclic and collective pitch blade mechanisms. Before proceeding, it is important to define the blowing momentum coefficient C, as
where the last definition only holds for two-dimensional incompressible flow (pj = pm). Typically, the jet velocity in ft/s is calculated from isentropic relationships as:
where the subscript d implies total conditions in the blowing plenum duct, subscript 03 is freestream, R = 1716 ft2/(s2 OR), and y =1.4 for air. A jet
X
U
z
THICKNESSICHORD, tlc
Fig. 3 Maximum lift of blown circulation control airfoils.
R. J. ENGLAR
28
expansion to the actual static pressure just outside the jet slot would yield higher calculated values of vj and thus C,, but would vary as the external flow conditions or shape changed, so would be hard to duplicate as a universal design parameter. Mass flow m is almost always measured under test conditions using appropriate flow meters, but can be calculated isentropically as well using compressible flow relationships." Before going any further, please note that there is nothing that prohibits the jet velocity from being supersonic unless the geometry is such that a shockdown back to subsonic flow causes the jet to detach from the curved surface. It will soon be seen that it is often advantageous to have a higher speed jet than a lesser speed one. The momentum term mvj can, of course, also be thought of as a jet thrust. 111. Applications of Circulation Control, Past and Present
A. Circular Cylinder Stopped-Rotor Aircraft An early application of CC was developed by the British National Gas Turbine Establishment (NGTE) in the mid-l960s, when it was desired to produce a stoppable-rotor VTOL aircraft. In this concept, a blown two-bladed rotor could produce very high lift per blade just to get the aircraft to hover, then be stopped and stowed within the helicopter fuselage for forward fixed-wing flight.697A circular-cylinder cross-section slotted-pipe rotor appeared to be an ideal solution, because, as Fig. 3 shows, its thickness/chord ratio of 1.0 presents the possibility of C, = 4.rr if flow can be made to stay attached. As the figure shows, values even greater than 4.rr were generated by blown CC cylinder rotor blades when excess thrust in the vertical direction (the jet flap effect) was included at higher t / c values. However, the high drag of a 100% thick circularcylinder airfoil proved to be a difficult problem and reduced the aerodynamic efficiency of these airfoils to unacceptable values. A similar circular lifting surface' was also pursued by NASA Langley in the 1960s to provide lift on takeoff and landing by blowing on the circular fuselage cross-section of a hypersonic aircraft, as well as for return after launch of missile or rocket boosters having circular cross-sections. Whereas lift coefficient values over 20 were measured at very low Reynolds numbers for an end-plated-cylinder tunnel model with multiple slots, a single-slotted cylinder produced Cl = 18 at C, = 6. This lift augmentation of only three times the input C, implied the need for a large air supply. The associated drag coefficient of over 9 gave a lift/drag ratio of only 2, or even less if the blowing coefficient were added to the drag to yield an equivalent drag coefficient. Clearly, high lift was available, but the lift-associated drag and required blowing coefficient posed serious problems.
B. Elliptic-Airfoil CC Rotor As interest in circular cylinder CC blades for helicopters was lessening in the United Kingdom, it was rising dramatically in the United States in the late 1960s as a possible means to increase rotorcraft performance while greatly simplifying the entire rotor system mechanical hardware. The US effort was centered at the
OVERVIEW OF CC PNEUMATIC AERODYNAMICS
29
Navy’s David Taylor Naval Ship R&D Center (DTNSRDC), where the approach taken was to develop lower-drag, high-lift rotor blade sections by converting the circular cylinder profile into a much thinner blown elliptic airfoil. These efforts also became the basis for fixed-wing efforts, and are presented here to clarify understanding of these CC pneumatic devices. Figure 4 shows several such single-slotted CC rotor elliptic airfoils, where the obtainable Cl is lower than for the cylindrical airfoil, but the required C, is a factor of 10-20 less. Note that this performance is all at angle of attack a = 0 deg, providing a nonpitching alternative to both the mechanical cyclic and collective angle of attack variation required of conventional rotor blades. Note the very high force augmentation, ACl/C, of 80, representing an 8000% return on the momentum invested. Also shown for comparison is a typical 30-deg jet flap applied to a 15% thick ellipse airfoil; the greatly reduced force augmentation of the jet flap is evident because the jet exits from the lower surface of the airfoil at a fixed angle. It should be clear that CC is not a jet flap, but achieves its high-lift capability because the stagnation stream line movement and resulting circulation can be controlled and increased well beyond that of a sharp trailing edge. Figure 5 shows the equivalent lift-to-drag ratio of sample elliptic CC airfoils, where the
YOMEMUM COEFFICIENT C
Fig. 4 Typical blown-lift capabilities of two-dimensional CC elliptic airfoils at a = 0 deg.
R. J. ENGLAR
30
equivalent drag in the denominator now includes a severe penalty for compressor power required as well as for intake ram pressure.lo9l1Maximum equivalent LID values roughly 6-7% greater than the conventional unblown rotor blade NACA 0012 airfoils (varying only a ) are seen for the 20% CC ellipse (at a = 0 deg), but at a lift coefficient 30% higher, at about 1.3. Furthermore, the Cl can be increased up to 6 or 7 if desired, but at a lesser LID,,. For additional comparison, if the equivalent drag is defined as merely adding C, to the measured drag (i.e., CDE= Cd C,), then LID,, values of over 120 at Cl = 2.5 are possible, all at a = 0 deg (almost three times the Cl of the 0012 airfoil at stall). The efficiency and simplicity of CC was obvious from these tow-dimensional airfoil results, and a serious effort to develop these CC airfoils was undertaken. Reference 12 summarizes much of this Navy effort at DTNSRDC for the years 1969 through 1983, as well as providing a summary of CC-related research conducted by other agencies (in the United States and abroad) outside the Navy from 1956 to 1983.
+
X
3'
a
Y
Pd
SECTIONAL LIFT COEFFICIENT, C/
Fig. 5 Equivalent efficiencies for CC and conventional two-dimensional airfoils.
OVERVIEW OF CC PNEUMATIC AERODYNAMICS
31
In 1979, a CC Rotor flight demonstrator based on a Kaman H-2 helicopter was flown with pneumatic aerodynamic and control systems replacing conventional mechanical cyclic and collective blade pitch. 13- l5 Whereas this flight vehicle was hindered by control system response phasing problems, which limited its flight test envelope, it did demonstrate the ability to substitute pneumatics for mechanical blade lift and control devices for hover and forward flight. It also led to the possibility of higher harmonic control of helicopters, where cyclic lift variations at frequencies higher than one per revolution were possible to eliminate rotor-induced vibrations. The absence of blade collective and cyclic pitch links is possible; they can be replaced by internal control cams or valves to vary blowing pressures.
C. Circulation Control Airfoil Development Considerable CC airfoil development was ongoing at this time, both experimentally and analytically. A number of CFD techniques using various Navier-Stokes codes have been developed and used to understand the relevant viscous flowfields. These will not be discussed here, but can be found summarized in much more detail in Refs. 12 and 16. A typical example of CFD-calculated streamlines and velocity vectors” is shown in Figs. 6 and 7 for a generic flat-sided semi-elliptic CC airfoil. Of particular interest here are the computed velocity vectors and streamlines downstream of the slot on the blown trailing edge (Fig. 7), where the stagnation point of the jet sheet appears to be turned nearly 130-140 deg from the jet exit. A considerable number of additional CFD analyses, both subsonic and transonic, have been conducted by various
investigator^.^^,^^,^^
Fig. 6 Computed streamlines for simplified CC airfoil (a= -2 deg, Cl = 4.6).”
32
R. J. ENGLAR
Fig. 7 Computed velocity vectors and streamlines," CC airfoil.
A number of experimental programs have also been conducted to understand the CC phenomenon and the details within the blown curved surface region. Two-dimensional laser-velocimeter measurements at Lockheed" for the same CC airfoil as in Figs. 6 and 7 showed mean velocities that confirmed the CFD results already presented. Again, jet flow turning to a separation point/stagnation streamline approximately 130-145 deg from the slot was seen (Fig. 8). Experimental investigations by the present author" of a very similar generic airfoil (Fig. 9), used surface static pressure, static pressure across the jet, and a rotatable hot-film shear stress probe to measure the actual separation point location (where
Fig. 8 CC velocity vectors recorded by Lockheed Laser Doppler Velocimeter."
OVERVIEW OF CC PNEUMATIC AERODYNAMICS
33
G 2
w
0 U U
8
t 3
MOMENTUM COEFFICIENT, C, NONDIMENSIONAL CHORDMSE STATION. x/C
Fig. 9 Two-dimensional semi-ellipse CC model geometry, plus measured lift and static pressures as functions of C, and slot height.
shear stress = 0) as a function of blowing and slot height. The resulting C,and C, distributions are seen in Fig. 9. As Fig. 10 shows, jet turning as high as 170175 deg was measured for this airfoil. At a constant C,, greater turning occurred with a smaller slot height because the resultant jet velocity and entrainment are higher as jet area reduces. Figure 9 (left plot) shows that this greater velocity and jet turning clearly results in generation of higher Cl, where values close to nine are possible at a = 0 deg (although tunnel flow impingement occurs here). Figure 9 (right plot) presents associated static pressure distributions on the airfoil. These analytical and experimental data confirm the effectiveness of blowing to greatly deflect the entire flowfield and then strongly increase the circulation and lift on these very generic airfoils, to the point that very high lift is produced without wing flaps and slats and at Odeg angle of attack. Some additional information on generic CC airfoil performance is provided in Ref. 20. One last note on CC airfoil performance: as previously mentioned, smaller slot height yields a larger return in Cl at constant C, than does a larger slot height, primarily because of greater vj/Vm and extra flowfield entrainment. Figures 9 and 10 show this trend. However, if the static pressure coefficient just outside the slot exit (C,,,)is known or can be determined, a new parameter defined in Fig. 11 can be used (when vj is expanded to this local condition to yield CBLc)to collapse the different slot height results (left) into a single curve (right).
R. J. ENGLAR
34
Fig. 10 Blowing jet separation point location measured by hot-film shear stress probe.
c,
%LC
CONVENTIONAL MOMENTUM COEFFICIENTS {EWANDED TO fREESTREAM CONDITIONSI
LOCAL MOMENTUM COEfflClENlS IEXPANDED TO LOCAL WNDITIONS AT SLOT)
Fig. 11 Comparison with momentum coefficients based on local jet exit static pressure (right) and variation with slot height.
OVERVIEW OF CC PNEUMATIC AERODYNAMICS
35
D. X-Wing Aircraft An extraordinary use of unusual CC airfoils is the X-Wing VTOL ~ e h i c l e , ~ al - combined ~~ rotary/fixed-wing aircraft equipped with a fourbladed rotor, which was designed to take off and hover with the same nonmechanical cyclic and collective benefits as those already described. However, forward flight at speeds roughly twice the limit on conventional rotors could be achieved using a “reverse velocity” blown rotor/wing concept (Fig. 12). Typically, as vehicle speed increases, the retreating blade of a rotor sees a resulting velocity that is the difference between the vehicle forward speed and the blade rotational velocity; this can rapidly become a reverse flow at the blade trailing edge, an unacceptable region that moves further outboard as speed increases. Lift on that “stalled” blade segment can actually be negative; the rotor might not be trimable in roll, and drag increases dramatically. The X-wing avoids this problem at high speeds by employing CC on each end of the blade (Fig. 12), and a “clever” control system can blow whichever slot is currently on the airfoil’s trailing edge. Thus, the airfoil never experiences flow from the “wrong” direction. The entire system can be simplified even further by use of simultaneous blowing from both leading and trailing edges of the double-ended airfoils24 (Fig. 12, right). Note that even if the flow is coming from the wrong direction (dashed curve), the dual-slotted airfoil still yields 80-90% of the single-slotted airfoil’s lift, even when the leading edge is counter the conventional direction (compared to little, zero, or negative lift from a conventional airfoil). This allows rotorborne flight at a much higher speed until eventual conversion to a fixed wing in an X-configuration is achieved, with the representative TE slots on each blade of the now fixed wing being used for roll and pitch control without moving surfaces. This concept was actually “flown” full-scale in the NASA Ames 40 x 80 ft tunnel and successfully completed the transition from hover to stopped-wing using pneumatics. Two representative configurations from Refs. 16 and 25 are shown in Fig. 13.
DUAL PLENUM AIRFOIL SECTION
HELO DIRECTION
Fig. 12 Dual blowing on a reverse velocity rotor2’ and blown lift of dual-slotted CC airfoils.
36
R. J. ENGLAR
LIGHTLY LOAOEO LOW TIP SPEED TAIL F A N
Fig. 13 X-wing rotor configuration with rotor stopped, and control systems.
E. Circulation Control Wing (CCW) The high-lift capability independent of angle of attack, which was demonstrated by the CC rotor airfoils above, led to the application of CC as a simplified very-high-lift device for STOL aircraft. The airfoil in Fig. 1 is representative of this simplified pneumatic concept, where both the mechanical TE flap and the LE flap or slat have been replaced with nonmoving pneumatic systems. Primary development of the concept took place in conjunction with CC rotor development efforts at the Navy's DTNSRDC12,16,26-30in the late 1960s to early 1980s. Initially, the concept was modeled as a small add-on device" that would convert the wing flap's sharp TE into the round CC Wing (Fig. 14, right), which was tested at DTNSRDC in specialized two-dimensional high-lift test facilities." Compared to results from a family of more conservative blown flaps (Fig. 15, from Ref. 30), the CCW profiles showed two significant advantages. They could generate greater Cl than the blown flap, because of much greater streamline displacement, and had no sharp TE to limit streamline turning, or for the same chord-length device, could generate the same incremental Cl at much less C, required. An alternative 180 deg rotatable CCW TE is also shown in Fig. 14 (left), which, although it may be mechanically simpler, pays the penalty of losing wing area in the blown high-lift mode. Numerous two- and three-dimensional wind-tunnel evaluations and feasibility studies led up to the flight test of a fixed CCW device on an A-6/CCW STOL demonstrator a i r ~ r a f t ~in l - 1979. ~ ~ Flow visualizations in Fig. 16 show a full 180 deg of jet turning on a static 1/8-scale model of the test aircraft in the DTNSRDC tunnel, and Fig. 17 shows the CCW installation on the fixed flap of the A-6 flight-test aircraft. Because this was a proof-of-concept flight test, the CCW device was not retractable and the air supply lines were mounted externally and cross-ducted in the fuselage, where they connected to the high-pressure
OVERVIEW OF CC PNEUMATIC AERODYNAMICS
37
CONVENTIONAL
Fig. 14 Retractable/storable CCW trailing edges.
bleed ports of the standard J-52-P8A turbojet engine. Results using only available bleed air from the engines confirmed maximum C, values 120% greater than the conventional Fowler flap, or, even more applicable, 140% increase in the usable lift coefficient at takeoff/approach angles of attack. Also confirmed were 3035% reductions in the takeoff and approach speeds resulting in 60-65% reductions in takeoff and landing ground roll distances, and yielding values as short as 600-700 ft. This full-scale confirmation of CCW also implied that
FLAP CHORD LENGTH
MINIMUM Cp REQUIRED
Fig. 15 Comparisons between CCW and blown flap airfoils at a = 0 deg.
38
R. J. ENGLAR
Fig. 16 CCW jet turning on the Ad/CCW wind tunnel model at DTNSRDC.
there was sufficient extra CLgenerated to increase the liftable payload by 75% if the conventional takeoff ground roll distance were used. Also shown was that the additional lift-induced drag resulted in much steeper glide slopes on approach, where higher engine power settings (which could also be used for quicker response during waveoff) were offset by this excess drag.
Fig. 17 A-6/CCW STOL flight demonstrator aircraft.
OVERVIEW OF CC PNEUMATIC AERODYNAMICS
39
Fig. 18 WVU STOL demonstrator CC airfoil?'
A smaller CCW demonstrator based on a prop-driven BD-4 general aviation aircraft had been flown earlier by West Virginia University (WVU).'6,35,36In the flight-tested configuration, the CC blown cylinder was mounted at the TE of a hinged flap that rotated 180 deg aft to increase the effective high-lift area by 20%, and included a boundary layer control (BLC) suction slot at the flap hinge upper surface (Figs. 18 and 19). The blowing air was supplied by an onboard 200 hp compressor (APU), which provided enough air for blown ailerons in addition to the CCW. The section lift coefficient on the blown CCW wing section was increased by a factor of nearly 2.5 with blowing. Wing downwash on increase to a factor of 1.92, but provided threethe tail reduced trimmed C,, dimensional lift augmentations of ACL/C, = 15.2, a significant increase should the required airflow be available from a general-aviation aircraft engine, for example, if using a supercharger or turbocharger.
Fig. 19 WVU BD-4 based STOL demonstrator air~raft.3~
40
R. J. ENGLAR
CORRECTED
mnm. F&, MS x 1
6
Fig. 20 Thrust performance of J-52-PSA turbojet engine with bleed.
Both of these fixed-wing flight programs demonstrated the feasibility of CCW as an operational STOL system in terms of high lift, short takeoff and landing, and simplicity, but also identified issues still to be resolved. Among these were the drag of the device in cruise flight (WVU solved this but at the cost of a mechanical 180-deg rotating flap that stowed in the aft wing cavity, and GTRI solved it with the dual-radius airfoil discussed in Sec. 111.F.) and, of course, the need for an onboard air source. Figure 20, which presents turbojet engine ground test data31 taken during the A-6/CCW program, show that the airflow acquired from highpressure compressor bleed ports could be increased up to three to four times that of the standard engine spec bleed limit without overheating, but obviously at the cost of takeoff thrust lost. Similar data for turbofan engines show that engine core bleed is much more costly in thrust loss (although lower-pressure fan bleed is possible), and thus the idea of an ejector to trade excess pressure for extra mass flow appears feasible. However, the need to reduce CCW drag in cruise is a necessity for operational aircraft.
F. Advanced CCW Airfoils DTNSRDC and Grumman took two approaches to the drag p r ~ b l e m , ~ ~ ’ ~ ~ - ~ one a fixed simple radius reduction and the second a very-small-chord deflectable CCW flap. From the nondeflectable ~ t a n d p o i n ta, ~supercritical-type ~ airfoil was employed as the baseline because it already had a bluff base thickness between 0.005~and 0.010~.CCW rounded and semiround (96 deg arc instead of 180 deg)
OVERVIEW OF CC PNEUMATIC AERODYNAMICS
41
designs were tested, including a series of smaller radii, looking for reduced drag without loss of lift augmentation. The CCW/Supercritical airfoil was developed primarily from a low-speed standpoint, where a TE radius of 0.009~was found to produce very little drag penalty yet have superb lifting capability. Figure 21 shows its lift curves at constant C, compared with a family of mechanical multi-element flaps. Not only do the no-moving-parts CCW airfoils generate the same or greater lift as the maximum Cl of a triple-slotted-flap airfoil with mechanical slat, they also do so at a = 0 deg. Note that the large leading edge of the supercritical airfoil provides a natural nonmoving LE device, which generated similar stall angles to the mechanical slat. One further benefit is the cruise drag polar (Fig. 22), which is slightly higher unblown than the baseline supercritical airfoil, but with very slight blowing can reduce c d to less than the baseline while also increasing lift, both at constant a. For clarity here, measured c d includes C, because the wind-tunnel balance cannot easily separate blowing thrust from drag. That is why Fig. 22 shows negative drag recorded with blowing. This is accounted for in the equivalent drag term, LID,,, where C, is included (see Ref. 10 for a more detailed explanation). Additional benefits of blown CC airfoils at speeds up to transonic were shown in the compressible flow tests of Ref. 40, where blowing was seen to produce a very favorable boundary layer/shock interaction, drag reduction, and increased Cl (Fig. 23). The second approach to the drag problem was a simple CCW flap with a curved upper surface and a sharp trailing edge (Fig. 24, from Refs. 38 and 41). Here, a short-chord flap (less than 0 . 1 0 ~ pivots ) about a hinge on the lower surface and exposes a smaller-radius CCW surface downstream of the tangential slot. This radius is approximately the airfoil thickness at the slot location, less the slot height. The upper surface of this flap is a second arc of much larger radius, the radius being chosen to keep the arc close to the airfoil aft contour. As the small flap is deflected on this dual-radius CCW airfoil, the large radius produces an arced CC aft surface with a turning arc much larger than the flap deflection angle.
MULREWEHT MECHANICAL ~ W F T AIRFOILS
NDWVHG-PART CCWlsUPERCRlTlcAL AIRFOIL
Fig. 21 CCW/Supercritical, dual-radius CCW, and conventional mechanical flap airfoil comparisons.
42
R. J. ENGLAR
Fig. 22 Low-speed drag polars for CCW/Supercritical airfoil.
Freedream Mach Amber, Iy,
Fig. 23 Transonic lift caused by blowing for three pneumatic ellipse airfoils!'
OVERVIEW OF CC PNEUMATIC AERODYNAMICS
43
Supercritical Contour
Fig. 24 Dual radius CCW airfoil with LE blowing.
For the 90-deg flap shown here, the jet turning angle is about 135 deg (compare to Figs. 8 and lo), limited by the TE comer. With flap retracted to 0 deg, the airfoil is in a sharp-TE cruise configuration. The slight mechanical addition provides unblown camber as well. The leading edge employs an inverted tangential slot to replace any mechanical flap there. Lift data for the 90 deg flap configuration are also shown in Fig. 21, where Cl increases of 35% over the CCW/Supercritical airfoil occur, with considerably greater increases over the conventional flaps. Figures 25 and 26 also show additional advantages of this configuration: the ability to dramatically interchange lift and drag as the small-chord CCW flap
2-DCCW SUPERCRITICAL AIRFOIL, DUAL- RADIUSFLAPS, DRAG POLARS, THE PENALTY FOR LIFT??
Fig. 25 Drag polars of CCW dual-radius airfoil at various flap angles.
R. J. ENGLAR
44
2-D CCW/SUPERCRITICAL AIRFOIL, DUAL-RADIUS CCW,
u, degrees
Fig. 26 Dual-radius 90-deg flap CCW airfoil lift as functions of
(Y
and C I * ~ ~ .
is deployed (Fig. 25) and the increased lift and stall a as LE blowing is activated (Fig. 26). The thrust/drag interchange in Fig. 25 implies the potential for high lift and drag for STOL approach (remember, induced drag due to high lift is not included in this data for two-dimensional airfoils) or high lift and reduced drag for takeoff. Figure 26 shows the capability of this nonmoving LE device to reattach flow, prevent stall, and dramatically increase C , . Figure 27 combines the above data in terms of Cl vs LID,, where the equivalent drag coefficient is defined as CDE= Cd C, to account for the blowing required to yield these drag changes. These data include four CCW flap angles and various LE and TE blowing values. Figure 27 includes a locus of achievable Cl vs the associated efficiencies in comparison to the clean cruise airfoil (flap = 0 deg, C,E = 0, , = 0). This plot confirms the ability of CCW airfoils to generate very C high lift and associated drag (reduced LID,) for approach, plus much higher L / D e at somewhat lower Cl for takeoff and climbout. Because of the latter, the 30-deg CCW flap at reduced C, appears to be an excellent configuration for takeoff. One additional benefit results for the 0-deg flap CCW case. When in cruise, drag is low because of the sharp TE, but should blowing be initiated without flap deflection (Fig. 28), significant lift is generated by the flap curvature, while drag reduction occurs due to thrust recovery. Note the comparison to the NASA Energy Efficient Transport slotted, flapped airfoil. Not only is lift
+
Next Page OVERVIEW OF CC PNEUMATIC AERODYNAMICS
45
CCW Dual-Radius Airfoil DRAG POLARS: the Penalty for Lift?
2-D Lift Coefficient, C,
Fig. 27 Lift and equivalent efficiencies of dual-radius CCW airfoil.
greater for the CCW cruise airfoil, the drag polars move into the thrust recovery region. From these results, one can also immediately realize the potential of this high-lift system as a nonmoving roll/yaw device. Blowing only the undeflected right wing’s flap will produce a lift (roll with right wing up) and favorable yaw (nose left), thus yielding favorable roll/yaw coupling from a nonmoving surface, instead of the usual adverse roll/yaw coupling from a conventional aileron. A study41 was conducted for NASA Langley Research Center to evaluate the effectiveness of applying this concept to an Advanced Subsonic Transport. Here, the dual-radius CCW of Fig. 24 was applied to a 737 wing characterized in Fig. 29. The typical 15 moving elements per wing were replaced with the CCW single element flaps and LE blowing, yielding perhaps a maximum of three components per wing (the outboard CCW flap became the aileron, and blowing differentially on the CCW flap replaced the spoilers for roll). Using only fan bleed air (and the associated lower thrust lost), replacing the conventional flaps with CCW was able to triple the usable lift at takeoff and produce the ground roll reductions shown in Fig. 30. For lighter aircraft weights, blown takeoff rolls of 400-500 ft are possible with 0 kn headwind, about a third that of the conventional aircraft; with a 20 kn headwind (wind over deck), 200-300 ft
Previous Page
46
R. J. ENGLAR
Fig. 28 Comparison of cruise dual-radius CCW (Odeg flap) with mechanical flap airfoil.
737 WlNOlFLAPICONTROL SYSTEM
Fig. 29 Pneumatic airfoils simplify wing complexity.
OVERVIEW OF CC PNEUMATIC AERODYNAMICS
47
WODIO.0 kt.
Qrou WoIgM; lb
Fig. 30 737/CCW takeoff ground rolls, sea level, at 0 kn headwind.
rolls were predicted for the 737/CCW configuration. The tradeoff in increased liftable weight at constant ground roll is possible here as well. For the conventional 737’s ground roll of 1350 ft at W = 70,000 lb, the 737/CCW can increase the liftable weight to near 100,000 lb with the same runway length, a 43% increase in gross weight (fuel, payload, etc.). Climb angle over the 50-ft obstacle is slightly less for the CCW aircraft41than for the conventional one. These pneumatic benefits are not limited to subsonic transports. Reference 42 reports on the application of CCW to a generic High-speed Civil Transport (HSCT) configuration developed by GTRI for NASA Langley Research Center. Here the intention was to reduce takeoff wing area required (and thus increase cruise performance) by employing simplified CCW flaps and trading increased lift coefficient for reduced wing area. The extra lift would also allow a reduction in takeoff/approach angle, and eliminate the need for fuselage nose droop, aft fuselage upsweep, extended nose gear, and/or synthetic vision. Reference 42 reports 90-deg jet turning on the CCW flap, over 100% increase ,,,,C , and 45% increase in stall angle. High-lift generation from various in blown devices and blown canards is shown in Fig. 31. Note the effect of the blown canard on stall a. In-house feasibility studies43 have been conducted and confirm this potential for pneumatic wings/canards on high-speed aircraft.
R. J. ENGLAR
48
Angle of Attack,a,deg
Fig. 31 Lift augmentation on the GTRI HSCT/CCW semispan model with blown
IV. Powered Lift and Engine Thrust Deflection A. CCW/USB Mechanical flaps have been employed to entrain and deflect thrust from engines mounted on the wing upper surface (upper surface blowing, USB), and it was envisioned that the entrainment capabilities of CCW could do the same without the mechanical complexity; thus the CCW/USB c o n ~ e p t ~ ' , ~was ~-~* born (Fig. 32). Subsonic wind-tunnel investigation^^^ at DTNSRDC showed no-moving-part pneumatic capability to entrain and turn USB engine thrust well past the 60 or so degrees of a mechanical USB system, but also continuing through 90 deg, and then rotating the thrust forward as a thrust reverser through 165 deg (Fig. 33). The possibility then exists for high lift and thrust reversing all in one system just by varying the CCW blowing rate, with a possibility of VTOL in between (depending on installed thrust levels). Wind-on data (Fig. 34) show very interesting lift-drag polars at a = 0 deg, with the ability to vary lift and drag by blowing alone, independent of angle of attack. The enhanced lift capability is far more than mere thrust deflection (i.e., ACL = CT [sin a,, a]).It
+
WRIPBLEMUSTOEFW3lD+l UJETOCC.PLENUHPFESWW
U A R m
Fig. 32 USB and CCW/IJSB powered-lift concepts.
OVERVIEW OF CC PNEUMATIC AERODYNAMICS T = ENQINETHRUST (LB) 01 * ANQLE OF ATYXK q = DYNAMIC PRESSURE
49
L! 0 (CCW ALONE)
P, = SLOT PRESSURE mV, = SWT MOMENTUM FT T + IhVi T. = RESULTAKTTHRUST
0 25.4 1,LB 0 48 6T L5
Fig. 33 CCW/USB model static thrust deflection by blowing only.
results from the increased velocity from the engine exhaust being entrained onto the blown lift surfaces, and the greatly increased circulation lift beyond the powered wing only. A full-scale ground test was performed by the present author at NASA Ames with the CCW/USB mounted behind one engine of the NASA Quiet Short-haul
THRUST
DRAG
Fig. 34 CCW/USB model lift-drag polars, (Y = 0 deg.
50
R. J. ENGLAR
Fig. 35 CCW/USB test assembly on the QSRA aircraft.
Research Aircraft (QSRA), with the aircraft mounted on a force b a l a n ~ e . ~ ~ , ~ ’ - ~ l Figure 35 shows the installation behind the left inboard engine of the QSRA. Thrust deflections as high as over 100 deg were recorded behind this single operating engine. These data are expected to improve if the two engines per wing are operated together and the two exhaust sheets converge for even better turning. As a result of this test, Navy feasibility studies46were conducted for a sea-based turbofan-powered STOL aircraft using both CCW and CCW/USB (Fig. 36). These studies, based on the preceding powered model wind-tunnel tests, showed takeoff ground rolls of 100-200 ft (Fig. 37), varying with weight, blowing, and thrust levels, and resulting from powered C, values of 8-9.
Fig. 36 Proposed CCW/USB Navy STOL aircraft.
OVERVIEW OF CC PNEUMATIC AERODYNAMICS
51
GROSS WEIQHT. LBllOm
Fig. 37 Takeoff ground rolls for proposed CCW/USB Navy STOL aircraft.
B. CC/Jet Deflection In a related effort,52CC entrainment was also applied to high-performance aircraft to yield thrust deflection for much higher engine exhaust velocities, where lesser jet turning could still provide excellent STOL potential due to higher thrust/weight ratio. An example is shown in Fig. 38. In-house unpublished experimental work by the present author provided similar studies, where we were able to deflect supersonic jets from rectangular nozzles by more than 80 deg, using blowing jet momentum values around 10% of the engine thrust.
BLOWING WY)ENTW,
&I,
US.
Fig. 38 Pneumatic thrust deflection of rectangular jet exhaust (left, from Ref. 52) and unpublished static test results of a similar configuration (right).
52
R. J. ENGLAR
C. Pneumatic Channel Wing A configuration using similar thrust deflection capability of a CC trailing edge has recently been under development by GTRI for NASA Langley Research Center. Called the Pneumatic Channel Wing (Fig. 39), it employs blowing at the TE of a 180 deg channel (similar to the much earlier but unblown Custer Channel Wing) to entrain the propeller’s thrust, augment the velocity in the channel, and thus generate high-powered lift. Figure 40 (from Refs. 53 and 54) shows typical GTRI wind-tunnel lift data as a function of both blowing and thrust compared to the baseline unblown channel wing configuration, where untrimmed CL,, is increased by a factor of over 7 to a value of 10.5- 11. Reference 54 shows predicted takeoff ground rolls of less than 100 ft on a hot day at 3000 ft altitude using wing angle of attack of only 10 deg. Further details of this concept’s capabilities and the associated data are found in the paper by Englar and Campbell in this volume, and also in NASA CP 2005-213509.
Fig. 39 Conceptual pneumatic channel wing and semispan model in GTRI MTF tunnel.
OVERVIEW OF CC PNEUMATIC AERODYNAMICS
53
a,degrees Fig. 40 Pneumatic channel wing lift from thrust and blowing?4
V. Other Aircraft Applications
A. CC Propeller In a manner somewhat similar to the CC rotor above, CC airfoils have also been incorporated into general aviation propeller designs to replace complex and expensive mechanical variable-pitch blades with fixed-pitch pneumatic blades that change aerodynamic and thrust characteristics through mass flow variation to each blade. Figure 41 shows a proposed application, where the propeller blade airfoil is the CCW/Supercritical type of Fig. 21. References 55, 56, and 57 discuss feasibility studies, and concluded that such a pneumatic variable-pitch propeller was possible and held interesting promise depending on the details and costs of an air compressor (such as an aircraft supercharger or turbocharger) to supply the blowing. The study also envisioned supersonic jet blowing to be a possible problem, but much of the CCW data already presented above have blowing pressures well above choked (sonic).
54
R. J. ENGLAR
Flow Control Valve
Compressor
Fig. 41 Circulation control propeller system.
B. Moment Control, Stability Augmentation, and Induced Drag Reduction The preceding data and applications show the ability to pneumatically augment or modify lift and drag without use of moving parts (except possibly very short chord dual-radius CCW devices) and with a high rate of return on input jet momentum. The application to a pneumatic rudder or even winglets can provide side force generation as well.58 It should be obvious that augmenting the aerodynamic force capability of any control surface by blowing can also either increase the control power or reduce the required area of the device, with associated benefits including maintaining stability levels but reducing cruise drag. A few further and less obvious examples of pneumatic control devices will now be mentioned; many others can be found in Refs. 12, 16, and 58. The aft suction peak downstream of the CCW slot (usually at 95% chord or greater) produces very large nosedown pitching moment, which, besides having to be trimmed, also produces greatly enhanced longitudinal pitch stability. In fact, the A-6/CCW flight demonstrator had such large negative values of dCM/dCl that the center of gravity of the flight-test aircraft was moved aft by an additional 10- 15% chord to aid in trim, and the aircraft still had greater longitudinal stability than the conventional A-6, flaps A clever application of CCW for moment control is shown in Fig. 42 on a forward swept wing.59 Previously, increasing blowing on an aft-swept trailing edge pulled the center of pressure (CP) outboard and aft, but during this tunnel e~aluation,~’ the CP was made to move outboard and thus forward with blowing. The amount of xcp movement and the resulting moment were controlled by which segments of the TE slots were blown, and by how much. Pneumatic roll control by differential wing blowing can produce phenomenal rolling moment increments (Fig. 43), where only one wing of a CCW configuration is blown. A second innovation is also shown here: letting the slot continue around the wing tip. Now, high suction peaks at the wing tip, having a maximum moment arm, can yield even greater rolling moment. For reference, a conventional 0.20-chord aileron deflected down 30 deg on the outboard 50% span of
OVERVIEW OF CC PNEUMATIC AERODYNAMICS
55
WEIGHT FLOW-0.25 Iblsec
Fig. 42 CCW applied to forward swept wing for pitching moment reduction.
this wing produced an incremental hCrol1= -0.03. Figure 44 suggests one further advantage of the CC wing tip, where blowing down around the tip directly counteracts the tip vortex rollup and relocates it further outboard, creating an effective aspect ratio increase. Figure 44 shows the effective drag reductions due to tip blowing*l on an already high-aspect-ratio CC rotor blade. At the higher C, values where induced drag usually dominates, C , reductions of 1719% are seen, with greater percentage reductions at lower CL. Lower aspect ratio aircraft wings using this technique should yield even greater CD reduction. One can also alter the spanwise lift distribution with spanwise tapered blowing to approximate an elliptic distribution, and thus minimize induced drag both in cruise and during climbout. In 1986, the present author experimentally applied blowing from a tangential slot along the nose of a generic high-a vortex-lift configuration, thus turning the
Coefficients based on full span and area
Fig. 43 Roll due to CC wingtip blowing, (Y = 0 deg.
56
R. J. ENGLAR
WNG LIFT COEFFICIENT
Fig. 44 Tip blowing for induced-drag reduction.
fuselage into a side-force and yawing-moment generator. Other investigators have more recently tried similar schemes, but the results shown in Fig. 45 summarize these effects. At a = 35 deg, the conventional rudder was useless because of fuselage blockage and separated flow (see C , = 0 curve), but blowing on the right side of the nose restored directional stability when the vehicle was yawed to the left, and vice versa. Large side force values were also generated by blowing.
C. Microflyer and Pulsed Blowing A combination of all of the above force and moment control applications has been pursued recently at GTRI relative to a very small unmanned aerial vehicle (UAV), the Pneumatic Microflyer.60 Jet turning on a small-scale, lowReynolds-number wing is illustrated in Fig. 46. Pneumatic lift and control surfaces will be driven by gas generated by a GTRI proprietary engine powering the flapping wings of a 6-in.-span flying-insect-like UAV. The opportunity also exists here to take advantage of pulsed blowing, investigated in Refs. 61 and 6 2 for application to blown flaps and in Ref. 6 3 relative to CCW. Here, for properly shaped blowing wave forms, mass flow required was greatly reduced
OVERVIEW OF CC PNEUMATIC AERODYNAMICS
t c
57
0.40
I 0.30
I!
a 0.20
CY SIDE FORCE
a. 10
COEFF.
0.00
-0.10 -O.’O
1 t
I!
O-I5 a 0.10 y1
1
I
0.201
I c CN YAWING MOMENT COEFF.
s I
cn 0.05
2
0.00
-0.0s -0.10
-loo
-5O
SIDESLIP ANGLE,
0
11
+50
+loc
NOSE RICHT-
Fig. 45 Tangential forebody blowing for yaw and side force control, a = 35 deg.
experimentally by up to 40 to 50% (Fig. 47), or conversely, greater lift could be generated by the same mean mass flow levels. Also, as a simplifying means, all pneumatic Microflyer control moments would be generated by differential blowing, rather than by very small moving mechanical parts.
VI. Nonflying Applications of Circulation Control A number of nonflying applications have been investigated, where the CC phenomenon was used to augment or modify flowfields for unique purposes. In order to provide pitch and/or yaw control for submarines without using mechanical stem planes, a dual-slotted “pneumatic” stem plane64 was designed for submerged applications (Fig. 48). Here, up or down pitch of the submarine (or right or left yawing moment) could all be provided by blowing the appropriate slot. Towing basin tests of this concept verified that blowing water from the slots when underwater was equally as effective as pneumatic devices (even if the power required might be higher), and provided the opportunity for smaller stem-, bow-, or sail-planes, or avoided the possible control plane jam problem of moving hydrodynamic surfaces.
58
R. J. ENGLAR
Fig. 46 Low-Reynolds-numbermicroflyer wing with CCW turning.
Applications of pneumatics similar to the CC rotor were both conceived as the CC fan (Fig. 49, from Ref. 65) and the CC windmill. Here, variation in blowing parameters through the individual blade slots could vary the output of the fan, or conversely, for a pneumatic windmill, vary the sensitivity of each individual blade to the incoming wind angle and strength, as well as the radial load distribution on the blades. For the windmill, blade pitch would not be required to change mechanically for maximum performance or avoidance of rotor overspeed. More recently, application of pneumatic concepts to improve the aerodynamic performance of automotive vehicles has been heavily pursued at GTRI. Tests on European Formula 1cars (Fig. 50) have verified that proper application of blowing can dramatically increase the download frequently required for higher-speed cornering of these cars, or reduce the required wing area and its associated drag (note the absence of the conventional inverted fore and aft multi-element wings). The high suction (negative static pressure) difference across a blown lifting wing inverted on a race car can also entrain sufficient flow to provide cooling through a radiator located therein. Figure 51 shows a Formula SAE car with an aft blown wing including a pneumatic radiator (unit developed and tested at GTR166 with assistance from the GT Motorsports team). It is now possible to have a multipoint aerodynamic race car design that had previously been prohibited by the “nonmoving-aerodynamic-components”rule. More details of these concepts and of testing of this device are found in another paper by Gaeta and Englar in this volume and in NASA CP 2005-213509,2005. A GTRI program originally intended only to reduce aerodynamic drag on production cars for increased fuel economy has recently led to additional benefits. Blowing on the curved aft panels67 of a generic streamlined car (Fig. 52) showed drag reductions of up to 35%, but also drag increases of over 100% by blowing different elements, which could be used as a form of aerobraking. Lift could also be increased by up to 170% over the unblown car, or conversely, a lower surface slot could also yield downforce if so desired. GTRI tests also showed that yawing and pitching moments could be dramatically changed by
OVERVIEW OF CC PNEUMATIC AERODYNAMICS
8
m a a
I
59
60
R. J. ENGLAR Optional End Plate
Fig. 48 Blown model stern plane design, two-slotted.
FAN INFLOW
Fig. 49 Circulation control fan concept.
OVERVIEW OF CC PNEUMATIC AERODYNAMICS
61
Fig. 50 Pneumatic Formula 1 car model in GTRI tunnel.
blowing, and lateral and directional stability could be restored by blowing only one side of the slot. Interestingly, the blowing required for all of this could be provided by turbochargers or superchargers now being installed on highperformance cars. These experimental data for automobiles have now been extended to and adapted in a GTRI program for the Department of Energy68969to improve the aerodynamics, performance and economics of heavy vehicles (HV; i.e., large tractor/trailer trucks). Figure 53 shows blowing on all four aft comers of the trailer; this combination is able to reduce drag, turbulent separated flow,
Fig. 51 GTRI-patented pneumatic aerodynamic heat exchanger installed on formula SAE race car by GT Motorsports Team.
62
R. J. ENGLAR
Fig. 52 GTRI Pneumatic Futurecar model for drag reduction, showing jet turning.
spray, and aft suction on the back doors. Blowing the lower slot only can increase download and aid in braking or provide traction in wet/icy weather, while blowing the top slot only can generate lift and thus reduce effective weight on the tires and rolling resistance. Blowing either side slot can offset yaw due to gusts or sidewinds (which can yield a large component of increased highway drag), or can help to restore lateral/directional stability. Because the response of the blowing system can be virtually instantaneous (pressure of only 1314 psig can produce sonic jet velocity), safety of operation is very promising, including the ability to prevent jackknifing by generating opposite yawing moment for the trailer. Blowing on the trailer top leading edge also appears promising, because it can provide not only a boundary layer control device, but also can entrain flow up through the cab/trailer gap and eliminate strong separation and vorticity there, plus enhance cooling. Wind-tunnel investigations of this concept on a smaller-scale model of a blown pneumatic heavy vehicle (PHV)68,69 have shown drag reductions of up to 80% relative to a baseline generic HV
Fig. 53 Pneumatic heavy vehicle configuration with potential for 5 blowing slots.
OVERVIEW OF CC PNEUMATIC AERODYNAMICS
63
Fig. 54 Pneumatic heavy vehicle test rig undergoing road tests.
model, plus the ability to increase drag if needed for braking, as well as provide side forces and lateral/directional control in side winds. They have also confirmed that blowing only one vertical side slot at the rear of the trailer can eliminate the destabilizing yawing moments due to sidewinds and generate counteryaw in the opposite direction if needed. These tunnel tests have led to development of a full-scale test vehicle and on-road test program of a PHV test rig (Fig. 54), now ongoing for DOE. More data on this pneumatic groundvehicle program are found in another paper by Englar in this volume as well as in NASA CP 2005-213509,2005.
VII. Conclusions Capabilities The high flow-entrainment capability of tangential blowing over curved aerodynamic surfaces has been shown in the preceding discussions to yield augmentation and control of virtually all aerodynamic/hydrodynamic forces and moments by simplified means, which frequently require no moving external components. The capabilities of the CC devices demonstrated include the following: A.
1) Two-dimensional lift coefficients as high as 20 without moving parts and similar high Cl for download as desired in automotive applications. This extra high lift can also provide aircraft Super-STOL capability or the downsizing of wing area for more efficient cruise. 2) Lift augmentations ACl/C, of 80 and very effective boundary layer control.
64
R. J. ENGLAR
3) Drag reduction due to flow reattachment and thrust recovery, or drag increase due to flow turning and lift-induced drag, and the ability to pneumatically activate these as needed by the pilot or driver-this is particularly applicable in automotive usage. 4) Aerodynamic moment increases from blowing or differential blowing to provide large control increases compared with those of mechanical devices, or to allow control surface downsizing. 5) Pneumatic engine thrust deflection to 165 deg or more without moving surfaces. 6) Pneumatic propellers or rotor blades to achieve variable thrust and control moment without mechanical cyclic pitch. 7) Automotive applications to vary all forces and moments, including racing vehicle download and drag, without moving parts, using only onboard air sources such as turbochargers. Also, a low-drag aerodynamic heat exchanger using pneumatic-generated pressure difference can cool the vehicle while controlling aerodynamic forces and moments. B. Future of Circulation Control The preceding capabilities offer the potential for aerodynamiclhydrodynamic vehicles simplified by pneumatic multipurpose sui$aces synergistically augmenting lif, drag, moments, control, stability, and propulsive functions without any moving mechanical parts. The force augmentation capability also offers the potential for reduction in wing and control surface areas for improved cruise performance, or multipoint designs with lift/control surfaces sized for optimal points of operation. Future investigations could include improved pulsed blowing to even further reduce the required input mass flows, or to simplify the operation of complex devices such as higher harmonic rotors. Application of CC pneumatics to automotive and hydrodynamic vehicles offers the use of aerodynamic surfaces for functions not currently employed, such as aerodynamic drag reduction or increase, download, heat exchange, thrust augmentation, and stability and control. The opportunity to incorporate all of these devices into a synergistic blown vehicle from the initiation of the design, rather than as an add-on, offers the potential for a very effective and efficient multipurpose vehicle, in which the pneumatic effectiveness, including the propulsion system air supply source and the control systems, is incorporated from the very beginning. A perfect example of how CC could be applied to a new and unique Super-STOL vehicle would be its application to the new NASA ExtremeSTOL concept aircraft, where desired goals include C, of 10, balanced field lengths of 2000 ft or less and, of course, the necessity to trim and control this vehicle at very low speeds, plus the ability to interchange drag increase and drag elimination between approach and takeoff operations, respectively.
References ‘“The Use of Slots for Increasing the Lift of Airplane Wings,” NACA Translation, PW 635, Aug. 1931 (Proceedings L’Aeronautique, June 1931). ’“Wings with Nozzle Shaped Slots,’’NACA Translation, TM 521, July 1929; Berichte Der Aerodynamischen Vereuchsenstalt in Wien, Vol. 1, No. 1, 1928.
OVERVIEW OF CC PNEUMATIC AERODYNAMICS
65
3Metral, A. R., “On the Phenomenon of Fluid Veins and their Application, the Coanda Effect,” AF Translation, F-TS-786-RE, 1939. 4Sproule, R. S., and Robinson, S. T., “Combined Intelligence Objective SubCommittee Report,” WF Document Library Item 5, File No. IX-1, X-2, XII-1, D52.420127, 1944. ’Voedisch, A., Jr., “Analytical Investigation of the Coanda Effect (Project No. FP188),” Air Material Command, Wright Field, Dayton, OH, Rept. F TR-2155-ND, April 1947. 6Cheeseman, I. C., and Reed, A. R., “The Application of Circulation Control by Blowing to Helicopter Rotors,” Journal of Royal Aeronautical Society, Vol. 71, No. 848, 1966. ’Cheeseman, I. C., “Circulation Control and Its Application to Stopped Rotor Aircraft,” AIAA Paper 67-747, Oct. 1967. 8Lockwood, V. E., “Lift Generation on a Circular Cylinder by Tangential Blowing from Surface Slots,” NASA Langley Research Center, Technical Note D-244, May 1960. 9 Englar, R. J., “Circulation Control Pneumatic Aerodynamics: Blown Force and Moment Augmentation and Modification; Past, Present and Future,” AIAA Paper 20002541, presented at AIAA Fluids 2000 Meeting, June 2000. “Englar, R. J., and Williams, R. M., “Test Techniques for High Lift Two-Dimensional Airfoils with Boundary Layer and Circulation Control for Application to Rotary Wing Aircraft,” Canadian Aeronautics and Space Journal, Vol. 19, No. 3, 1973 pp. 93-108. 11 Englar, R. J., “Two-Dimensional Subsonic Wind Tunnel Tests on a Cambered 30Percent-Thick Circulation Control Airfoil,” NSRDC, Technical Note AL-201, AD 91341 lL, May 1972. 12 Englar, R. J., and Applegate, C. A., “Circulation Control-A Bibliography of DTNSRDC Research and Selected Outside References (Jan. 1969 through Dec. 1983),” DTNSRDC-84/052, Sept. 1984. 13Wilkerson,J. B., Barnes, D. R., and Bill, R. A., “The Circulation Control Rotor Flight Demonstrator Test Program,” American Helicopter Society, Paper AHS 79-5 1, May 1979. 14Mayfield, J., “Aeronautical Engineering-Navy Sponsors Coanda Rotor Program,” Aviation Week and Space Technology, 31 March 1980, pp. 69-74. ”Wilkerson, J. B., Reader, K. R., and Linck, D. W., “The Application of Circulation Control Aerodynamics to a Helicopter Rotor Model,” American Helicopter Society, Paper AHS-704, May 1973. 16Nielson,J. N. (ed.), “Proceedings of the Circulation Control Workshop, 1986,” NASA Ames Research Center, NASA CP-2432, Feb. 1986. ”Shrewsbury , G., “Numerical Evaluation of Circulation Control Airfoil Performance Using Navier-Stokes Methods,” AIAA Paper 86-0286, Jan. 1986. 18Novak, C. J., and Cornelius, K. C., “An LDV Investigation of a Circulation Control Airfoil Flowfield,” AIAA Paper 86-0503, Jan. 1986. 19 Englar, R. J., “Experimental Investigation of the High Velocity Coanda Wall Jet Applied to Bluff Trailing Edge Circulation Control Airfoils,” DTNSRDC, Report 4708, Aero Report 1213, AD-A-019-417, Sept. 1975; also M.S. Thesis, Dept. of Aerospace Engineering, Univ. of Maryland, College Park, MD, June 1973. ”Wood, N., and Nielson, J., “Circulation Control Airfoils Past, Present, and Future,” AIAA Paper 85-0204, Jan. 1985. ”Williams, R. N., Leitner, R. T., and Rogers, E. O., “X-Wing: A New Concept in Rotary VTOL,” presented at AHA Symposium on Rotor Technology, Aug. 1976.
66
R. J. ENGLAR
”Reader, K. R., and Wilkerson, J. B., “Circulation Control Applied to a High Speed Helicopter Rotor,” DTNSRDC, Rept. 77-0024, Feb. 1977. 23Rogers, E. O., Schwartz, A. W., and Abramson, J. S., “Applied Aerodynamics of Circulation Control Airfoils and Rotors,” presented at 41 st Annual AHS Forum, May 1985. 240ttensoser, J., “Two-Dimensional Subsonic Evaluations of a 15-Percent Thick Circulation Control Airfoil with Slots at Both Leading and Trailing Edges,” NSRDC, Rept. 4456, July 1974. 25Williams, R. M., and Cheeseman, I. C., “Potential Acoustic Benefits of Circulation Control Rotors,” presented at AHS Meeting on Rotor Acoustics, NASA Langley Research Center, May 1978. 26Englar, R. J., “Investigation into and Application of the High Velocity Circulation Control Wall Jet for High Lift and Drag Generation on STOL Aircraft,” AIAA Paper 74-502, June 1974. 27Englar,R. J., “Subsonic Two-Dimensional Wind Tunnel Investigations of the High Lift Capability of Circulation Control Wing Sections,” DTSNRDC, Rept. ASED-274, April 1975. ”Englar, R. J., “Circulation Control for High Lift and Drag Generation on STOL Aircraft,” A I M Journal of Aircraft, Vol. 12, No. 5, 1975, pp. 457-463. 29Englar,R. J., Trobaugh, L. A., and Hemmerly, R. A., “Development of the Circulation Control Wing to Provide STOL Potential for High Performance Aircraft,” AIAA Paper 77578, June 1977. 30Englar, R. J., “Circulation Control Technology for Powered-Lift STOL Aircraft,” Lockheed Horizons, No. 24, Sept. 1987. 31Englar, R. J., Hemmerly, R. A., Moore, H., Seredinsky, V., Valckenaere, W. G.,and Jackson, J. A., “Design of the Circulation Control Wing STOL Demonstrator Aircraft,” AIAA Paper 79-1842, Aug. 1979; also published in Journal of Aircruft, Vol. 18, No. 1, 1981, pp. 51-58. 32Englar,R. J., “Development of the A-6/Circulation Control Wing Flight Demonstrator Configuration,” DTNSRDC, Rept. ASED-79/01, Jan. 1979. 33Mayfield, J., “Circulation Control Wing Demonstrates Greater Lift,” Aviation Week and Space Technology, March 19, 1979. 34Pugliese,A. J., and Englar, R. J., “Flight Testing the Circulation Control Wing,” AIAA Paper 79-1791, Aug. 1979. 35Loth, J. L., Fanucci, J. D., and Roberts, S. C., “Flight Performance of a Circulation Control STOL Aircraft,” AIAA Paper 74-994, April 1974; also published in Journal of Aircruft, Vol. 13, No. 3, 1976, pp. 169-173. 36Roberts, S. C., “West Virginia University Circulation Control STOL Aircraft Flight Test,” WVU Aerospace, Technical Rept. No. 42, July 1974. 37 Englar, R. J., “Low-Speed Aerodynamic Characteristics of a Small Fixed-TrailingEdge Circulation Control Wing Configuration Fitted to a Supercritical Airfoil,” DTNSRDC, Rept. ASED-81/08, March 1981. 38Englar,R. J., and Huson, G.G.,“Development of Advanced Circulation Control Using High-Lift Airfoils,” AIAA Paper 83-1847 July 1983; also published in Journal ofAircraft, V O ~21, . NO. 7, 1984, pp. 476-483. 39Carr, J. E., “An Aerodynamic Comparison of Blown and Mechanical High Lift Airfoils,” AIAA Paper 84-2199, Aug. 1984. 40Englar, R. J., “Two-Dimensional Transonic Wind Tunnel Tests of Three 15-PercentThick Circulation Control Airfoils,” NSRDC, Technical Note AL-182, AD 882-075, Dec. 1970.
OVERVIEW OF CC PNEUMATIC AERODYNAMICS
67
41Englar, R. J., Smith, M. J., Kelley, S. M., and Rover 111, R. C., “Development of Circulation Control Technology for Application to Advanced Subsonic Transport Aircraft,” AIAA Paper 93-0644, Jan. 1993; also published in Journal of Aircraft, Vol. 31, NO. 5, 1994, pp. 1160-1177. 42Englar, R. J., Niebur, C. S., and Gregory, S. D., “Pneumatic Lift and Control Surface Technology Applied to High Speed Civil Transport Configurations,” AIAA Paper 970036, Jan. 1997. 43Mavris, D. N., Kirby, M. R., Lee, J. M., Qui, S., Roth, B., Tai, J., and Englar, R. J., “Systems Analyses of Pneumatic Technology for High Speed Civil Transport Aircraft,” GTRI Final Technical Rept. A-5676, Oct. 1999. 44Ni~hols, J. H., Jr., Englar, R. J., Hams, M. J., and Huson, G.G.,“Experimental Development of an Advanced Circulation Control Wing System for Navy STOL Aircraft,” AIAA Paper 81-0151, Jan. 1981. 45Harris,M. H., Nichols, Jr., J. H., Englar, R. J., and Huson, G.G.,“Development of the Circulation Control WingIUpper Surface Blowing Powered-Lift System for STOL Aircraft,” Proceedings of the ICASIAIAA Aircraft Systems and Technology Conference, Paper ICAS-82-6.5.1, Aug. 1982. 46Yang,H. T., and Nichols, Jr., J. H., “Design Integration of CCW/USB for a Sea-Based Aircraft,” Paper ICAS-82-1.6.1, Aug. 1982. 47Englar, R. J., Nichols, Jr., Harris, J. H., Eppel J. C., and Shovlin, M. D. “Circulation Control Technology Applied to Propulsive High Lift Systems,” Society of Automotive Engineers, Paper 841497, Oct. 1984. 48Lowndes, J. C., “Aeronautical Engineering: Studies Show Lift Coefficient Tripling,” Aviation Week and Space Technology, Dec. 1, 1980. 49Englar,R. J., Nichols, Jr., J. H., Hams, M. J., Eppel, J. C., and Shovlin, M. D., “Development of Pneumatic Thrust-Deflecting Powered-Lift Systems,” AIAA Paper 86-0476, Jan. 1986. ”Eppel, J. C., Shovlin, M. D., Jaynes, D. N., Englar, R. J., and Nichols, Jr., J. H., “Static Investigation of the CCW/USB Concept Applied to the Quiet Short-Haul Research Aircraft,” NASA, TM 84232, July 1982. 51Shovlin, M. D., Englar, R. J., Eppel, J. C., and Nichols, Jr., J. H., “Large-Scale-Static Investigation of Circulation-Control-Wing Concepts Applied to Upper-Surface-Blowing Aircraft,” NASA, Technical Paper 2684, Jan. 1987. 52Bevilaqua,P. M., and Lee, J. D., “Design of Supersonic Coanda Jet Nozzles,” in Proceedings of the Circulation Control Workshops 1986, NASA, CP 2432, pp. 289-312, Feb. 1986. 53Englar, R. J., and Campbell, B. A., “Development of Pneumatic Channel Wing Powered-Lift Advanced Super-STOL Aircraft,” AIAA Paper 2002-2929; presented at AIAA 20th Applied Aerodynamics Conference, June 25, 2002. 54Englar, R. J. and Campbell, B. A., “Experimental Development and Evaluation of Pneumatic Powered-Lift Super-STOL Aircraft,” NASAIONR Circulation Control Workshop, March 2004; also published in NASA CP 2005-213509, 2005. 55Braslow, A. L., “Aerodynamic Evaluation of Circulation Control Propellers,” Bionetics Corp., NASA Contractor Report 165748, June 1981. 56Taback, I., Braslow, A. L., and Butterfield, A. J., “Circulation Control Propellers for General Aviation, Including a BASIC Computer Program,” NASA Contractor Rept. 165968, April 1983. 57Gamer, D., “No Moving Parts, The Circulation Control Airfoil and Fluidic Propeller,” EAA Sport Aviation, Vol. 37, No. 3, 1988, pp. 27-30.
68
R. J. ENGLAR
58Wilson, M. B., and von Kerczek, C., “An Inventory of Some Force Producers for Use in Marine Vehicle Control,” DTNSRDC-79/097, Nov. 1979. 59Wellman, L. K., and Jacobsen, C., “Wind Tunnel Investigation of the Application of Circulation Control to a Forward Swept Wing,” DTNSRDC/ASED-82/05, June 1982. 60 Englar, R. J., “Pneumatic High-Lift and Control Surfaces Applied to Micro-Aerial Vehicles,” Proceedings of GTRI International Conference on Emerging Technologies for Micro Air Vehicles, Feb. 1997. 610yler, T. E., and Palmer, W. E., “Exploratory Investigation of Pulsed Blowing for Boundary Layer Control,” North American Rockwell, Rept. NR72H-12, Jan. 1972. 62Walters, R. E., et al. “Circulation Control by Steady and Pulsed Blowing for a Cambered Elliptic Airfoil,” West Virginia Univ., Dept. of Aerospace Engineering, Rept. TR-32, July 1972. 63Jones,G. S., and Englar, R. J., “Advances in Pneumatic-Controlled High-Lift Systems Through Pulsed Blowing,” AIAA Paper 2003-3411; presented at AIAA 21st Applied Aerodynamics Conference, June 2003. 64Englar, R. J., and Williams, R. M., “Design of a Circulation Control Stem Plane for Submarine Applications,” NSRDC, Technical Note AL-200, March 1971. 65F~ry, R. J., and Whitehead, R. E., “Static Evaluation of a Circulation Control Centrifugal Fan,” DTNSRDC, Rept. 77-0051, AD A041-463, June 1977. %aeta, R. J., and Englar, R. J., “Pneumatically Augmented Aerodynamic Heat Exchanger,” Paper presented at NASA/ONR Circulation Control Workshop, March 2004; also published in NASA CP 2005-213509, 2005. 67Englar, R. J., Smith, M. J., Niebur, C. S., and Gregory, S. D., “Development of Pneumatic Aerodynamic Concepts for Control of Lift, Drag, Moments and Lateral/Directional Stability of Automotive Vehicles,” Society of Automotive Engineers, Paper 960673, Feb. 1996; also published in SAE SP-1145, “Vehicle Aerodynamics,” Feb. 1996. 68EnglarR. J., “Drag Reduction, Safety Enhancement and Performance Improvement for Heavy Vehicles and SUVs Using Advanced Pneumatic Aerodynamic Technology,” 2003 SAE International Truck and Bus Meeting and Exhibition, Society of Automotive Engineers, Paper 2003-01-3378, Nov. 2003. 69 Englar, R. J., “The Application of Pneumatic Aerodynamic Technology to Improve Performance and Control of Advanced Automotive Vehicles,” NASA/ONR Circulation Control Workshop, March 2004; also published in NASA CP 2005-213509, 2005.
Chapter 3
Exploratory Investigations of Circulation Control Technology: Overview for Period 1987-2003 at NSWCCD Robin Imber* Naval Air Systems Command, Patuxent River, Maryland and Ernest Rogerst and Jane Abramsont Naval Sur$ace War$are Center-Carderock Division, West Bethesda, Maryland
Nomenclature A = area of CC slot, or area of foil planform AR = aspect ratio C, = momentum coefficient of slot flow (rizvj/qS) CL = lift coefficient ( L / q S ) C, = drag coefficient ( D / q S ) C, = power coefficient CT = thrust coefficient c = chord length D = drag force d = diameter, or camber line offset h = CC slot exit height (gap) L = lift force rit = mass flow (pAV) PR = pressure ratio q = dynamic pressure (&pV2) S = planform area of lifting surface t = thickness of airfoil
*Aerospace Engineer. 'Aerospace Engineer, retired. This material is declared a work of the U.S.Government and is not subject to copyright protection in the United States.
69
R. IMBER, E. ROGERS, AND J. ABRAMSON
70
V = velocity a = angle of attack ACL/AC, = lift augmentation ratio (slope of CL vs C, curve) p = density u = rotor solidity, cavitation index
Subscripts
i = induced j =jet I. Introduction EGINNING in 1967, when the Naval Surface Warfare Center, Carderock Division (NSWCCD) was known as the David Taylor Model Basin (DTMB), researchers there were involved with many circulation control (CC) exploratory projects, including CC-airfoils, CC-centrifugal fans, dual-directional CC-airfoils, CC-fixed wings, including the A6 aircraft modification, CC-rotorcraft, including XH2-CCR and X-wing, CC-hydrodynamic applications, and valving systems for CC. The first Circulation Control Workshop (unpublished) was held at DTMB in 1971, and the second was held at the National Aeronautics and Space Administration, Ames Research Center, in 1986.' Papers, presentations, and reports of the research performed from 1967 to 1985 at what is now NSWCCD (there have been several name changes since 1967) are cited in the proceedings from the 1986 CC Workshop and more can be found in Ref. 2. This overview is intended as a brief summary of the highlights of six of the major CC experimental investigations that have taken place at NSWCCD since the second CC Workshop, specifically between 1987 and 2003, and was presented at the 2004 Circulation Control Workshop held in Hampton, Virginia.3 The following investigations are discussed: 1) The Dual-Slotted Cambered Airfoil, LSB; 2) The Self-Driven Rotary Thruster, TIPJET; 3) The Annular Wing, CC-Duct; 4) The Circular Wing, CC-Disc; 5 ) The Miniature Oscillatory Valve, CC-Valve; and 6) The Dual-Slotted Low Aspect Ratio Wing, CCHydrofoil. For further details regarding these investigations, the reader is encouraged to examine the publications that are referenced for each of the projects. Used throughout this summary is a frequently used measure of CC performance, the lift augmentation ratio. This ratio is defined as the ratio of the gain in lift (ACL)to the change in slot flow momentum (AC,). In this review, the ratio is determined from experimental data by assessing the slope of the lift response in the low blowing (low C ), range, where the response is usually linear.
B
11. Dual-Slotted Cambered Airfoil (LSB) The Dual-Slotted Cambered Airfoil, also referred to as the LSB (lower surface blowing), was designed and tested in 1987 by Abramson and colleague^.^,^ The inclusion of a lower surface slot along with the usual upper surface slot provides the ability to produce lift in both positive and negative directions. The presence of camber in this model, along with the objective of preserving the contour of a
INVESTIGATIONS OF CC TECHNOLOGY AT NSWCCD
71
Fig. 1 Photograph of the dual-slotted cambered airfoil (LSB).
proven (parent) single-slot CC airfoil, means that the geometric properties of the lower-slot region are not the same as those of the upper-slot region. Additionally, when operating in the lower surface blowing mode, the CC section functions with negative camber, all with unexplored consequences at that time. A photograph of the LSB model is shown in Fig. 1 and a cross-section drawing of the model is shown in Fig. 2. The LSB has a 17% thickness ratio with 1.1% camber. It was constructed with a 12-in. chord and a 36-in. span. The upper slot is located at 96.8% chord and the lower slot is slightly further aft at 97.0% chord. The airfoil was experimentally evaluated in the NSWCCD 8 x 10 ft wind tunnel configured with two-dimensional wall inserts. Testing included three blowing modes: upper surface only, lower surface only, and dual blowing. The wind tunnel dynamic pressure ranged from 20 to 60psf, Reynolds number ranged from 0.8 to 1.4 x lo6, and angle of attack (AOA) ranged from - 10 to +10 deg. Two slot height-to-chord (h/c) ratios were set: 0.0013 and 0.0020. The maximum momentum coefficient (C), was 0.22. One of the main design goals was to have the dual-slotted model perform as well, when using only the upper slot, as the single slotted “parent” model. The dual-slotted model had the same cross-section as the parent model. Lift performance results from the single and dual slot models are shown in Fig. 3. The comparison shows that there was no detrimental effect in adding the second slot. The second design objective was to increase the control range so that force control in both directions was available. Figure 4 displays a plot of lift coefficient Upper Surface Blowing Slot Coanda Surface
Air Supply Ducts
Fig. 2 Cross-section of LSB airfoil!
Lower Surface Blowing (LSB) Slot
R. IMBER, E. ROGERS, AND J. ABRAMSON
72
ACr due to blowing
Momentum Coefficient, Cp
Fig. 3 Comparison of lift performance for dual-slotted LSB and single-slot parent airfoil?
against momentum coefficient, and reveals that the goal of doubling the control range was met. An unanticipated finding was that the performance of the lower slot, in terms of measured lift augmentation ratio, was noticeably better than that for the upper slot (80 compared with 60). This empirically unexpected performance enhancement illustrates the need for a well-validated computational code (computational fluid dynamics; CFD) to help guide future CC designs. Another finding was the effect of simultaneous blowing5 At the only ratio of dual blowing examined,
Lift Coefficient
Momentum Coefficient
Fig. 4 Control range increase demonstrated with upper and lower slot ~apability.~
INVESTIGATIONS OF CC TECHNOLOGY AT NSWCCD
73
where the lower slot flow momentum level was 25% that of the upper slot, activation of the lower slot decreased lift. This dual-slotted airfoil helped pave the way for another dual-slotted CC investigation discussed later, the CC-Hydrofoil. Summarizing the key findings from the Dual-Slotted Cambered Airfoil investigation, it was found that 1) incorporation of a lower slot did not affect performance of the upper slot; 2) the available lift control range was doubled, as expected; 3) a lift augmentation ratio of 80 for the lower slot was obtained; and 4) simultaneous blowing can be used to decrease (control) the lift increment produced by single-slot operation.
111. Self-Driven Rotary Thruster (TIPJET) Experimentally investigated in 1991 by several NSWCCD engineers, the SelfDriven Rotary Thruster was the first integrated lift/reaction-drive rotor system combining Cbanda CC aerodynamics with cold cycle reaction drive technologies.6-8 The rotor was developed as part of the TIPJET unmanned air vehicle, shown conceptually in Fig. 5 . The design involves a stoppable two-bladed rotor concept where, after lifting off vertically in rotary mode and accelerating forward, the rotor transitions to a fixed wing to enable high-speed flight. A “cold cycle” gas generator, such as the fan stage of a turbofan engine, supplies the compressed air for both the circulation control and the tip jets that provide rotor drive torque. Figure 6 shows a sketch of the completely pneumatic rotor. Circulation control slots are located along most of the rotor blade span on both the leading and trailing edges. Reaction drive nozzles are located at the rotor tips. Thus, a single source of air pressure provides flow for the CC slots to augment rotor lift (vertical Y
Fig. 5 TIPJET vertical takeoff and landing unmanned air vehicle with circulation controlled stoppable rotor.6
74
R. IMBER, E. ROGERS, AND J. ABRAMSON
No drive shaft, unarticulated, flat-pitch blades
NOZZLE DRIVE FORCE
AIR SUPPLY
Fig. 6 Sketch of TIPJETcompletely pneumatic rotor?
thrust), while at the same time providing the rotor torque drive via the tip-jet nozzles. (The full-scale application concept called for in-flight controllable slot gap settings.) A detailed investigation of the TIPJET rotor in hover took place in 1991.7 The primary objective of the hover experiment was to evaluate the interactions between the lift and drive systems. Drawings of the aluminum rotor model are shown in Fig. 7 and specifications of the model are listed in Table 1. The 80in. rotor blade is tapered, with no twist and zero pitch angle. The thickness and camber varies linearly with radius from the 25% to 95% span locations. Figure 8 is a photograph of the blade tip region, showing the CC slot along the span and the tip-drive nozzle. During the hover test, the rotor could be driven by either an electric drive motor that enabled the rotor to be operated at selected rpm settings while investigating specific performance attributes, or by the tip-jet reaction drive. To better understand the performance of the integrated lift/drive system, a detailed investigation was conducted with the tip nozzles closed and the rotor mechanically driven. Figure 9 shows the experimental data for the measured rotor thrust coefficient as a function of momentum coefficient for several slot height settings. The slope of the curve is 29 at the lower values of CJu. This ratio is higher than that of any previously tested CC rotor. As shown in Fig. 9, this measure of efficiency was independent of the four slot heights tested. To determine if the level of understanding of the performance of the fully pneumatic rotor system was sufficient for successful incorporation into a flight vehicle, numerical calculations were developed and compared to the experimental performance. The results of this comparison, shown in Fig. 10, indicate excellent correlation for both the rotor thrust developed and drive power required. The ultimate goal of the experimental rotor investigation was to determine the aeromechanics of the model rotor in self-drive mode. The behavior of the rotational speed in response to pressure input was unknown at the beginning of
INVESTIGATIONS OF CC TECHNOLOGY AT NSWCCD
SLOT HEIGHT ADJUSTMENT SIDE-BY-SIDEOPPOSING SCREWS
Fig. 7 Drawings with details of TIPJET pneumatic rotor?
75
76
R. IMBER, E. ROGERS, AND J. ABRAMSON
Table 1 TIPJET rotor specifications’ Blade Fixed pitch angle (OJ, deg Rotor diameter, ft Number of blades Chord, in. 25% span 93% span Solidity ratio Geometric twist, deg Airfoils Thickness ratio ( t / c ) Camber ratio (d/c) Trailing edge radius (rt,/c) Slot height ( h / c ) Tipjet nozzles (rectangular) Area/nozzle, in.*
0 6.67 2 7.95 5.40 0.110 0 25% span 0.213 0.053 0.05 variable
93% span 0.170 0.011 0.03 variable
0.764
the test. When a slot height and tip nozzle area are set, the blade pressure input is the only determining factor of the operating condition. Shown schematically in Fig. 11, as pressurized air is introduced into the rotor, the rotor rotates in reaction to the flow from the tip nozzles. As the nozzle flow increases, the rotor lift will increase as a result of using a cambered airfoil and, most especially, as a result of the increased circulation due to the CC slots. This additional lift increases the required drive torque, which then limits the rotational rate for a given air pressure setting. In essence, pressure input simultaneously influences the lift and produces the torque drive. Identifying the behavior of a system coupled by these two effects was one of the objectives of the experiment.
Fig. 8 Tip region of the fully pneumatic rotor model.
INVESTIGATIONS OF CC TECHNOLOGY AT NSWCCD
77
c
f SLOT MOMENTUM COEFF. / SOLIDITY, C ~ / O Fig. 9 TIPJET rotor thrust performance when mechanically driven at constant rpm.’
It was discovered that, in full self-drive mode, the rotational speed is stable and exhibits a self-limiting maximum for a given ratio of slot area to nozzle area. The data in Fig. 12 reveal the nature of the self-limiting rotational speed. Rotational tip speed is shown as a function of blade pressure for several slot height settings. At each of the slot settings, the pressure input response of the rotor is to increase the rotational rate until a limiting tip speed is reached. It was demonstrated that the value of the limiting tip speed is a function of the slot height setting; increasing the slot height results in a lower limiting tip speed. However, because CC-based lift is not solely dependent on local velocity, the lift response is not limited to a tip speed maximum. Figure 13 shows that the rotor lift is essentially always the same linear function of the applied pressure, independent of actual rotational rate (compare Figs. 12 and 13). A major finding was that a non-shaft-driven completely pneumatic rotor inherently seeks a rotational rate that results in lift being a near-linear function of the blade pressure input, and this linear lift is easily controllable by pressure throttling. In addition to the specific TIPJET vehicle application, this capability can be applied to systems that require mechanically simple, easily controlled thrusters. Summarizing some of the key findings from the Self-Driven Rotary Thruster investigation, it was found that 1) a lift augmentation ratio of 29 was obtained when in the mechanically driven mode; 2) the pneumatic rotor inherently seeks equilibrium and the self-limiting rotational rate is a function of slot-to-drivenozzle area ratio (the resulting rotor lift is a near linear function of the blade pressure); 3) there is a significant impact on induced power efficiency because of the non-lifting tip nozzle region; and 4) the presence of the tip nozzle jet has no discernible impact on the external aerodynamics of the lift system.
78
R. IMBER, E. ROGERS, AND J. ABRAMSON
. D
c
Fig. 10 TIPJET performance numerical analysis correlation with experimental results.'
cc slot area
Slot flow
1
t
r3
I
Internal air
Rotor
Drive
rotor drive torque
Fig. 11 Conceptual schematic of TIPJET rotational rate equilibrium.
INVESTIGATIONS OF CC TECHNOLOGY AT NSWCCD
79
d a, a,
Q
v)
Blade Root Pressure Ratio, PR ,oo,
Fig. 12 TIF'JET rotor characteristics when self-driven via tip-jet nozzles?
IV. Annular Wing (CC-Duct) In the mid-1990s the performance characteristics of an annular wing, having both inner and outer trailing edge circulation control slots, were explored. The focus of this investigation was to apply full, or partial, perimeter trailing
Blade Root Pressure Ratio, PR,,,,
Fig. 13 Relationship of blade pressure to thrust developed.' (Data and symbols are the same as in Fig. 12.)
80
R. IMBER, E. ROGERS, AND J. ABRAMSON
Fig. 14 Annular wing model installed in free jet wind tunnel?
edge CC fluid ejection to a propulsor duct to enhance maneuvering control on watercraft via a thrust vectoring capability. An existing CC-Duct model with inner and outer trailing edge CC slots had been borrowed from West Virginia University. The model has been used in the 1970s to investigate the attributes of variable diffusion for ducted fans on aircraft.’ The CC-Duct model is shown in Fig. 14 as it was installed in the Atlantic Applied Research Corporation open-jet acoustic tunnel in 1993. The model originally had a motor housing and stator that were removed for the CC-Duct investigation discussed here. The objective for removing the motor housing and stator was to have a simple configuration in which to establish an understanding of the performance attributes, and to provide a data set for correlation to basic ring-wing theory. No propeller was present in any of the test series. A cross-section of the top of the CC-Duct is shown in Fig. 15 and its geometry is presented in Table 2. The model is 18411. in diameter with a 10-in. chord and a 20% thick uncambered foil section. The inner and outer slots are located around the full trailing edge circumference at 97% chord. During the Duct investigation, metal foil tape was applied and burnished well to the external surface of the model, over the slot region, in order to temporarily block off portions of the slot. This commonly used technique for exploratory CC research provides the ability to control the distribution of slot flow. By using the tape, several configurations, focusing on potential attributes of the Duct, were tested.
INVESTIGATIONS OF CC TECHNOLOGY AT NSWCCD
81
Dimensions in inches
Fig. 15 Annular wing model cross-section (at top).’
The model and experimental arrangement provided the opportunity to examine many interesting flow control configurations. Icons representing six basic configurations, or modes of operation, and a brief explanation of each, are displayed in Table 3. The dotted line on each of the icons represents the trailing edge of the CC-Duct. The solid lines represent the portion of either the inner or outer slot that is open and where the fluid ejection occurs. The author suggests reviewing Table 3 before reading further. As a simple illustration of force vectoring capability, a long strand of yam was positioned in the center of the CC-Duct. The photograph on the left of Fig. 16 shows the configuration with no CC blowing. The yam is aligned with the tunnel free stream velocity, along the longitudinal axis of the Duct. The photograph on the right of Fig. 16 was taken for a complimentary halves configuration, where the outer lower slot is active and the upper inner slot is active. The yam strand is now at an angle to the free stream, indicating the wake deflection due to the force vectoring brought about by the active flow control. Quantitative data from the configuration on the right in Fig. 16 are shown in Fig. 17 as a plot of force developed as a function of C ., The reference area used
Table 2 Annular wing model specifications Model geometry Outside diameter Inside diameter Chord Slot gap hlc Slot position d/c (16.2110) AR (effective)
Dimensions, in. 18.2 14.2 10
0.009 0.0009 0.970 c 1.62 2.1
R. IMBER, E. ROGERS, AND J. ABRAMSON
82
Table 3 CC annular wing modes of operation as a marine propulsor duct Aft view of ducta
Slot ejection configuration
Effect
Operation benefit
Inner slot only
Increased duct flow-through: accelerating nozzle
Higher prop efficiency
Outer slot only
Decreased duct flow: diffusion
Reduced cavitation
Complementary quadrants
Side-force: yaw
Steerage
Complementary quadrants
Side-force: pitch
Depth keeping
Alternating
Vortex generation: very high drag, no side force
Braking: crash-back
Both slots
Drag reduction, auxiliary thruster
Cruise efficiency, dock side positioning
aDashed line is trailing edge; solid lines represent active slot.
for the force coefficients is duct length times duct diameter (c x d). The lift force developed for this configuration, even at zero model pitch angle, is more than twice that available on a passive duct. When interpreting the CC performance of ring wings, in comparison to that of flat wings or airfoils, it is important to be observant of how the performance parameters are nondimensionalized. In Fig. 17, note that the CL versus C, performance curve for the duct is close to that of some CC airfoils (e.g., see Fig. 4). This result occurs even though the finite wing effect of shed wake vorticity causes a downwash that should decrease the net lift response to roughly half that of a two-dimensional CC foil. The explanation for the apparent two-dimensional-like performance level has to do with the reference areas used for the coefficients. Consistent with the practice for CC airfoils and wings, the duct C, is based on the full slot length (Tx d x c). At the same time, as is standard practice, CL is defined based on projected area (c x d ) , without consideration that there are two lifting surface areas involved: the upper and lower halves of the ring-wing. Therefore, the CL that should be compared to that of an airfoil is half the CL shown in Fig. 17, and thus
INVESTIGATIONS OF CC TECHNOLOGY AT NSWCCD
No Blowing
83
180 I 180 deg blowing
Fig. 16 Lateral force capability; wake deflection with asymmetric trailing edge CC blowing?
matches what would be expected from a wing having an as ect ratio (AR) of 2.1, which is the equivalent AR for the geometry of the duct."(See the later discussion and performance of the CC-Hydrofoil "flat" wing, which has about the same aspect ratio.)
Slot Flow Momentum Level CF = mV,/( % pVm2ndc)
Fig. 17 CC annular wing lift and drag performance' (the drag force is in the direction of a negative propulsive force).
84
R. IMBER, E. ROGERS, AND J. ABRAMSON
Test results met expectations including that the drag (plotted as negative thrust) is a linear function of C , as shown in the experimental data, and derived as in the following equations. From the experimental data,
CL = 1 o G From Hoerner and Borst"
AR = 2.1
(2)
and for lift-induced drag
then
One of the findings from the CC-Duct investigation was the ability for braking, or control of induced drag, without a change in net lift. Alternating the active inside and outside slots every 90deg creates two pairs of counterrotating vortices. Figure 18 shows the measured performance demonstrating this capability. The drag is about the same as it was when lift was being developed. Figure 19 represents a VSAERO potential-flow solution of surface pressure and wake filaments for a CC-Duct braking configuration. (A discussion of
Slot Flow Momentum Coefficient (Cp) Fig. 18 Braking configuration measured data; lift-induced drag without net lift?
INVESTIGATIONS OF CC TECHNOLOGY AT NSWCCD
85
Wake Filaments looking upstream
Fig. 19 Braking configuration computational solution."
potential flow solution techniques related to this work, and CC applications in general, can be found in Ref. 11.) The key findings from the CC-Duct investigations include 1) lift and side force can be generated using specific blowing segments; 2) at zero angle of attack (AOA), forces of almost 2.5 times a conventional ring-wing are possible; 3) a braking force via induced drag is available without the development of lift; and 4) the performance met expectations and it was found that the performance can be predicted using a potential flow code, although the slot flow requirements have to be estimated from the historical CC airfoil database.
V. Circular Wing (CC-Disc) In 1995, Rogers and Imber created a circular wing model with a full perimeter circulation control capability. l 2 The purpose was to investigate the effectiveness of CC on very low aspect ratio wings and to explore the attributes of a CCenhanced omnidirectional type of control surface or vehicle. The CC-Disc, also known as the Coanda Disc, was tested in the 8 x loft Subsonic Wind Tunnel at NSWCCD, employing a six-component external balance for force and moment measurement. Figure 20 is a photograph of the anodized aluminum 2-ft-diam. model and Table 4 lists the model specifications. The Disc has a 19% thick cross-section with 2.4% camber. Figure 21 shows a drawing of the
86
R. IMBER, E. ROGERS, AND J. ABRAMSON
Fig. 20 Circular wing model with full perimeter circulation control.
centerline cross-section with surface pressure tap locations. The baseline slot height was 0.032 in. The major configurations investigated are shown in Fig. 22. The circular icons represent a planform view of the Disc with the shaded sections representing the perimeter sections where fluid ejection occurred. In Fig. 22, the free-stream flow would be directed from the top to the bottom of the page. The three groups of configurations were 1) increasing area centered about the trailing edge; 2) constant area at variable azimuth; and 3) lateral and asymmetric variations. The same slot tape-over technique used in the CC-Duct test was employed for flow control configuration changes on the CC-Disc. Results from the first configuration group are shown in Fig. 23 for the model at zero pitch angle, gradually increasing the flow ejection circumference region. Lift is shown as a function of the region of blowing, starting with unblown and then, centered about the trailing edge, increasing the perimeter region blown until there was full 360-deg fluid ejection. The lines on the plot are for constant
Table 4 Specifications of circular wing Diameter (chord) Reference area (S) Aspect ratio (AR) Thickness ( t l c ) Camber Coanda radius Slot position Slot lip thickness Slot height ( h ) hlc hlrs Pressure tap diameter
2 ft 3.14 ft2 1.27 19% 2.4% rs/c = 0.050 rte/c = 0.040 3.2%c from edge 0.026 in. 0.032 in. 0.0013 0.027 0.040 in.
\
INVESTIGATIONS OF CC TECHNOLOGY AT NSWCCD
87
Surface Pressure Taps
Centerline Cross-section
Fig. 21 Circular wing model with full perimeter circulation control.'*
blowing coefficient C,. The optimum, or highest lift, configuration varied somewhat with the C, level. The overall highest lift was obtained using a 225-deg perimeter of fluid ejection centered about the trailing edge. It is notable that high lift performance was obtained even with full perimeter blowing, showing that the omnidirectional configuration is viable. Figure 24 reveals lift performance results for the 225-deg perimeter configuration at several AOA and C, levels. The maximum C, was more than twice that available from a disc without circulation control. Both with and without blowing, the slope of C,/a closely matches the slope calculated from inviscid theory. The same configuration and data collection series is shown in Fig. 25 plotted as lift
lncreasina Area Centered About the Trailina edae:
Fig. 22 Circular wing test configurations: planform view (sections with blowing shown in black).
aa
R. IMBER, E. ROGERS, AND J. ABRAMSON
Configuration
Fig. 23 Lift as a function of azimuthal mass ejection coverage.'*
coefficient squared against drag coefficient. The measured induced drag matches the prediction of lifting surface wing theory. These matches suggest that, for low aspect ratio wings, there are no basic effects unique to lift development by means of the Coanda form of CC.
Angle-of-Attack (deg) Fig. 24 Influence of angle of attack?
INVESTIGATIONS OF CC TECHNOLOGY AT NSWCCD
89
CD Fig. 25 Aerodynamic efficiency: induced drag with incidence?
As discussed in more detail in Ref. 12, the circular wing with CC has two aerodynamic centers: 1) lift due to angle of attack and 2 ) lift due to circulation control. The pitching moment map in Fig. 26 demonstrates the results of these two centers. The pitching moment was established about the center of
moment
Fig. 26 Pitching moment map resulting from two aerodynamic centers?
90
R. IMBER, E. ROGERS, AND J. ABRAMSON
CP Fig. 27 Sensitivity to slot gap setting?
the disc and the plot can be used to determine the combination of pitch angle and level of circulation control that will provide a specific desired maneuvering effect. During part of the CC-Disc investigation, the slot height around the perimeter was changed to determine the sensitivity of wing performance to slot gap setting. For the range of C, shown in Fig. 27, there was essentially no change in lift performance for a 4:l slot gap change. The lack of performance change with the large change in slot gap is considered a desirable attribute, because it allows flexibility in the selection of a slot-flow supply system. For many of the configurations investigated, upper and lower surface pressures were measured every 10 deg in azimuth. A sample of the upper surface pressure distributions of four configurations is shown in Fig. 28. The area of fluid ejection is easily seen as the darker shaded regions around the perimeter, which have significant low pressure values and steep pressure gradients. The radial lines that appear in Fig. 28 are a product of the plotting software and not a true characteristic of the overall pressure distribution. There were many key findings from the Circular Wing test. The investigation demonstrated that 1) circulation control is effective on very low aspect ratio lifting surfaces and, for a circular planform, can provide an omnidirectional capability when full perimeter blowing is applied; 2) with at least 225 deg of flow control around the perimeter the lift produced was more than double that of an unblown circular wing (the limit to augmented lift is believed to be the result of excessive wall jet turning); 3) roll control was demonstrated using asymmetric blowing; 4) lift control without change in pitching moment was demonstrated when blowing only the lateral edges; and 5 ) sensitivity to the 4:l change in slot gap is minimal.
INVESTIGATIONS OF CC TECHNOLOGY AT NSWCCD
91
High
pressure
Low pressure Full perimeter blowing
Fig. 28 Upper surface pressure results showing four configurations?
VI. Miniature Oscillatory Valve (CC-Valve) for Unsteady Wing Load Reduction There were two main objectives for this project. The first objective was to demonstrate that mass ejection in the trailing edge region of a hydrofoil could be used to cancel periodic unsteady hydrodynamic loading. The second objective was to show that a practical closed-loop control system could be devised and that the required oscillatory valving could be miniaturized and incorporated into the trailing edge region of a hydrofoil. The focus of the miniature valve design and control demonstration was to develop the capability to cancel unsteady foil forces and be automatically adaptive to upstream disturbances. Fry and Jessup designed and tested the slot control valve in 1993.13-15 An overall sketch of the 15-in. chord, dual-slotted hydrofoil used in this demonstration is shown in Fig. 29. The flow ejection for this application was normal to the surface, as in a jet flap. (The concept could be adapted to the production of tangential mass ejection). Figure 30 shows a trailing edge section-cut of the
dual-slotted, surface-normal mass ejection into the boundary layer
Fig. 29 Dual-slotted hydrofoil used for miniature valve proof of ~ 0 n c e p t . l ~
92
R. IMBER, E. ROGERS, AND J. ABRAMSON
electromagnetic actuator
14
Fig. 30 Trailing edge of dual-slotted hydrofoil: rotor pivots to open/close upper lower slots. Constant fluid pressure results in high-response-rate,efficient system.
model and Fig. 3 1 depicts the actuator mechanism construction. A small rocker valve embedded in the trailing edge uses an electromagnetic actuator attached to a permanent magnet assembly to produce a high-frequency response of the rocker valve. As shown in Fig. 30, there is an optional nonmovable trailing edge tail section. The main feature of the actuator was that it controlled the slot exit area and not the fluid pressure. A schematic of the water tunnel installation is shown in Fig. 32. The hydrofoil was attached to one side of a 24-in. water tunnel test section. A freewheeling propeller provided a periodic upstream flow disturbance. An external pump and fluid lines delivered fluid to the rocker valve assembly region. A controller was employed to send signals to the magnet assembly installed in the trailing edge body of the hydrofoil.
Permanent Magnet Assembly
Actuator
Coil Assembly
Rotor Valve Assembly
Fig. 31 Actuator mechanism constr~ction.'~
INVESTIGATIONS OF CC TECHNOLOGY AT NSWCCD
93
24-inch water tunnel
Fig. 32 Operational schematic of actuator test.
During operation, fluid pressure to the trailing edge of the foil is not throttled; it is simply redirected as needed. This low-rate perpendicular ejection of mass into a boundary layer, from a slot at the trailing edge, may mean that the flow effect responsible for any change in foil lift is the same as the flow effect attributed to a Gurney flap. (The term “low-rate’’ ejection is used to make a distinction from the high momentum level of a true jet flap.) As an example of the experimental results, two plots of force against frequency are shown in Fig. 33. The top plot shows the periodic force spectrum produced by the hydrofoil due to the upstream flow disturbance. The bottom plot shows the force variation with the active flow control system operating. As shown, the targeted hydrofoil load spikes were successfully eliminated by the system, with three frequencies simultaneously reduced. The key findings from the Miniature Oscillatory Valve project include the following: 1) functional model-scale actuators can follow steady or time-varying input signals up to 500 Hz;2) hydrofoil forces were successfully varied up to 110 Hz;and 3) alleviation of a high-frequency periodic hydrofoil loading is feasible.
VII. Dual-Slotted Low Aspect Ratio Wing (CC Hydrofoil) In 2002, Rogers16 was the principal investigator for an in-depth low aspect ratio hydrofoil investigation that employed dual-slotted trailing edge CC. The experimental investigation took place primarily in the Navy’s 10-ft-Large Cavitation Channel (LCC) and is extensively documented in Ref. 16. The question as to the effect of cavitation on the performance of a CC-foil had been pondered for many years and was finally answered during this research.
94
R. IMBER, E. ROGERS, AND J. ABRAMSON
Frequency (Hz)
Fig. 33 Load cancellation effectiveness. Two plots show three frequencies of unsteady lift simultaneously reduced to the broadband fl00r.l~
There are several compelling reasons to incorporate circulation control fluid dynamics on underwater vehicles. The buoyancy of waterborne vehicles means that they can and do operate down to extremely low speeds, where conventional control surfaces have very limited force generation capability. Because the forte of CC is to leverage the momentum flux from a slot to make a planar surface produce much greater force than otherwise possible, it becomes an attractive consideration for low-speed maneuvering enhancement. Furthermore, CC augmentation has its best pumping-power efficiency at low speed, in terms of the control force advantage over a conventional surface. Photographs of the model installed in the water tunnel on a reflection plane, and of the dual-slotted cross-section, are shown in Fig. 34. A more inclusive view of the test section is provided in Fig. 35. The half-span model of aspect ratio 2.0 has an uncambered 20% thick elliptical profile essentially identical to a previously tested CC airfoil, thus allowing a comparison of two- and threedimensional (finite wing) performance. The slot-height-to-chord ratio of approximately 0.0018 was maintained over the full 2-ft span of the tapered planform. The fluid pressure in the dual plenums could be individually regulated and model loads were measured by a multicomponent balance. Fundamental lift performance is shown in Fig. 36 with lift coefficient as a function of C, for the model at 10 deg AOA. At low C, levels, the wing performed slightly better than expected, producing a lift augmentation ratio of 36 in the initial linear portion of the curve. Transition from a linear to a squareroot-like response to C, occurred, as expected, at the higher blowing levels. An important discovery was made early in the test. With only the upper slot blowing, lift roll-off occurred at a much lower C, than expected, as shown on
INVESTIGATIONS OF CC TECHNOLOGY AT NSWCCD
Fig. 34 Dual-slotted low aspect ratio circulation control hydrofoil.'6
Fig. 35 Dual-slotted hydrofoil installed in the Large Cavitation Channel.
95
R. IMBER, E. ROGERS, AND J. ABRAMSON
96 3.5
AOA = 10°
3
with 2nd
slot flow
2.5 2 CL 1.5
single slot (max conventional)
1 0.5
with 5% lower slot Cµ no lower slot flow
R52
0 0
0.1
0.2
0.3
0.4
0.5
0.6
Cµ Cp total (upper + + lower)
Fig. 36 Lift performance and benefit of dual-slot activation.16
the curve in Fig. 36 labeled “single slot”. It was concluded that excessive turning of the wall jet was causing the loss in lift. The lower slot was then employed to produce a very small counter flow, no larger than 5% of the upper flow, to see if it would prevent the excessive turning. This dual-flow configuration produced the greatly improved performance shown in the upper line in Fig. 36. Note that there was no performance penalty at low C, for the dual-flow (where the benefit was not needed) and the investigated C, range extended to 0.5, a very high value for CC. (Recall that the prior dual flow experiment on the LSB airfoil had used a much higher percentage of second slot flow, 25%, with an accompanying decline in lift.) The comparison of actual to expected performance for the hydrofoil, shown in Fig. 37, shows excellent agreement. For the three-dimensional foil the average value of C, vs C, is about 50% of that seen in the corresponding twodimensional airfoil data. This is the same ratio of three- to two-dimensional performance as found for the C, vs AOA on conventional wings of the same aspect ratio as compared to an airfoil. Also, similar to the circular wing discussed earlier, the induced drag performance of the hydrofoil matched predictions based on conventional lifting line theory. One of the major test objectives was to determine where the minimum pressure occurs on the model and what the impact of subsequent cavitation would be on the ability of the jet to induce circulatory lift, or even to remain attached. Cavitation occurs when the minimum pressure reaches the value corresponding to the vaporization of water, about 0.5 psia depending on temperature. is the term for the absolute value of the pressure The cavitation index, sigma (a), coefficient that will result in vaporization and is a function of the test section static and dynamic pressures.
INVESTIGATIONS OF CC TECHNOLOGY AT NSWCCD
97
3.5 AOA = 0o w/ 2nd -slot
3
airfoil data
2.5
CL
2
CD
1.5
wing data
prediction
prediction
CD
1
drag pred.
0.5 0
G3
0
0.05
0.1
0.15
0.2
0.25
0.3
Slot Momentum Momentum Coefficient Coefficient (total), (total), Cµ Cy
Fig. 37 Comparison of actual and expected performance.16
The data plotted in Fig. 38 show that after the onset of cavitation, the lift continued to increase in response to increasing duct pressure. Eventually the lift began to roll over, but not abruptly. At no time did the Coanda jet detach prematurely from the trailing edge due to cavitation. For this particular model 2 AOA = 0° single slot
σ = 13.5
1.5
10.2 6.6
CL
1 onset of cavitation on Coanda surface for σ = 6.6
0.5
sigma = 13.5 sigma = 10.2 sigma = 6.6
σ = Cpmin at which water would vaporize 0 0
0.05
Cµ CP
0.1
0.15
Fig. 38 Lift response to Coanda surface cavitation developrnent.I6
98
R. IMBER, E. ROGERS, AND J. ABRAMSON
Fig. 39 Cavitation induced by decreasing tunnel static pressure at moderate lift coefficient.I6
design, cavitation initiated on the nozzle lip face; see the sketch in Fig. 39. The photograph in Fig. 39 shows some interesting flow visualization on the CCHydrofoil, compliments of the cavitation that caused the white bubbles of vaporized water. Cavitation is not likely to occur operationally, but if it does, it is not catastrophic to the fundamental CC effect. Another advantage of the dual slots is the ability to vector the jet thrust. In fact, in static conditions, as representative of extremely low speed operations, the direction of jet thrust can be vectored essentially through a full 360deg because the two jets merge to form a free planar jet. The photographs in Fig. 40 show qualitatively the results obtained when sequencing through a range of pressure differentials between the upper and lower slots. The tests were conducted in air, and air pressure was used in the model. The wall jet vector directions are visualized with yam tufts. In the photograph at the top left, only the lower slot is active, and the ejected air follows the curved trailing edge and vectors out the leading edge of the wing. In the top center photograph, a small amount of upper surface slot flow is introduced, which results in lifting the wall jet off the wing surface. In the top right image, the upper flow is increased to produce a vertical thrust vector effect from the planar jet formed by the merger of the two slot flows. The bottom set of images represent a continued increase in the upper slot pressure until it equals the lower pressure
INVESTIGATIONS OF CC TECHNOLOGY AT NSWCCD
99
single slot blowing (lower) 180 deg redirectionof the wall jet
Fig. 40 Model checkout in air and dual-slot operation with no freestream, 0.2 psig. The two wall jets merge to form a steerable planar jet.16
(bottom right photograph) and the jet flow direction is now 180 deg from the direction shown in the top left photograph. Verification and quantitative data for this thrust vectoring capability were measured in water using a load cell and revealed a thrust efficiency of 70-80%. Among the many findings from the CC-Hydrofoil test, the investigation demonstrated that 1) cavitation has a benign effect on the Coanda wall jet and there is no performance detriment with the onset of cavitation; 2) wake velocity profile filling is viable with dual slots; 3) a low flow rate from the second slot can eliminate one form of the CC lift limit; and 4) dual slots permit 0 to 360 deg static thrust vectoring and this merged-dual-jet mode may be viable as a jet flap for lift augmentation at extremely low speeds, where the coefficient of momentum would be too high for a viable CC mode.
VIII. Status of Design Capability To date, design implementation of CC technology has been based on the historical CC airfoil database and potential flow solutions (two-, three-dimensional) where local increment in lift is specified directly as an empirical relationshi between slot momentum flux and two-dimensional lift augmentation. R Whereas potential-flow-based techniques can readily address the “what-if” of a proposed CC application, they cannot guide the “how-to” in terms of trailing edge design details and the subsequent mass flow requirements, nor can they identify the performance boundaries. To support future CC applications, there exists a need for viscous-based computational codes to guide the subtle design details of the Coanda trailing edge region, as well as the upstream contour. Until a CFD code is available that has been properly and thoroughly validated for sensitivity to CC contour changes
100
R. IMBER, E. ROGERS, AND J. ABRAMSON
(airfoil data exists for this), design details will have to continue to be based on the historical database and intuition. Reliance on past practices as the only guide will result in a level of conservative design that may fail to uncover the full performance potential of lift augmentation by active flow control. Another benefit of future CFD will be the ability to develop additional insight into the fluid dynamics of circulation augmentation. For that insight to be valid, the challenge is to ensure that the numerically modeled flow physics is correct and not just fortuitous in producing correlation with a given set of experimental data. At NSWCCD, and other organizations, there are ongoing CFD CC validation efforts.
IX. Conclusions There have been many diverse experimental CC investigations at NSWCCD since 1986, the time period reviewed in this summary. Each of the projects built on lessons learned from previous experiments dating back to the 1970s for fixed and rotary wing, air and water applications. The experimental results increased insight into the fundamentals of CC and were used to correlate computational codes for the evaluation of various proposed applications. The six exploratory investigations revealed many new findings and, although similar in many ways, all were unique in basic geometry. Of the four dual-slotted models, one was a two-dimensional foil, two were three-dimensional foils (one with tangential mass ejection, the other with perpendicular ejection), and the fourth was an annular wing (duct). The dual slots provided either increased control range, extension of lift limit, or increased maneuvering steerage compared with a single-slot configuration, or, as in the case of the miniature oscillating valve, unsteady load reduction. Two of the investigations took place in water and four in air; however, each would be viable in either fluid regime. Of relevance to hydrodynamic applications, there had been the question of what would happen if cavitation occurred in the Coanda-slot region, which is where minimum pressure occurs. In a special series of experiments, it was revealed that the onset of cavitation did not have a disruptive effect on performance. Lift continued to be augmented in response to increased slot flow with an eventual smooth roll-off in lift as the cavitation became more extensive. Cavitation resulting from incorporating CC is not foreseen as an issue for presently contemplated hydrodynamic applications. Four of the investigations were low aspect ratio geometries and these helped to extend the understanding of the viability of CC on short-span surfaces. The successful omnidirectional demonstrations of the CC-Disc suggest future application to very maneuverable low AR vehicles or to appendages. It is helpful to know that analysis of the low AR investigations determined that there are no basic effects unique to wing lift developed by means of the Coanda form of CC, as compared to the classical lift development approaches. The TIPJET was the only rotary device reviewed and is unique in that it is driven in the rotary mode by the same air source that provides the blade lift augmentation. The findings from the TIPJET hover test were significant in furthering rotary-wing CC knowledge and in demonstrating a novel application of CC.
INVESTIGATIONS OF CC TECHNOLOGY AT NSWCCD
101
In conclusion, the six diverse exploratory applications presented all met basic performance expectations, suggesting a certain degree of maturity of this technology. These experimental results, plus those reported by other organizations, now make for a rather extensive body of knowledge on the subject, ready to encourage additional creative applications and to support a variety of full-scale applications. Complementing this broad range of empirical knowledge, the anticipated emergence of thoroughly validated viscous-based two- and three-dimensional computational codes for CC will contribute to the achievement of the full performance potential of active flow control by efficiently allowing design refinement of how the circulation control effect is implemented.
References ‘Nielsen, J. N. (ed.), Proceedings of the Circulation-Control Workshop 1986, NASA Ames Research Center, NASA/CP-2432, Feb. 1986. 2 Englar, R. J., and Applegate, C. A., “Circulation Control-A Bibliography of DTNSRDC Research and Selected Outside References (Jan. 1969-Dec. 1983),” DTNSRDC-84/052, Sept. 1984. 31mber, R. I., “Exploratory Investigations of Circulation Control Technology: Overview for Period 1987-2003 at NSWCCD,” NASAICP-2005-213509, Proceedings of the 2004 NASAIONR Circulation Control Workshop, compiled by G. S. Jones and R. D. Joslin, March 2005. 4Abramson, J. S., and Rogers, E. O., “Design of a Circulation Control Airfoil Having Both Upper and Lower Surface Trailing Edge Slots (Model LSB17),” DTNSRDC/TM16-86/03, Sept. 1986. ’Abramson, J., “Characteristics of a Cambered Circulation Control Airfoil Having Both Upper and Lower Surface Trailing Edge Slots,’’NSWCCD-50-TR-2004/030, April 2004. 6Reader, K. R., Abramson, J. S., Schwartz, A. W., and Biggers, J. C., “Tipjet VTOL UAV Summary: Volume 1-1 200-Pound Tipjet VTOL Unmanned Aerial Vehicle,” DTRC/AD-89/01, Jan. 1989. ’Schwartz, A., and Rogers, E., “Hover Evaluation of an Integrated Pneumatic Lift/ Reaction-Drive Rotor System,” 30th Aerospace Sciences Meeting and Exhibit, AIAA Paper 92-0630, Jan. 1992. ‘Schwartz, A. W., and Rogers, E. O., “TIPJET VLAR UAV: Technology Development Status,” Presented at the 20th Annual Symposium and Exhibit of the Association for Unmanned Vehicle Systems, June 1993. 'Waiters, R. E., and Ashworth, J. C., “Experimental Investigation of a Circulation Controlled Shrouded Propeller,” West Virginia Univ., Morgantown, WV, TR-39, Feb. 1974. “Hoerner, S. F., and Borst, H. V. (ed.), Fluid-Dynamic Lift, Hoerner Fluid Dynamics, Vancouver, WA, 1985. “Rogers, E. O., and Abramson, J., “Selected Topics Related to Operational Applications of Circulation Control,” NASA/CP-2005-213509, Proceedings of the 2004 NASAIONR Circulation Control Workshop, compiled by G. S. Jones and R. D. Joslin, March 2005. ‘’Imber, R., and Rogers, E., “Investigation of a Circular Planform Wing with Tangential Fluid Ejection,” 34th Aerospace and Sciences Meeting and Exhibit, AIAA 96-0558, Jan. 1996. 13Fry,D. J., and McGuigan, S., “Hydrofoil Circulation Control Via a Miniature Valve for Alternating Flows Between Two Exit Slots,’’ CDNSWCISHD-1401-02, Dec. 1993.
102 14
R. IMBER, E. ROGERS, AND J. ABRAMSON
Fry, D. J., Louie, L. L., and Jessup, S. D., “A Water Tunnel Evaluation of a Novel Actuator and Active Control System to Cancel Unsteady Foil Forces,” CDNSWC/ SHD-1401-04, Dec. 1993. ”Louie, L., Fry, D. J., and Jessup, S. D., “An Active Control System to Cancel Unsteady Foil Forces,” DE-Vol. 75, Active Control of Vibration and Noise, American Society of Mechanical Engineers, New York, 1994. 16Rogers, E. O., and Donnelly, M. J., “Characteristics of a Dual-Slotted Circulation Control Wing of Low Aspect Ratio Intended for Naval Hydrodynamic Applications,” 42nd Aerospace Sciences Meeting, AIAA 2004-1244, Jan. 2004.
1I.A. Experiments and Applications: Fundamental Flow Physics
Chapter 4
Measurement and Analysis of Circulation Control Airfoils F. Kevin Owen* Complere Inc., Pacific Grove, California and
Andrew K. Owent University of Oxford, Oxford, England, United Kingdom
Nomenclature c = airfoil chord CL = lift coefficient CL, = infinite aspect ratio lift coefficient C, = blowing momentum coefficient U = mean axial velocity U, = edge velocity u' = rms velocity fluctuations x = streamwise position y = crosswise position a = airfoil angle of attack aeff= airfoil effective angle of attack ai= induced flow angularity
*Consultant. 'Research Assistant. Department of Engineering Science. Copyright 0 2005 by the authors. Published by the American Institute of Aeronautics and Astronautics, Inc., with permission.
105
106
F. K. OWEN AND A. K. OWEN
I. Introduction IRCULATION control (CC) airfoil concepts have been studied extensively for more than four decades. These studies have included low-speed airfoil, helicopter rotor, and flight demonstrator configuration^.'-^ In these, and other studies of CC, the sharp trailing edges of otherwise conventional airfoils are replaced with rounded or bluff surfaces, typically with either circular or elliptic cross-sections, with thin tangential blowing slots located on the aft upper surface. These rounded trailing edges allow the rear stagnation point to move. This movement is controlled by the relative blowing momentum of fluid injected through the slots, and by the properties of the external flow field. By blowing through the slot, a jet sheet is issued, which, as a result of the balance of centrifugal force and subambient static pressure within the jet, remains attached to the airfoil. At low blowing rates, this Coanda effect entrains upper surface boundary layer flow and prevents trailing edge separation. As the blowing momentum is increased, the rear stagnation point is moved further around the trailing edge and the wake deflection angle is increased. An effective camber is introduced, and the lift is increased. Blowing rates can be adjusted until the airfoil static pressure distribution is that predicted by inviscid potential flow. With increased blowing, the jet controls the location of the airfoil stagnation points, and therefore the circulation and lift. However, eventually there comes a point where there is no longer a balance between the static pressure and centrifugal force and jet blow-off occurs, with a corresponding dramatic decrease in lift. Lift values greater than those predicted by inviscid potential flow theory are generated in the CC regime. Pneumatic camber similar to a mechanical high-lift system can be obtained. However, CC lift augmentation is far more efficient than conventional high-lift devices, because they only have to overcome the viscous losses in the flow. By compensating for the viscous losses, the flow field more closely resembles the ideal inviscid case. Accordingly, lift augmentation several times that attainable with jet flap or blown devices has been achieved. Unfortunately, the precise determination of CC airfoil performance for design and computational fluid dynamics (CFD) assessment purposes is difficult to achieve. The most serious problem encountered in testing these high-lift devices is the interference produced by wind tunnel test section wall separation. Owing to the strong adverse pressure gradients on the airfoil upper surface, strong secondary flows can be generated in the sidewall boundary layers. The problem is further compounded by significant spanwise circulation gradients, because circulation must decrease toward the wall. Even at moderate lift, these factors can generate trailing edge vorticity more characteristic of a three-dimensional than an infinite span wing. A great deal of research and analysis is still required in order to properly establish a reliable database for full-scale model development and CFD code validation. To address this shortfall, a wind tunnel investigation has been conducted of a two-dimensional CC airfoil section equipped with trailing edge blowing. The tests were conducted in the NASA Ames 2 x 2 ft Variable Density Transonic Wind Tunnel over a range of freestream Mach number and unit Reynolds numbers. Detailed nonintrusive flowfield measurements of the mean flow
C
MEASUREMENT AND ANALYSIS OF CC AIRFOILS
107
and turbulent properties were obtained in the airfoil wake for a number of different blowing coefficients. In this paper, some of these results have been related to the CC airfoil performance obtained from direct surface pressure measurements. The analysis shows that wind tunnel wall interference can have significant influence on high-lift test results. This influence must be accounted for before wind tunnel test data can be used for design extrapolation or for turbulence modeling and CFD assessments. Corrections have been made for finite aspect ratio (AR) wind tunnel wall interference in order to provide interference-free benchmark data for turbulence modeling and CFD code development and validation. 11. Experimental Details
The work described in this report was conducted in the NASA Ames 2 x 2 ft Variable Density Transonic Wind Tunnel at a freestream Mach number of 0.5 and at a unit Reynolds number of 3.2 x 106/ft. The test model spanned the test section and was held at zero angle of attack for the present work. The model was a symmetric 6-in. chord airfoil, 20% thick, 3% camber ellipse with a nominally circular arc trailing edge. An adjustable, nominally 0.010-in. tangential blowing slot was located on the upper surface, 1-2% before the usual upper surface separation point, at the 96% chordwise location. Transition strips were attached to the airfoil section at the 17% chord on both the upper and lower surfaces. The 1.25-mm-wide strips consisted of 0.13-mm nominal diameter glass beads. Transition effectiveness was verified by the sublimation technique. A regulated 3000psig air system was utilized to supply the internal plenum of the model, and a maximum internal pressure of 60 psig was attainable. The resulting high internal contraction ratio ensured adequate two-dimensionality of the jet exit flow. The jet exit velocity was calculated from isentropic relationships referenced to tunnel static conditions. There were a total of 91 pressure taps on the model, 59 of which were positioned along the centerline. Of these taps, 24 were on the upper surface and 35 on the lower surface. The airfoil performance data were obtained by direct integration of these centerline pressure taps.4 The flow-field measurements were obtained using a two-component laser velocimeter with conditional sampling ~apability.~ The effective sensing volume approximated a cylinder with a 200 diameter and 3 mm length, with its axis aligned with the cross-stream direction. Detailed measurements of the mean axial and vertical velocities, turbulent intensities, and turbulent shear stress distributions were obtained.
111. Sample Results Examples of laser velocimeter wake measurements at 5% chord downstream of the trailing edge for a zero angle of attack airfoil case are shown in Figs. 1 and 2. These results show the effects of jet blowing on the near-wake axial velocity profiles. In the zero blowing case, there is a small wake displacement due to the airfoil camber that produces lift at zero angle of attack. There is also a large region of reversed flow typical of a blunt body recirculation zone. With a small amount of blowing (C, = 0.024), there is a significant downward wake
F. K. OWEN AND A. K. OWEN
108
0.1 Wake Location, Y/C
,. . >
0.05
U
0 −0.2 0 −0.05
−0.4
s
B
0.2
0.4
0.6
0.8
1
1.2
−0.1 −0.15 Wake Velocity,U/U U/Uee
Fig. 1 Wake velocity profile (C, = 0, x/c = 0.05).
displacement and a fuller profile wake that has been energized by the jet. These data, and other profiles obtained at various distances downstream of the airfoil trailing edge, can be used to determine wake deflection angles, wake deficit recovery, and mixing and shear layer growth downstream. It is evident that the Coanda effect of the flowing jet relocates the aft stagnation point upstream along the airfoil lower surface, resulting in a downward displacement of the wake. This trend of increased wake displacement continues with higher blowing rates until there is no longer a balance between local static pressure difference and centrifugal force required for continued jet attachment. At this point we get what is referred to as jet blow-off. Jet effectiveness is destroyed, and there is a rapid drop in wake displacement and in the measured lift coefficient. However, in the present case, Fig. 3 shows that blowing can produce effective angles of attack, determined from measured wake displacement, of almost 20 deg in the present 0-deg airfoil angle of attack (AOA) case before blow-off occurs. From these data we can calculate the infinite AR lift coefficients from inviscid potential flow theory and thus assess airfoil lift performance. These results are shown in Fig. 4, where the lift at zero blowing agrees well with zero offset
Wake Location, Y/C
Y >
0.3 0.25 0.2 0.15 0.1 0.05 0 −0.05 0 −0.1 −0.15
0.2
0.4
0.6
0.8
1
Wake Velocity, U/Ue UlUe
Fig. 2 Wake velocity profile (C, = 0.24, x/c = 0.05).
1.2
MEASUREMENT AND ANALYSIS OF CC AIRFOILS
109
01
0
0.005
0.01
0.015
0.02
0.025
0.03 0.035
Blowing Momentum Coefficient, C,
Fig. 3 Measured wake angles.
camber predictions. However, these results are significantly higher than the lift computed from the measured airfoil surface pressure distributions. However, as expected, we have seen seed particle deposits on the test section windows, which suggest that strong secondary flows are generated in the wind tunnel sidewall boundary layers. This shed vorticity will induce unknown flow angularity in the freestream flow ahead of the model, thus changing the airfoil’s effective AOA. However, from the wake measurements, we are able to calculate the induced flow angularity as a function of jet blowing momentum coefficient. These results, calculated assuming a semi-elliptic lift distribution, are shown in Fig. 5 . With this information, we are able to compute the finite AR lift coefficients that are shown in Fig. 6 . These results are in excellent agreement with the surface pressure, direct lift measurements shown in Fig. 7. This comparison shows that sidewall effects are indeed significant, because agreement is not reached until an induced freestream downwash for a fully three-dimensional wing is introduced, that is, CL = 2m,f
2.5
01
0
0.005
0.01
0.015
0.02
0.025
Blowing Momentum Coefficient, C,
Fig. 4 Infinite AR lift coefficients.
0.03 0.035
F. K. OWEN AND A. K. OWEN
110
0
0.005
0.01
0.015
0.02
0.025 Blowing Momentum Coefficient, C,
0.03
0.035
Fig. 5 Induced flow angularity.
1.2
,
I
01
0
0.005
0.01
0.015
0.02
0.025
0.03
0.035
Blowing Momentum Coefficient, C,
Fig. 6 Calculated finite AR lift coefficients.
01
0
0.005
0.01
0.015 0.02 0.025 0.03
Blowing Momentum Coefficient, C,
Fig. 7 Measured lift coefficients.
0.035
MEASUREMENT AND ANALYSIS OF CC AIRFOILS
111
0.1-
40
-nit -".
I
I
18.
50
WakeTurbulence Level, u'/Ue%
Fig. 8 Wake turbulence profile (C, = 0, x/c = 0.17).
where
Wake turbulence measurements indicate that large-scale fluctuations are introduced by jet blowing and that wake unsteadiness may well be present at the higher blowing rates just before jet detachment. In the no blowing case shown in Fig. 8, small-scale turbulence dominates, and local RMS turbulence intensities are related to the local mean velocity gradients as in a planemixing layer. Thus, using the measured local turbulence levels and the measured local mean velocity gradients, we can calculate the effective mixing length for this flow. There is good agreement between this calculated mixing length to wake width ratio of 0.2 compared to the nominal value of 0.18 for a plane-mixing layer. However, once jet blowing is initiated, as shown in Fig. 9, a wide highly turbulent core develops that is indicative of high turbulent kinetic energy production in the blown jet wake. Turbulent length scales are increased by a factor of three, an indication of large-scale turbulent mixing andor wake unsteadiness.
0.1
2
$
-1 t
I
0.05-
.Q
-
Ll ~1
Y
i!
0 0 -0.05-
10
20
-0.1",
I"
WakeTurbulence Level, u'/Ue%
Fig. 9 Wake turbulence profile (C, = 0.009, x/c = 0.17).
3
112
F. K. OWEN AND A. K. OWEN
IV. Conclusions New CC test measurements and analysis have been presented that show the need for caution when attempting to use wind tunnel test results for CFD code validation, or for design purposes. In particular, the results have identified the quantitative extent wall influence can have on CC test results; for example, lift augmentation reduced from 68 to 42. The results also suggest that turbulence models must be modified to account for the effects of unsteady, large-scale turbulent mixing. The agreement between the measured and the calculated finite AR lift coefficients suggests that if we know the effective angle of attack, then simple inviscid theory may well be adequate for lift coefficient predictions. In turn, the analysis suggests that two-dimensional CFD computations could well be meaningless unless the airfoil effective angle of attack is known. Full three-dimensional calculations will probably be required to account for wall interference; that is, effective angle of attack and effective camber, especially at high lift. Estimates of the errors caused by non-uniform flow due primarily to wall boundary layer separation are essential. These initial investigations suggest that angle of attack corrections of at least - 1.5 CL will be required. Clearly, this can be a substantial correction factor, because lift coefficients well in excess of 2.0 can be expected for high-lift systems. Effects on the estimated drag coefficient are even more acute. Typical drag coefficients show errors of over 100% at induced angles of less than 1 deg. Indeed, at lift slopes typical of those at transonic speeds, angle of attack errors of 0.01 deg can lead to drag measurement uncertainty of more than one drag count. Clearly, in any high-lift experiments, accurate estimates or measurements of induced flow angularity must be made before useful design estimates or meaningful comparisons with CFD calculations are undertaken. A detailed review and analysis of finite AR CC experiments must be conducted to assess wind tunnel wall effects on experimental data previously reported in the literature. Although induced flow angularity is a fundamental consequence of the flow around finite AR lifting wings, our experiments and calculations show that these problems could be ameliorated to some extent by testing higher AR wings, and by measuring the induced flow angularity upstream. References ‘Kind, R. J., and Maull, D. J., “An Experimental Investigation of a Low Speed Circulation Controlled Airfoil,” The Aeronautical Quarterly, Vol. 19, 1968, pp. 170- 182. ’Cheeseman, I. C., and Seed, A. R., “The Application of Circulation Control by Blowing to Helicopter Rotors,” Journal of the Royal Aeronautical Society, Vol. 71, 1966, pp. 451-464. 3Englar, R. J., “Development of the A-6 Circulation Control Wing Flight Demonstrator Configuration,” DTNSRDC Rept. ASED-79/01, Jan. 1979. 4Wood, N. J., and Conlon, J. A., “The Performance of a Circulation Control Airfoil at Transonic Speeds,” AIAA Paper 83-0083, Jan. 1983. 50wen, F. K., “Application of Laser Velocimetry to Unsteady Flows in Large Scale High Speed Wind Tunnels,” International Congress on Instrumentation in Aerospace Simulation Facilities, Inst. of Electrical and Electronics Engineers Publ. 83CH1954-7, September 1983.
Chapter 5
Some Circulation and Separation Control Experiments Dino Cerchie,* Eran Halfon,+ Andreas Hammerich,* Gengxin Han,s Lutz Taubert,* Lucie-Trouve? Priyank Varghese,* and Israel Wygnanski** University of Arizona, Tucson, Arizona
Nomenclature c = chord length
C , = drag coefficient ( D / q c) CDp= form drag coefficient ( s ( p - p,) dy/q c) 2 1/2 C , = integrated force coefficient (Ct CO,) C, = lift coefficient ( L / q c) CMac= moment coefficient about the aerodynamic center C, = pressure coefficient ( p - p , ) / q CQ = steady volume flow coefficient (Q/SU,) C, = steady momentum coefficient [(2 h / c ) ( U s l o t / U,)’] (c,) = oscillatory momentum coefficient [(h/C)(USlotMax/ d = reference length, diameter f = frequency of excitation F+ = nondimensional frequency (f d l U,) h = slot height J =jet momentum q = dynamic pressure ( J p U k )
+
u,)~I
*Research Associate. ’Research Assistant. Currently at Tel-Aviv University, Ramat-Aviv, Israel. *Research Assistant. gPostdoctoral Fellow. TResearch Assistant. Currently at L’Ecole Nationale Supdrieure de Mdcanique et d’Adrotechnique, Poitiers, France. **Professor. Copyright 0 2005 by the American Institute of Aeronautics and Astronautics, Inc. All rights reserved.
113
114
D. CERCHIE ET AL.
Q = volume flow through the slot Ree = Reynolds number (U,8/u) U , = freestream velocity UJ = slot velocity UJ = maximum slot velocity a = angle of attack or slot location on a circular cylinder Sf = flap deflection 8 = angular distance from the leading edge of a cylinder
I. Introduction THIN jet being emitted tangentially from a slot milled in a circular cylinder or other convex, highly curved surface, alters its direction and wraps itself around the surface. A circular cylinder can turn a jet around and alter its direction by more than 180 deg. The centrifugal force acting on the deflected jet is balanced by the pressure difference between the surface of the cylinder and the ambient fluid. Integrating this pressure results in a force that is approximately equal to twice the jet momentum emitted at the slot (Fig. 1). Blunting a trailing edge of an airfoil and blowing over its upper surface will deflect the fluid downward, changing the “Kutta condition,” and provide a powerful means of increasing the usable lift. This is loosely referred to as supercirculation. One may divert the flow around a blunt trailing edge by using suction, as it was aptly demonstrated by Prandtl,’ who removed the boundary layer from one side of a circular cylinder and attached the flow on the side of the suction slot and generated lift. This idea
A
m
B
Fig. 1 Streamlines representing a wall jet flowing around a cylinder.
CIRCULATION AND SEPARATION CONTROL EXPERIMENTS
115
was applied by Schrenk to thick airfoils that were otherwise plagued by early separation.2 Steady suction has been characterized historically by a volume flow coefficient CQ, because its primary aim was to remove low-momentum fluid from the boundary layer of a given freestream. Excessive suction could also provide circulation control (CC), which implied an increase of lift above and beyond the expected value generated by incidence and camber. The use of conformal mapping correctly predicted the lift generated by a strong, slot s ~ c t i o n , ~ which was directly proportional to the sink strength associated with the suction and depended on the location of the slot on the airfoil. The suction contribution to lift is given by ACL = 2 c Q cot(c$/2), where 4, in this case, represents the location of the slot in the mapped “circle plane”. The drag penalty associated with suction is very large (ACD= 2cQ), and it was theoretically predicted and experimentally verified by this model. Slot suction for the purpose of lift enhancement (CC) did not withstand the test of time because of the associated drag increase and the large ducts that were required to remove the low-pressure, external fluid. As the thickness of airfoils diminished with the quest to increase speed, they could not accommodate large internal ducting. Nevertheless, surface suction and multiple slot suction is still considered to be useful for drag reduction and for delaying transition to turbulence. The integration of propulsion with lift generation is a long-sought dream advocated by many researcher^.^ The advent of jet propulsion seemed to offer such an opportunity, but it quickly became apparent that materials withstanding the heat were too heavy and too costly for aeronautical applications. In most instances (the application to MIG-21 is an exception), only the compressed air generated prior to combustion by turbojet engines was ducted to slots and blown over flaps to augment their lift. A number of production aircraft used this form of lift augmentation (e.g., Lockheed F104 Starfighter, Blackburn NA39 Buccaneer, Dassault Etandard-IVM). In the application of blowing, a distinction is made between boundary layer control (BLC) and circulation control (CC). The first function of the jet, as it blows over the surface, is to increase the mean kinetic energy of the fluid within the boundary layer so that the latter may advance without separation into a region of rising pressure, for example, over the upper surface of a highly deflected trailing-edge flap. An adequate jet momentum is expected to generate a lift coefficient that is approximately predicted by a potential flow solution. In this regime of boundary layer control, the lift increment is roughly proportional to the first power of the jet momentum (ACL oc C,). An increase of jet momentum augments the lift further, but this augmentation is only proportional to the square root of the jet momentum (ACL oc JC,). This is the regime of supercirculation, where the jet departs from the trailing edge with sufficient downward momentum to increase appreciably the circulation around the wing. Poisson-Quinton4 is credited with establishing these criteria, as well as the critical value of (C,),,, that empirically determined the momentum required to pass from one flow regime to the other over an airfoil with a deflected flap at arbitrary angle 8, Circulation control may also be obtained by blowing the jet obliquely from the trailing edge of the wing, as was done on pure “jet-flap” experiments; however, there
116
D. CERCHIE ET AL.
C, =0.24 Calculated using row of sinks
Fig. 2 Calculated and measured streamlines around a cylinder.
is a substantial gain in lift when the jet is blown over a suitably designed solid flap.4 A number of theoretical methods have been developed for predicting the ACL resulting from supercirculation. Stratford attempted to calculate the lift by assuming that the “jet-flap” was equivalent to a physical flap.5 More realistic assumptions were made by Helmbold,6 S p e n ~ eLegendre,* ,~ and Woods,’ who replaced the jet by a vortex sheet originating at the trailing edge. Woods used the hodograph method, whereas Spence7 and Malavard” linearized the problem, assuming small incidence and small jet deflection. In all the theoretical models, the mixing of the jet with the ambient flow is neglected. In reality, the jet entrains fluid from its surroundings and that entrainment is well represented by placing a suitable distribution of sinks along its path” (Fig. 2). When a strong jet flows over a curved flap or the upper surface of an airfoil, this distribution of sinks contributes to circulation,’2 which is also proportional to JC,. When the jet is emitted from the trailing edge of bluff bodies (e.g. circular or elliptic cylinders), the entrainment that takes place on both sides of the jet contributes to form drag.” Some aspects of the ideal flow models are controversial and they have not been entirely resolved to date, for example, the prediction that the entire jet momentum should be recovered as thrust regardless of the jet’s initial inclination angle relative to the oncoming stream. This result was proven experimentally up to a flap deflection of 60 deg, at which approximately 90% of the jet momentum was recovered as thrust as long as the value of C, was quite large. At larger flap deflections, the C, required to overcome separation and other “real flow” effects (mixing) became excessive, and the thrust recovery almost entirely vanished when the flap deflection exceeded 90 deg. The effects of steady blowing, steady suction, or periodic excitation on circulation and drag are assessed presently. This report represents an ongoing research with the purpose of improving our understanding of each technique and to sorting out the leading parameters that affect, control, and manipulate the flow. We shall start by examining the flow over a flapped, conventional, symmetrical airfoil, the NACA 0015 (Fig. 3a). The Kutta condition is fixed and the impact of the increased circulation is easily recognized when compared to the standard airfoil performance. Thereafter, we have replaced the normal, 26% chord simple flap with a stubby, 8% chord flap consisting of a circular cylinder that blends into a wedge having an included angle of 40 deg at its
CIRCULATION AND SEPARATION CONTROL EXPERIMENTS
117
a)
b) Stream
Fig. 3 Airfoil models used.
trailing edge. This configuration was extensively studied at S t a n f ~ r d in ' ~ conjunction with strong steady blowing. The circular trailing edge facilitates the generation of supercirculation, but the trailing edge wedge predetermines the Kutta condition provided the flow over the flap is attached (Fig. 3b). The flow over the small, blunt, and concave trailing edge brought into focus the need to investigate the controlled flow over a concave surface in the presence of adverse pressure gradient more extensively. Such flows were investigated over wall-mounted humps, started by S t r a t f ~ r d , 'who ~ coined the concept of a boundary layer that is maintained on the verge of separation over an extensive distance. When periodic excitation was applied to such a boundary layer,15 the skin friction was increased while the shape factor was reduced, and it thinned and stabilized the boundary layer and enabled it to better overcome the imposed pressure gradient. If the pressure recovery region at the rear of the hump is made steeper, the boundary layer separates, but it has to reattach farther downstream due to the presence of the long flat surface that extends beyond the trailing edge of the hump. The control of this flow is reduced to control of a separation b ~ b b 1 e . l ~Because ~ ' ~ the hump used in Refs. 16 and 17 is based on Glauert's GLAS I1 airfoil (GLAS stands for Glauert's Laminar Airfoil Section), it was investigated in the present context (Fig. 3c). The flow around a thick elliptical cylinder was later examined. Its maximum thicknessto-chord ratio is 30%, and its leading and trailing edges are circular. This geometry easily lends itself to a change in the actuation location and in the slot width. The pressure gradient near the leading edge resembles the pressure gradient experienced by a standard airfoil, whereas the flow near the trailing edge is complicated by the fact that the Kutta condition is not well defined. The circular cylinder was the last test article to be examined, because it is the most widely researched flow, but it might be the most difficult one to control. The Kutta condition is not determined and the parameters affecting flow reattachment interact and affect the circulation in a more complex manner than on previous configurations due to the strong coupling between the flows near the leading and trailing edges. It is believed that by increasing the complexity and the number of degrees of freedom that are associated with the various configurations selected, the dominant variables controlling the flow will be identified. Typical questions to be answered include the following:
118
D. CERCHIE ET AL.
1) What is the best method or a combination of methods to increase lift? 2) Is C, the unique parameter that governs BLC and CC control and are they occurring sequentially as C, is increased beyond a prescribed threshold level ( CJcrit?
3) Is separation effectively controlled by suction? 4) Is C, displaced by CQ when suction is used for LBC or CC? 5) When and why is periodic excitation (active flow control, AFC) more effective than blowing or suction? 6) How sensitive is each method to the location of the actuation, and how is it affected by the configuration on which it is employed? The present chapter focuses on some of these questions, in an attempt to categorize the effects of the leading parameters in a rational manner.
11. Discussion of Results
A. Flow Control over an Airfoil with a Conventional Flap Most aerodynamic control of lift experiments begin with a standard NACA airfoil and then either progress in the direction of more custom lofting, lift augmentation devices or flow control to achieve not only the desired loads, but more favorable distribution of the load along the airfoil surface. We will discuss the impact of the total load and distribution of the load on a standard airfoil using both a trailing edge flap and flow control. Data were collected using a NACA 0015 airfoil with a simple 26% chord flap at Re < 5 x lo5. A schematic drawing is included in Fig. 3a, showing a cross-section through the airfoil model. Some early observations carried out by Greenblatt and Wygnanski indicate that the flow over a deflected flap at Sf= 20 deg separates around a = -2 deg." Even at a = 0 deg, both steady blowing and periodic excitation are beneficial. Consider injection of momentum at C, = 3% (Fig. 4). For a flap deflection of Sf = 20 deg, both steady blowing and periodic excitation at very low frequency generate a lift increment of ACL = 0.5 relative to the baseline airfoil performance, whereas periodic excitation at F+ = 1.1 generated an inferior lift increment of only ACL = 0.35. Repeating the same experiment at a lower C, of 1.2% shows slightly lesser periodic excitation performance at F+ > 1.1, and even poorer steady blowing performance (see Fig. 5 for Sf= 20 deg). At Sf= 35 deg and at the high C, of 3%, both steady blowing and low frequency excitation (F+ = 0.3) peaked out by generating ACL = 0.4 and 0.52 relative to the baseline flapped airfoil, respectively. An increase in the flap deflection beyond this angle caused a reduction in the lift increment generated by steady blowing until, at Sf> 50 deg, the injection of steady momentum became detrimental to the generation of lift (i.e., the baseline CL exceeded the value obtained by using steady blowing). The efficacy of the low-frequency periodic excitation at C, = 3% did not deteriorate with increasing flap deflection beyond Sf= 35 deg, whereas the excitation at the higher frequency of F+ = 1.1 improved with increasing flap deflection until the two curves crossed over around Sf= 65 deg. At the lower level of C, = 1.2%, the increase in flap deflection beyond Sf= 35 deg rendered the steady blowing
CIRCULATION AND SEPARATION CONTROL EXPERIMENTS
119
1.61.41.2-
Re = 300K -#-Baseline AFC = 3%, F+= 0.3
1.o-
I*
+AFC=3%,F -#- Blowing C = 3%
0.8-
+=1.1
P
0.6 !
I
20
I
I
60
40 Flap deflection Sf(')
Fig. 4 Effect of blowing and AFC at C, = 3%, as a function of flap deflection on a NACA 0015 at a = 0 deg.
ineffective, if not detrimental, whereas even higher frequency excitation remained effective (Fig. 5 ) . The pressure distribution associated with the three modes of flow control at C, = 3% and Sf= 35 deg is plotted in Fig. 6 . The constant, low pressure on the upper surface of the flap ( x / c > 0.75) indicates that the baseline flow was
'L
1.81 1.6
1
1.4 121.o -
Re = 300K f
AFC = 1.2%, F+= 1.1
P
"." .
+
+AFC = 1.2%, F = 2.5
0.8 -
--C Blowing C I
20
I
40 Flap deflection
P
= 1.2% I
4(")
60
Fig. 5 Effect of blowing and AFC at C, = 1.2%,as a function of flap deflection on a NACA 0015 at a = 0 deg.
D. CERCHIE ET AL.
120 C
P
-3
-2
Re = 300K α = 0°, δ f = 35° Baseline + AFC = 3%, F = 0.3 + AFC = 3%, F = 1.1 Blowing C µ = 3%
-1
0 0.0
0.5 X/C
1.0
1
Fig. 6 Pressure distribution over NACA 0015 with 26% chord deflected flap.
totally separated over the deflected flap. The flow was partially attached by the periodic excitation at F+ = 1.1 and completely attached by the low-frequency excitation at F+ = 0.3 and by the steady blowing. The reattachment of the flow over the flap changes the circulation around the airfoil and has a far-reaching effect on the upstream pressure distribution all the way to the leading edge of the airfoil. The acceleration of the flow upstream of the slot is of particular interest in this case. It seems reasonable to examine various flow control mechanisms providing identical circulation and C ., This approach is of practical interest, because a potential designer may be required to generate a prescribed lift by various techniques available and should be familiar with the consequences, such as drag, pitching moment, momentum input, and so on, associated with generating the required lift. During the experiments discussed here, the flap was deflected at two angles; Sf= 20 and 40 deg, with periodic excitation being applied through a 0.06 in. slot at the interface of the main element and the flap shoulder. Figure 7 includes two pairs of angle of attack (AOA) sweeps with and without AFC (periodic excitation) at the two different flap deflections discussed previously. Some features are immediately apparent in this figure. All of the configurations share the same dCL/da when a > 0 deg. The deflection of the flap on the model increases the effective camber of the model, even if the flow over the flap is separated, causing a shift upward (or to the left) of the C, vs a curves. However, when the flow over the flap separates (see curve corresponding to Sf= 40 deg that uses AFC in Fig. 7),there is a shift to the right with a dCL/ d a % 0 in the range -4 < a < -2 deg. This effect is not seen as clearly for the baseline airfoil sweep with Sf = 20 deg because of the low Re of the experiment, although the flow separates partially from the flap a x - 2 deg. The other two curves, plotted in Fig. 7, are not expected to have a discontinuity in dCL/da. The baseline flow over the flap that is deflected at Sf = 40 deg is separated over the entire range of a, considered, and the periodically excited flow at Sf = 20 deg is attached over the flap until a = astall. The excitation level for Sf = 20 deg (F+ = 0.9, (c,) = 2.2%) was specially selected in order to overlap the lift
CIRCULATION AND SEPARATION CONTROL EXPERIMENTS
121
CL 2
1
+
Re = 200K F = 0.9 δ f = 20° Baseline δ f = 20° = 2.2% δ f = 40° Baseline δ f = 40° = 2.2%
0 -10
0
α (°)
10
20
Fig. 7 NACA 0015 airfoil performance with AFC.
curve generated when the flap deflection of the basic airfoil was Sf = 40 deg. This enables a detailed comparison to be made between the effect of flap deflection and periodic excitation. In fact, the curve for Sf= 20 deg with AFC falls on top of the curve for S,= 40 deg without AFC until the occurrence of stall. Although the stall angle is somewhat higher for Sf= 20 deg in conjunction with AFC, the resulting CLma,is approximately the same for both cases. When the same AFC is applied while S,= 40 deg, the maximum lift coefficient, C,, = 2.25 at a = 10 deg. At negative angles of attack (-8 < a < -4 deg), the ACL generated by the application of AFC to S f = 40 deg is commensurate with the ACL observed at 20 deg flap deflection for a < astall, because the flow over the flap is attached for both 8, values in the respective range of a. Inspite of the flow separation from the upper surface of the deflected flap at S,= 40 deg, the airfoil continues to generate a higher lift than for Sf= 20 deg, primarily due to the deflection of the flow by the lower surface. In the discussion that follows, we examine pressure distributions measured on the surface of the airfoil model, which produced three different lift coefficients (CL= 1.0, 1.35, 1.5). These “sectional” cuts through the (CL vs a ) curves show the different approaches that a designer could select to produce a specific lift and are marked in Fig. 7 to aid the reader. We consider this to be an important technique to evaluate different flow control strategies, rather than simply look at the relative benefit in performance that the control can provide at a fixed geometric configuration. Figure 8 shows four pressure distributions that generate c, = 1.0. In the absence of AFC and with the flap deflected to 40 deg, a slight pitchdown attitude ( a = -2 deg) generates a small suction peak near the leading edge (LE) and mild adverse pressure gradient along the upper surface of the entire main element. The flow over the flap is undoubtedly separated and, consequently, there is a drag penalty associated with this configuration. When Sf= 20 deg for the baseline airfoil, the incidence must increase to a = + 2 deg in order to
D. CERCHIE ET AL.
122
Cp
-2
Re = 200K F+ = 0.9 δ f = 20° α = 2° Baseline δ f = 20° α = -2° = 2.2% δ f = 40° α = -2° Baseline δ f = 40° α = -8° = 2.2%
-1
0 0.0
0.5
1.0
X/C 1
Fig. 8 Pressure distribution that develops at C, = 1.0 for the NACA 0015.
maintain the same lift, and a larger suction peak is created at the LE, while separation on the flap is pushed slightly farther downstream. When the appropriate level of AFC is applied while maintaining Sf = 20 deg, the AOA can once again be returned to a = - 2 deg, resulting in a more uniform pressure distribution over the upper loft and reducing the suction peak near the LE. Because a is the same for the two flap deflections as the total circulation, the flow near the LE is identical over the upper surface, as is the pressure distribution in the range 0 < x / c < 0.4 for the two cases considered (Sf= 40 deg and Sf= 20 deg with AFC). At x/c > 0.5 and in the absence of AFC, the pressure remains constant over the upper surface of the airfoil and the deflected flap, because it is dominated by the “base pressure” (C p RZ - 0.7) of the recirculating region downstream. On the other hand, when AFC is applied and the flap is only deflected at 20 deg, the flow accelerates upstream of the slot (i.e., for 0.6 < x/c < 0.74). The flow over the flap is fully attached with a pressure coefficient at the trailing edge being positive ( C p x 0.25), suggesting that the flow downstream of the trailing edge continues with its downward momentum, generating perhaps a “jet flap” effect. Increasing the flap deflection to 40 deg while maintaining the AFC results in an attached flow with C p x 0 at the trailing edge (TE), while heavily loading the aft region of the flapped airfoil. This is achieved while maintaining a favorable pressure gradient over the entire upper surface of the main element by placing the airfoil at a = - 8 deg. In this case the flow acceleration upstream of the slot is magnified. This behavior would be especially advantageous for laminar flow applications where delay of transition is important. It is easy to identify the upstream influence of the AFC along the upper surface of the main element. Figure 9 shows the pressure distributions over the model that generated CL = 1.35. The two cases that share the same angle of attack and lift (a= 2 deg, S - 20 deg, C, = 2.2% and Sf= 40 deg, C, = O%), indicate that fthe pressure distributions on both the upper and lower surfaces over the upstream half of the airfoil are almost identical. The case with AFC shows the flow over the
CIRCULATION AND SEPARATION CONTROL EXPERIMENTS C p -4 CD
123
Re = 200K F+ = 0.9 δ f = 20° α = 8° Baseline
-3
δ = 20° f δ = 40° f δ = 40° f
-2
α = 2° < C > = 2.2% µ α = 2° Baseline α = -4° < C > = 2.2% µ
-1 0 0.0 1
0.5
1.0
X/C
2
Fig. 9 Pressure distributions that develop C, = 1.35 for the NACA 0015.
flap is attached with a pressure coefficient close to zero at the TE. The heavily deflected flap in the absence of AFC has the same C p % -0.7 at the TE as it did at CL = 1.0, indicating that in both cases the recirculating wake region has approximately the same dimension. It implies that the flow over the flap was completely separated at both angles of incidence and that the additional lift was generated by the main element. Once again, the flow with AFC accelerates before reaching the slot. This reduces the adverse pressure gradient on the upper surface, making the airfoil less susceptible to stall at this value of CL.The basic airfoil with the flap deflected at 20 deg also generates CL = 1.35, but at a = 8 deg. Under these conditions, the flow is still separated over the flap, but the wake is narrow as evidenced by Cp % - 0.2 at the TE. One may reattach the flow to a deflected flap at 40 deg through AFC enabling CL = 1.35 at a = -4 deg, due to the increased suction on the aft portion of the main element upstream of the slot (i.e., at 0.4 < x / c < 0.74). In conclusion, the results described in Fig. 9 are similar to those associated with C L = 1, but with the effect of AFC being accentuated. Therefore, not only does the AFC prevent flow separation downstream of the actuation location, thereby increasing the circulation, but it also lowers pressure on the upper surface upstream of it, enhancing the lift. The prime benefit is in the form-drag reduction, which was reduced by a factor of four in the range 0.5 < CL < 1. The total drag was reduced only by a factor of two and there is some uncertainty in the drag estimate. Nevertheless, the effect of AFC is significant and will be discussed in full in Section D, p. 144. The pressure distributions at a given CL,with AFC being applied, suggest that a can be significantly reduced depending on the deflection of the flap needed to produce the same lift coefficient. For the lower flap deflection case, the suction peak is reduced by approximately 40% and the flow over the flap is fully attached with a TE pressure coefficient close to that of the freestream. At the higher flap
124
D. CERCHIE ET AL.
deflection, the suction peak is further attenuated when sufficient lift is generated, even at zero AOA. Although the flow over the flap may not be fully attached, the upstream effect of the AFC is strong enough to load the main element sufficiently to generate the necessary lift. This behavior is not unique to this airfoil. One can conclude from the data presented that AFC contributes through three distinct mechanisms to airfoil performance: first by preventing flow separation on a deflected flap (this mechanism was investigated by Nishri and Wygnanski” and Darabi and Wygnanski2’); secondly, by enhancing circulation through the inviscid jet flap effect; and, thirdly, associated with turbulent entrainment of the flow and the reduction of the static pressure both upstream and downstream of the slot location. Poisson-Quinton, in Ref. 4, identified the first two mechanisms using steady blowing. Acceleration due to entrainment, while being present in those cases, is more prominent when oscillatory excitation is used. The demarcation between (BLC) and (CC) was quite well defined when steady blowing or suction were used to control the flow, because CC implied “an artificial increase in circulation over that which could be expected from incidence and camber in unseparated flow”.21 Because the lift over streamlined bodies (over which the flow is totally attached) is predicted by inviscid solution, a comparison of pressure distribution both measured and calculated was essential. The pressure distributions calculated from viscous and inviscid solutions using the “Xfoil” program, assuming that the flow is entirely turbulent in the viscous case, are plotted in Fig. 10, together with the measured results with and without the use of AFC. When Sf = 20 deg, a = 2 deg and, in the absence of actuation, the experimental results agree quite well with Xfoil’s viscous prediction, with the exception of the base pressure observed over the separated flap. The experimental data for the forced flow suggest that the flow over the flap was attached as a result of the excitation and, as a consequence, the pressure over the entire upper surface was reduced (Fig. 10). The measured pressure distribution resulting from excitation at (CY) = 2.2% at F+ = 0.9 fell short of the expected inviscid values, suggesting that this level of excitation is below the (CJcrit that separates the BLC and the CC regimes. Similar conclusions may be drawn for the results obtained for Sf= 40 deg and a = -4 deg, except that the ideal flow solution overpredicts the forced results by a larger amount, and the viscous solution does as poorly in predicting the base-flow pressure distribution. Both examples (Fig. 10) confirm the suggestion that periodic excitation, at the level and frequency used, keeps the flow attached (i.e., controls separation), but does not enhance the circulation above the normal inviscid limit. A complimentary example where the enhancement of circulation was achieved without reattaching the flow over the flap was provided by H. Nagib (personal communication, 2003) who examined the control of the flow over a three-element airfoil with a slot located at the shoulder of a highly deflected, simple flap. In this case, periodic excitation affected the pressure distribution on the main element and on the leading edge slat without causing reattachment to the flap itself (Fig. 11). Circulation was increased and the pressure upstream of the actuation was lowered, in spite of the fact that the pressure over most of the flap remained unchanged. Because the upstream effect of AFC may be more significant than the downstream effect, it is possible that the demarcation
CIRCULATION AND SEPARATION CONTROL EXPERIMENTS
125
NACA 0015, a=2', &m=200 CP
Fig. 10 A comparison between pressure distributions calculated using viscous and inviscid solutions and the measured pressures on the basic airfoil and after using AFC (all data taken on the NACA 0015).
between BLC and CC is artificial and that the effect of AFC on circulation occurs concomitantly with BLC.
B. Control over a Truncated NACA 0015 Flapped Airfoil Typical CC airfoils have circular trailing edges in order to make use of the Coanda effect and generate maximum circulation. The addition of a cusp provides the ability to control the angle at which the jet departs from the trailing edge and control the circulation without altering the blowing emanating from the slot. The truncated NACA 0015 airfoil shown in Fig. 3 was widely used in the verification of jet-flap concepts.13 In the absence of strong blowing, the airfoil does not perform very well. Its maximum lift coefficient is, CL % 1.2 in
D. CERCHIE ET AL.
126
Rec = 0.75e6 Flap = 40 deg. alpha = 13 deg.
ADVINT/ATT 5% Model in NDF at IIT Nagib & Kiedaisch; 2002
Baseline, slat 2 F = 120 Hz, Uj/Uinf = 2.8
Cp CP Slot Location for AFC
-0.2
-0.1
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
x/c
Fig. 11 Pressure distributions over a three-element airfoil with and without CC (H. Nagib, private communication).
the absence of flap deflection (Fig. 12), but the rounded trailing edge generates a large form drag (Fig. 13) and a wider wake than is generally expected from the NACA 0015 at a given Re. In fact, by deflecting the flap to 15 deg, the C ,,, as well as the C, attained at small angles of incidence, is slightly reduced, but this reduction does not affect the form drag or even the total drag. Strong blowing approaching C, % 1 has been used in previous experiments for CC.13 In the present investigation C, 5 0.1, in order to use the upper limit of C, = 0.1 for comparison with data acquired by Hynes.13 For C, = lo%, CLmaxis increased to 1.5 in the absence of flap deflection, and it attained C , = 2.5 for S,= 60 deg (Fig. 12). In the absence of blowing, the minimum form drag is attained at incidence 4 < a < 6 deg, regardless of the flap deflection (Figs. 12 and 13). The total drag was determined from wake surveys and corrected for buoyancy (both Betz's and Jones's corrections were used; however, there was no difference between the two methods of correction). It behaves in a similar manner to CDp provided 8, > 30 deg (Fig. 14). It is interesting to note that the total drag is always lower than the CDpand the difference between the two increases with increasing SF Because the skin friction drag is generally positive, there has been a search for experimental error and uncertainty. It is possible that the number of pressure taps near the leading edge is insufficient, but it is equally plausible that vortices shed from a lifting airfoil over which the flow is partially separated (either due to high incidence or large Sf) increase CDp,making CDp> C ,. This excess is more noticeable whenever AFC is used. The constant presence of vortices downstream of
CIRCULATION AND SEPARATION CONTROL EXPERIMENTS
127
C, vs. a,Baseline &Steady Blowing
Fig. 12 C L - a curves on the truncated NACA 0015, for C, = 0% and C, = 10%.
a bluff body induces low “base pressure” near the base of such a body and contributes to form drag. This possibility should be examined more closely in the future, particularly near bluff bodies where the skin friction drag is negligible relative to CDp. We shall now focus on the effect of increasing C, on the characteristics of the airfoil when Sf= 30 deg, at a constant representative C, = 1. In the absence of blowing, CL = 1 is attained at a x 7 deg, but at C, = 0.1 it is achieved at a x -0.7 deg (Fig. 12). There is a coupling between a and the C, necessary to provide the required lift. This relationship is not linear (Fig. 15a), although it is explored in the region where dCL/da is constant (Fig. 12). The highest effect on reducing the incidence required to generate the necessary lift corresponds to 0.025 < C, < 0.075. The moment coefficient about the quarter chord location (the aerodynamic center) behaves in a similar manner, implying
D. CERCHIE ET AL.
128 C, vs. C,,,
Baseline, Uh+=12m/s, Re=2.P105
Fig. 13 Form drag polars for C, = 0% and C, = 10%.
that a desired pitching moment can be obtained at a prescribed lift by trading incidence with C, (Fig. 15a). It is interesting to note how the form drag increases with increasing C,, (Fig. 13), whereas the total drag turns to thrust with increasing C, (Figs. 14 and 15b). The increase in C D p is attributed primarily to the low pressure generated on the convex surface downstream of the flap shoulder due to the Coanda effect. The concave surface resulting from the presence of the cusp generates positive pressure, but the surface is too small to affect the CDp in a meaningful way (Fig. 16). The increase in incidence necessary to generate the proper CL at lower values of C, also results in an increased suction at the LE and a reduction in C D p . One may now examine the lift increment generated by increasing C, while maintaining a and afconstant (Fig. 17). It is interesting to note that at low levels of C,, ACL cc C;, and only at higher C,, it becomes linearly dependent on this parameter. This is contrary to the accepted n o t i ~ n ~ ” ’ ~ ’ ’ ~
CIRCULATION AND SEPARATION CONTROL EXPERIMENTS
129
CL vs. CD,Baseline, Ui,=12mls, Re=2.7*105
Fig. 14 Drag polars for C, = 0% at four flap deflections and for 8f = 30 deg, but for O
that in the BLC regime (i.e., at small C,) ACL cc C,, and it only becomes proportional to JC, in the CC regime. The difference may stem from the low values of C, that are presently considered. The critical C, distinguishing between the two regimes is 3.5 < C, < 5.5% in the range of flap deflections investigated. Suction at equivalent C, = 10% generates lower lift than blowing, although the stall angle increases somewhat by using suction. A comparison between the two methods is shown in Fig. 18 for S f = 30 deg. Whereas, for blowing, most of the added momentum coefficient is manifested as thrust (the CD of the baseline airfoil is 0.04 while with C, = 10% the CD x -0.06), for suction CD x 0.02, implying that some 80% of the suction momentum generated drag. It is interesting to note that, based only on CDp,suction generates apparent thrust, whereas blowing increases the drag. The explanation for this can come from comparing the pressure distributions corresponding to C, = 1 (Fig. 19). In one case there is a negative pressure peak over the deflected flap, whereas in the other it occurs near the leading edge.
D. CERCHIE ET AL.
130
a& Pikhing M o m e n t a C14 vs. Cp Variation of a & Pitching Moment for fixed CL=l FIap=3Oo
a) 9 8 7 6
..............
5 a 4
3 2 1 0
-1 -2
..............
.
J
.
.
I
4
0
6
> 4.5 10
8
c, [",'.I CDp & CrJ vs. C, @ C p l for various C,, %hap=300
b)
0.15
...... ...... ...... ................. ......A
........... , .......,......7 ............
0.1J 0
1
2
3
4
5
. . . I.
6
+CDp ..; .. - +- CD __.I
- - - -
.....:.
c-
......,......,
7
8
9
i 10
c, [%I Fig. 15 Dependence of a) (Y and CM,, and b) CD and CDpon C, at CL = 1.
C. Controlling the GLAS I1 Airfoil The Glauert Laminar Airfoil Section I1 (GLAS 11) has a maximum thicknessto-chord ratio of 31.4% and was designed to operate with massive suction through a slot located at 69% of the chord. The designer intended to have favorable pressure gradients over most of the upper and lower lofts, and maintain
CIRCULATION AND SEPARATION CONTROL EXPERIMENTS
131
C, profiles for different C,,,CL-l, qlSp=3O0
Fig. 16 Pressure distributions for various values of C, at C, = 1.
laminar flow over a large fraction of the airfoil surface. Suction provided a pressure discontinuity across the slot that led to a positive pressure along the entire concave recovery ram terminating with stagnation pressure at the trailing edge. With adequate suction!’ the measured L I D varied between 250 and 550 for C, > 1 and Re x lo6. In the absence of suction, L I D > 12 for the same Re, but C, was reduced to C, x 0.6. The L I D increased to approximately 30 at C,Sweep, Steady Blowing for different Flap Angles
Fig. 17 Dependence of CL on C, at (Y = 3 deg.
D. CERCHIE ET AL.
132
CL vs. a for various C,
CL vs. Cm and C,,
Fig. 18 Comparison between the performance of the airfoil using strong suction or blowing at C , = 10%. C,
- Comparison between Steady Suction & Blowing fixed C,-I ,C,=lO%,~,,,=3O0
Fig. 19 Representative pressure distributions on the truncated 0015 airfoil using suction at C , = 10%.
CIRCULATION AND SEPARATION CONTROL EXPERIMENTS
133
Re % 3 x lo6. It appeared that the flow was intermittently reattaching to the ramp just upstream of the TE, resulting in large drag oscillations. Blowing appeared to be less effective than suction, requiring larger mass flux to forcibly reattach the flow. Either way, the C, required to keep the flow attached was in excess of 20%. Many additional research papers followed, culminating in test flights on a glider. If comparable performance could be achieved using periodic excitation at considerably smaller C, levels, the use of GLAS-type airfoils could be revived. Furthermore, there has been a considerable interest in a hump that was placed on the wind-tunnel wall and whose shape represents the upper loft of a GLAS I1 airfoil for validation of CFD codes. The differences between the flow over such a hump and over the airfoil should be fully understood and properly documented. The dependence of CLon a is plotted in Fig. 20 for the baseline configuration at Re < 0.5 x lo6. dCL/da is not constant at these low Reynolds numbers, but the discontinuity in the slope diminishes with increasing Re. At Re = 1.17 x lo5 there is a sudden increase in the lift at a % 20 deg. With increasing Re, the kinks in the CL-a curves occur at lower incidence (e.g., a % 16 deg at Re % 1.7 x lo5) and they become more moderate. Also, the maximum lift experienced by the airfoil decreases from being CL,, = 1.7 at Re=1.17 x lo5 to C , = 1.05 at Re = 4.8 x lo5. The data acquired at the Compressed Air Wind Tunnel of the W L Z 3agrees fairly well with the present results. The CL-a curve reproduced from Ref. 23 was taken at Re = 4.06 x lo5 and could be obtained by interpolating the present data taken at 3.5 x lo5 < Re < 4.8 x lo5 (Fig. 19). No wind tunnel corrections were applied to either set of data. Figure 21 shows the typical pressure distributions measured on the surface of the baseline airfoil at 1.2 x lo5 < Re < 4.8 x lo5, corresponding to CL % 0.5, suggest that a laminar boundary layer could be maintained over the lower loft
C,
YS.
a - Baseline
Fig. 20 Dependence of C, on (Y and on Re for the baseline airfoil.
134
D. CERCHIE ET AL.
Baseline C, profiles at q 4 . 5 for different Re
Fig. 21 Pressure distribution measured on the baseline airfoil at constant C, but different values of Reynolds number.
of the airfoil up to x/c ranging between 0.5 and 0.7 due to the favorable pressure gradient existing on that surface. On the upper loft, however, the location where the C p is minimum depends strongly on Re as it changes from x/c = 0.07 at the lowest Re used presently tox/c = 0.33 at Re = 4.8 x lo5. The C p measured near the TE indicates that the flow is mostly separated in this region, although the base pressure increased with increasing Re suggesting that the mean size of the separated region was reduced. Applying the strongest available suction (C, x 19%), blowing (C, x 22%), or periodic excitation (at (C,) x 2.1%) to this airfoil at x/c = 0.62 and Re x 1.2 x lo5 results in a tremendous increase in lift and the straightening of the CL-a curves (Fig. 22). The actuation location was moved upstream because the levels of actuation mentioned above were unable to affect the flow at small angles of incidence at the original slot location suggested in the literature. The Australian researchers faced similar difficulties and they, too, moved the suction location. The chosen location of the slot corresponds to the separation line predicted by CFD and it will be discussed separately. Contrary to the observations of Glauert et a1.,22 blowing was more effective than suction, approximately doubling the CL attained at a < 10 deg. The maximum CL was obtained in both instances at a,,, = 24 deg; however, in the case of blowing, CL,, = 5, whereas by using suction, CLm,, = 3.5 only. Periodic excitation at F+ = 0.7 and approximately 1/10 of the steady momentum input for blowing performed almost as well as the steady blowing up to a x 14 deg. At higher incidence, the difference in C, between the steady blowing and the periodic excitation became noticeable, yielding CLm,, = 3.5 for the periodic excitation. Because the C,, obtained for the basic airfoil was only 1.7, suction and periodic excitation represent 100% increase in CLm,,, whereas the much stronger steady blowing represents almost 200% increase in this coefficient.
CIRCULATION AND SEPARATION CONTROL EXPERIMENTS
135
Fig. 22 Dependence of C, on a and on the control parameters used and the corresponding drag polars for Re = 1.17 X lo5.
The drag polars for the corresponding cases are also plotted in Fig. 22. They reveal very interesting features associated with each method of boundary layer and circulation control. The baseline drag at CL < 0 is approximately C , % 0.1 1. This number agrees very well with the National Physics Laboratory (NPL) results at comparable Re and C, values.23 The baseline pressure drag CDp= 0.086 was almost constant for all C, < 0, and it implies that the skin friction drag is approximately 0.024 for the negative lift coefficients. The application of suction at a < 6 deg generated a small dCL/da and it possessed relatively large C ,. For C, > 0.4, suction reduced the drag to C, x 0.02. This represents a substantial drag reduction relative to the baseline airfoil that attains C, x 0.4 around a x 20 deg. The form drag in this case is negative, indicating that there is a low pressure at the leading edge of the airfoil and a high pressure on the rear ramp. Strong, steady blowing generated thrust in addition
136
D. CERCHIE ET AL.
to lift enhancement (Fig. 22). The maximum thrust measured at CL % 1.4 is C, = -0.27, and it exceeds the total momentum input of C, % 22.6%. According to these results, blowing at high levels of C, is much superior to suction. It is well known that there is a drag penalty associated with the removal of boundary layer flow through the airfoil’s surface. According to Poisson-Quinton4this drag is equal to CDsink= 2 c Q = 0.0312 and, were it not for this penalty suction, would have generated thrust as well. When massive blowing is used, the jet momentum is recovered as thrust, but in addition there is a source flow that should contribute to thrust. In this case the thrust attributed to the source CD,,,,, = - 2 c Q % 0.034. The maximum thrust recovered may be CT = C, 2 c Q % 0.26. Because the energy spent to generate suction is similar to that of blowing, the superiority of steady blowing is clear in this case. The drag associated with periodic excitation is larger than either steady blowing or suction, but because the (C,) spent in this case is but a small fraction in comparison to the steady cases, the comparison is inappropriate. Repeating the same experiment at Re = 2.35 x lo5 reduces the respective C values by a factor of 4. In the case of periodic excitation, it also affected F+Y lowering it to F+ = 0.35. The baseline CL-a curve is now much more normal, particularly for a > 10 deg (Fig. 23), reaching a CL,, of 1.45. The CLma,generated by steady blowing dropped from CLmax= 5 to 3.1, whereas suction attained a CLmaXof 2.5 only. The efficacy of periodic excitation was further reduced as a result of the concomitant change in F+ yielding a CLm, of 2.25, in spite of the fact that for a < 14 deg it generates a higher lift than the steady suction does at a C, that is an order of magnitude larger. The drag polar of the basic airfoil suggests that the flow may be attached to the TE ramp at a = 18 deg corresponding to CL = 1.2 (Fig. 23). For this C,, the L I D of the basic airfoil is 15. Steady suction at C, = 4.7% generated identical total drag at CL = 1.2. There is a major difference in the form drag of these two cases. Whereas the CDpgenerated by the suction is so small and perhaps even negative, the CDpassociated with the basic airfoil is larger than the total drag (CDp= 0.113, whereas C , = 0.085). Active flow control (AFC) generates a higher total drag at this C, (C, = 0.1 l), whereas steady blowing reduces the C, to 0.015. In fact, at CL = 0.8 the drag associated with the steady blowing at C, = 5.5% vanishes, implying that a wake generated by this C, resembles a wake of a self-propelled, two-dimensional body. The actual C, required to propel an aircraft at this CL is higher because of the added induced drag. In contrast to the results accumulated on the NACA 0015 and its truncated derivative, CDpis usually smaller than C , except near the stall angle; it is possible that the concave curvature affects the result as well. The pressure distributionsover the airfoil at C, = 1 and CL = 2.1 and Re values of 1.17 x lo5 and 2.35 x lo5 are plotted in Fig. 24 for the three control mechanisms used. Because the same CL was obtained at different incidence angles as a result of differences in C, and the specific method of control used, a comparison of pressures near the LE is inappropriate. Nevertheless, the efficacy of the control scheme reflects on the incidence angle for which a given CL is achieved. Consider the case for C, = 2.1 and Re = 1.17 x lo5 (Fig. 24b). It appears that the flow is accelerating toward the slot for all three of the control schemes used. Suction, however, provides the least maximum acceleration, but its effect is felt
+
CIRCULATION AND SEPARATION CONTROL EXPERIMENTS
137
Uinf=I4mls,Re=2.35*1 O5
Fig. 23 The variation of C, with a and the corresponding drag polars at Re = 2.35 X lo5 and the various control parameters used.
farther upstream from x / c = 0.5 up to the slot located at x / c = 0.62. At the slot itself, suction brought the flow to stagnation (C, x l), whereas on the reminder of the ramp (0.65 < x / c < l), an almost constant, slightly negative base pressure is maintained (e.g., C= ., -0.2). Active flow control accelerates the flow upstream of the slot, but it also maintains a good pressure recovery downstream, culminating with C, = +0.2 near the TE. Strong blowing (at C, that is an order of magnitude larger than the AFC applied) provides a favorable pressure gradient just upstream of the slot and a very-low-pressure bubble immediately downstream. The bubble is generated by the oblique injection angle of the jet,
Next Page
P 0
rn
n
0
Fig. 24 Pressure distributions measured at CL= 1 and 2.1 for two values of Re and various control mechanisms.
Previous Page CIRCULATION AND SEPARATION CONTROL EXPERIMENTS
139
because the slot is inclined at 30 deg to the downstream surface. After reattachment the C p is still negative because the curvature of the surface is convex; however, at x/c % 0.68 the curvature of the surface changes sign and also the sign of the Cp, which becomes positive downstream of this chordwise location. The large positive pressure on the TE ramp contributes to the thrust generated by the steady blowing (Fig. 22). The measured C p can be larger than unity because the high-speed jet that emanates from the slot has a larger total pressure than the freestream (Cp,,, = 5). To attain CL = 2.1 at Re = 2.35 x lo5 requires a much larger incidence than for the lower Re, partly because of the lower C, available. In this case the same C, is obtained for AFC and for steady suction (in spite of the large difference in the level of actuation) at a = 18 deg, enabling a direct comparison between the pressure distributions upstream and downstream of the slot. It is clear (Fig. 24d) that steady suction is more effective in accelerating the flow upstream of the slot than AFC is; however, the latter is more effective in the pressure recovery region on the ramp. The maximum C p due to steady blowing measured around x/c = 0.7 was reduced from C p = +5 to C p = +1.25, because the total pressure of the freestream was increased by a factor of 4. It suggests that the regularly defined pressure coefficient so widely used on normal airfoils is not adequate in the case of blowing. The LE edge radius of curvature of the GLAS I1 airfoil is very small, pointing to a potential problem with this design. There is a large separation bubble on the lower surface of the airfoil that reduces greatly the favorable pressure gradient on this surface. This is most obvious when the flow over the ramp is separated (Figs. 24b, 24c, and 24d). Laminar flow on this airfoil is probably achieved on the upper surface around CL % 1 by a proper combination of incidence, BLC and Re. The application of AFC at low values of C, < 2% is considered in Fig. 25 with particular emphasis being placed on C, < 1% at incidence a = 0 and 6 deg. At a = 0 deg there is an increase in lift of ACL = 0.4 at C, < 0.4%, whereas at a = 6 deg the same increment in lift requires 0.4 < C? < 0.9% depending on the reduced frequency F+ used. The largest increase in CL for the smallest input in C, corresponds to a reduced frequency of 0.7 < F+ < 1. For F+ = 1.8 the sudden increase in CL requires a lower input of momentum, but the ACL is somewhat smaller. The pressure distributions taken in the region of transition from the “low to high” CL suggest that even a partial attachment of the flow downstream of the slot changes the circulation affecting the entire pressure distribution on the upper surface of the airfoil. Active flow control enables the boundary layer to overcome a very severe adverse pressure gradient existing on the convex part of the ramp (i.e., 0.6 < x/c < 0.7), even if it does not succeed in attaching the flow all the way to the TE (Figs. 26 and 27). The data presented in Fig. 26 correspond to C, = 0.25 and 0.75% for a = 0 and 6 deg, respectively. It is noted that the LE stagnation point occurs on the upper surface around x/c % 0.01 for a = 0 deg, and almost precisely at the LE for a = 6 deg. The existence of a bubble near the LE of the lower surface is observed. The acceleration of the flow over the leading 50% of the chord on the upper surface is followed by a smooth deceleration towards the slot and towards the inflection point on the surface of the TE ramp, provided the C , is close to the threshold value at
140
D. CERCHIE ET AL.
Effect of F’ and C, on C, at GO*,
Ui,pl4ml~
Effect of F’ and C, on C, at a=$,Ui,,=14mls
Fig. 25 Increase of C, with C, for a variety of F’ at two values of a.
which CL increases (Fig. 26). A further increase in C, results in a “spike” in the observed C pjust upstream of the slot and in a higher pressure recovery over the ramp (Fig. 27). These results seem to contradict the previous concept that required a threshold value of CPcritto overcome separation before circulation can be increased. In this case the control of separation (downstream effect) and the increase in circulation (upstream effect) occurred simultaneously. The effects of AFC on drag are shown in Fig. 28 for a = 6 deg, Re = 235 x lo5 and for three values of F+ ranging from 0.36 to 1.8. For F+ > 1, the drag is reduced to approximately one-third of its value in the absence of excitation, whereas for F+ = 0.36, the reduction is only two-thirds. Although the high-frequency periodic excitation reduced the total drag at all
CIRCULATION AND SEPARATION CONTROL EXPERIMENTS
141
C,-0.26%, GOO, U,,,+4mls
Fig. 26 Effects of F + on the pressure distributionfor C, = 0.25% and 0.75% at two values of a.
amplitudes (C, > 0), the excitation at F+ = 0.36 increases the total drag for 0 < C, < 0.6%. The highest frequency of excitation experimented with to date, F+ = 1.8, reduced the drag at the lowest amplitudes imposed. The response of CDpto the imposed excitation is quite different. First, CDpwas not increased by excitation at F+ = 0.36; secondly, this frequency was able to lower the CDpas effectively as F+ = 1.8; and thirdly, higher CDpresulted from excitation at F+ % 1. The present results are in general agreement with the parametric
142
D. CERCHIE ET AL.
F'=0.36,a=0°, Uim=14mls
F'=0.36,a=6', U i e l4mls
Fig. 27 Effects of C, on the pressure distribution for two values of F + and a.
study of Nishri and Wygnanskilg on the reattachment of flow to a generic flap by using AFC. The high CDpassociated with excitation at F+ x 1 results possibly from the strongest eddies that are consistently present over the ramp and that are generated by this frequency. Excitation at lower frequencies results in stronger eddies that are not always present over the surface, because their wavelength exceeds the length of the surface, whereas excitation at F+ >> 1 generates weaker eddies, so their constant presence has a lesser effect on the flow. Nishri and colleagues also observed that to maintain the flow attached requires a much smaller momentum input than to force a separated flow to reattach. In other words, a hysteresis should be present whenever the level of actuation
CIRCULATION AND SEPARATION CONTROL EXPERIMENTS a=6', Pand C, effect on CD, CD,,
143
at Ui,,,=14mls
Fig. 28 Effects of C, on the drag for three values of F + at a = 6 deg.
is increased until the flow reattaches or when the fluctuation level is decreased until the flow separates. Such a hysteresis was observed and is plotted in Fig. 29. A comparison among all three methods of control discussed is shown in Fig. 30 for a prescribed incidence of 6 deg. Active flow control results in a sudden increase in ACL % 0.6 when (C,) % 0.5%; suction requires a C, = 1.5% to obtain the same result, and blowing requires a higher threshold. Seifert et al. made similar observations on the NACA 0015 airfoil.24 At larger values of C,, blowing may surpass AFC, but the practical implications of this are questionable in view of the large momentum required. a=6*,FreqrSSOHz, F'11.8, U,,,=14mls
Fig. 29 Hysteresis effect caused by an increase or a decrease in C,.
D. CERCHIE ET AL.
144
a=6', C, sweep at U,,,=l4m/s CL
Fig. 30 Effects of suction, blowing, and AFC on lift at small values of momentum input.
D. Flow Around an Elliptical Airfoil The elliptical cylinder represents an aerodynamic body that has no definite Kutta condition at its trailing edge, but it generates pressure distributions that are similar to an airfoil at an angle of attack. The modified 30% ellipse has both leading and trailing edge cylinders that have adjustable slot widths and exhaust locations (measured as included angles either from the LE or TE; (Fig. 3). The focus in the present paper is only on fluidic actuation emanating from the TE cylinder. Because of the undefined Kutta condition, the elliptical airfoil has the adjustable circulation characteristics of a cylinder when flow control is introduced. Steady suction, steady blowing, and oscillatory excitation have been tested using the model. The effectiveness of the three flow control categories at a = 0 deg has been quantified (Fig. 31). The slot location c$ was varied around the TE of the model in order to generate the plotted data. The three flow types were run at a blowing coefficient of 1.9%. The oscillatory flow produced the best lift results, followed by suction and then blowing. The slot was inclined at 25 deg to the local surface to enhance mixing with the boundary layer. This may provide the reason for the poor performance of the blowing relative to the other control methods. If the exhaust angle had been tangential, blowing might have extended the attached flow region farther around the TE and increased the lift. The angle where the applied flow control provided the best lift moved toward the TE until the control was insufficient to counter the adverse pressure gradient upstream of the slot. As the momentum of the applied flow control was increased, the location maximizing the lift moved towards the TE, regardless of the method of control used. Only C, values generated by AFC are shown in Fig. 31, indicating that an increase in (C,) from 1.9 to 2.38% increases both C, and by at least 5 deg, to 130 deg.
,,+,
CIRCULATION AND SEPARATION CONTROL EXPERIMENTS
/>
0.41
Suction (C@= 1.go/,)
-@-
( f = 130Hz, = 1.9p)
+AFC w.w-
145
I
I
100
I
110
io
120
1
140
TE slot exhaust angle (")
Fig. 31 Comparison among three different types of flow control on a 30% thick ellipse.
The pressure distributions at C, x 0.44 give some indication where the lift is generated for the three different control types (Fig. 32). The pressure distribution over the elliptical body is nearly constant for the three cases from 15 to 85% chord. The real difference is at the leading and trailing edges. Active flow
CP
-2.0
AFC CL = 0.448 = 1.9% TE = 110°/0.030" Blowing CL = 0.436 Cµ = 1.9% TE = 110°/0.030" Suction CL = 0.441 Cµ = 1.9% TE = 120°/0.030" 110° Slot Location
-1.5 -1.0 -0.5
120° Slot Location 0.0 0.0 0.5
0.2
0.4
0.6
0.8
1.0
x /c
1.0
Fig. 32 Comparison among different flow control techniques at constant C, 30% thick ellipse.
146
D. CERCHIE ET AL.
control and steady suction produced the same increase in Cp at the LE when the control was applied at the TE. These two control approaches also produced the same pressure distribution along the lower surface near the TE. Active flow control and blowing looked similar along the upper surface near the TE where the higher velocity slot flows were entering the flow field. The lowpressure peak associated with the slot flow was not present at the TE for the suction case, as was expected. As a test note, the AFC data plotted represent the time-averaged pressure data. The data were gathered at 600 Hz, and 700 samples were gathered for each static port. Based on the number of static ports on the model, the typical data runs were just over a minute in duration. The effect of (C,) on Cp is plotted in Fig. 33. The AFC “off’ (baseline) condition is also plotted for reference. This baseline pressure distribution agrees well with the CFD results generated using NASA’s CFL3D program. The three AFC magnitudes increased the lift by increasing the velocity along the entire upper surface of the ellipse. The lower surface velocity is hardly affected by AFC. The application of AFC near the TE increased the circulation along the entire span of the airfoil, not just locally at the TE. A closer look at the Cp at the trailing edge region for various AFC magnitudes is presented in Fig. 34. The baseline condition shows symmetrically separated flow on the upper and lower surfaces from around 97% chord to the TE. Application of AFC resulted in a large, time-averaged increase in the velocity on the upper surface. The separated region shrank on the upper surface, but it was hardly affected on the lower one, until, for (C,) = 2.38%, the flow over the entire upper surface is attached.
Fig. 33 Supercirculation on a 30% thick ellipse using AFC (CFD courtesy of A. Hassan, Boeing).
CIRCULATION AND SEPARATION CONTROL EXPERIMENTS
cp-1 .o
130" Slot Location
ff=
147
0"
+-----
\
-0.6
I -0A -02 0.o
02 C
-m-
-dP>
AFC C;=0.619 = 2.38% Baseline CFL3D Re =340K Fully Turbulent
-
Fig. 34 Zoom into TE region of Fig. 33.
The effectiveness of AFC applied to the TE element of the ellipse as a function of a is plotted in Fig. 35. The data show a nearly constant benefit of AFC until a nears stall. The increase in C, is approximately 40%. Angle of attack sweeps were performed with steady suction being applied from the TE cylindrical cross-section (Fig. 36). Rather than increasing the blowing coefficient, the slot angle was altered for each of these sweeps. The intent was to assess the influence of a on the optimal slot location and compare these data to the results shown in Fig. 31 at a = 0 deg. The initial run was at C$ = 120 deg, or 5 deg from the peak performance point shown on Fig. 31. The lift benefit ACL from steady suction was nearly constant for a < 4 deg. The slot angle C$ was then reduced by 10 deg, but a,,, (i.e., a corresponding to CL,,,) increased only by 3 deg. Another 10 deg reduction in C$ only increased a,,, by 1 deg. From these data it is clear that the zero angle of attack data provide a good insight into the optimum performance location. They also shows that the upstream pressure gradient has a very big influence on the performance of control applied at the TE region. In fact, the first decrease in lift (stall) occurs for the flow separating from the TE cylinder; thereafter the lift increases again with increasing a until the flow separates from the LE of the ellipse (Fig. 36). The same test was repeated for a larger slot angle and larger suction coefficient (Fig. 37). The trends seen at the smaller slot-width were validated. The larger .,,,C , However, the a at which CLm,, was suction coefficient produced larger realized was not directly proportional to the change in the suction slot location
D. CERCHIE ET AL.
148
CL
1.6 F* = 0.35,TE = 130"/0.030", U = 1Oms-l
1
+Baseline A =l.2%
t=l.8%
1.2
0.8
0.4
0.0
1
0
5
10
15
20
a (") Fig. 35 Angle of attack sweep using AFC.
or its width. Figure 38 shows the effect of steady suction on the C, at a = 0 deg where the basic ellipse provides no lift. The lower surface velocity is decreased only slightly by the suction and almost at the same increment along the entire lower surface. The strongest effect occurs on the upper surface where the increase in velocity is constant from the LE to 90% of the chord. The change in the TE
CL
1
--W-
.2 a = 0",h = 0.030",Cp= 1.9%
Baseline
+120" 110" 100"
0.8
0.4
0.0 .d/
0
I
I
I
5
10
15
Angle of Attack a (")
Fig. 36 Suction on 30% ellipse: C, = 1.9% and h = 0.030 in., U = 10 d s .
CIRCULATION AND SEPARATION CONTROL EXPERIMENTS
149
Angle of Attack a (")
Fig. 37 Suction on 30% ellipse at C, = 3.5%.
static pressure is very evident on this plot and it is similar to the AFC data shown in Fig. 33. Comparing the pressure distribution at C, = 0.8 with and without the suction necessitates a comparison between a = 4 deg and a = 10 deg (Fig. 37), which are very differently loaded along the chord and generate different moment around the aerodynamic center. Active flow control tests at the TE were then conducted for a variety of slot angles, excitation frequencies, and amplitudes.
Fig. 38 Pressure distribution at C, = 3.5%.
D. CERCHIE ET AL.
150
ACL
Fig. 39 Scatter in data when C, used as parameter.
Data acquired at two slot locations are plotted in Fig. 39 against (CJ. The data did not collapse onto a single curve, indicating that some other parameter or parameters need to be included for this type of control and geometry. There are two broad clusters of points depending on the slot location. There is also a dependence on frequency with the lowest frequency of excitation generating the lowest ACL. Empirical correlation trials resulted in the data plotted in Fig. 40. This set of data appears to collapse all the results onto a single curve as a function of the product of (C,), the square root of F+, and the angle denoting the distance from the TE. The length scale used in the definition of F+ depends on the same angle, as it represents the length from the slot to the theoretical TE of the ellipse. In this case the data collapse onto a single curve whose generality is yet to be proven.
E. Controlled Flow Around a Circular Cylinder and Reexamination of Some Old Results The flow around the circular cylinder is discussed last because it represents the highest degree of complexity. Similarly to the ellipse it has no defined TE separation location or imposed Kutta condition. Unlike the previous four geometries discussed, the LE and the TE flows are closely coupled because of their immediate proximity. The flow is also sensitive to transition location, which is, in turn, sensitive to a variety of inflow parameters.
CIRCULATION AND SEPARATION CONTROL EXPERIMENTS
151
Fig. 40 Collapse of data based on empirically derived flow control parameter when AFC is used.
In the absence of an external stream, a wall jet created by steady blowing wraps itself around a convex surface of the cylinder, following it up to, and sometimes beyond half of its circumference (Fig. 1). In the example shown, a jet of momentum J emanating to the right from a slot located on top of the cylinder encircles it before separating to the left from its lower surface. The change in the direction of the flow generates a low-pressure region on the right-hand surface, which, when integrated, yields a side force whose magnitude is almost equal to twice the jet momentum. This force multiplier makes some applications of wall jets over curved surfaces very attractive, arousing interest in improving the understanding of this flow. One of the unique characteristics of the curved wall jet is its phenomenal rate of growth from the surface, and its high turbulence level, which is attributed to the streamwise vortices generated by a centrifugal in~tability.'~ The cylinder over which these measurements were made was carefully designed with the jet emerging tangentially to the surface after passing through a smooth contraction. The resulting flowfield is a consequence of instability. Therefore, it is important to know how sensitive this flow is to the detailed jet characteristics leaving the nozzle and to the initial width of the jet relative to the radius of the cylinder (Fig. 41). There is a large region around the surface of the cylinder where the pressure is constant in spite of the rapidly thickening jet flow. The pressure distribution is not seen to be very sensitive to the jet-related Reynolds number.
D. CERCHIE ET AL.
152
Fig. 41 Pressure distributions on a cylinder for various Reynolds numbers.
Two additional cylinders were constructed that were more suitable for wind-tunnel tests in which an external stream of variable velocity could be applied. One of the cylinders (2 in Fig. 42) was machined from two parts and has a continuous nozzle along its span. The figure shows that the nozzle design is a critical feature in determining the external flow characteristics. The best tangential design, nozzle 1, maintained attached flow at least 40-deg further around the cylinder than the other two slot designs. Nozzle 3, which had a segmented slot cut through the cylinder’s wall, demonstrated the worst performance. This poor tangential flow characteristic was previously discussed as a possible reason for the lower blowing control performance near the TE of the ellipse relative to suction and to AFC control. Using slot design 2, the pressure distributions around the cylinder, as a function of slot location relative to the freestream, were generated for a given-jet-to freestream velocity ratio of 14.5 (Fig. 43). The data possess the same trend as for the ellipse, where the maximum surface velocity continues to increase until the slot has rotated far enough from the natural baseline separation point (around 60-70 deg for that Reynolds number) that the added momentum can no longer hold the flow attached. The maximum Cp generated is -22 for an injection angle 120 deg measured from the LE of the cylinder. It is interesting to note that the region of separated flow on both sides of the cylinder is reduced as the performance of the cylinder is increased. For the smallest angle of 30 deg, the separated region extends from approximately 140 to 290 deg, whereas of extends only from 240 to 290 deg for the performance slot angle of 130 deg, which already exceeds the most effective location of 120 deg. A comparison of the pressure distributions for the three flow control approaches over a single slotted cylinder is plotted in Fig. 44. The slot location was rotated from the LE toward the TE and the data plotted
+=
CIRCULATION AND SEPARATION CONTROL EXPERIMENTS
153
Comparison of Pressure Distribution of C o d a Flow ReM=llOOO
Fig. 42 Pressure distributions on a cylinder for various slot geometries.
correspond to the best-performance location. The data are presented in a manner similar to airfoil data in percent of chord, using the cylinder’s diameter as the chord, rather than degrees around the cylinder. The two blowing cases illustrate a standard observation: the stronger the blowing, the farther Cp distribution for different slot locations Re=26000, U,/Um=14.5, b/R=O.Oll
Fig. 43 Pressure distributions on a cylinder for various slot locations.
154
D. CERCHIE ET AL. Comparison for three Cases
Fig. 44 Pressure distributions on a cylinder for three different flow control approaches.
downstream into the adverse pressure region the jet can be introduced for maximum performance. Note also that the suction and oscillatory (AFC) flow control have nearly the same pressure distribution from the LE to the slot, while the AFC input magnitude is less than one-third the suction magnitude. The AFC and steady blowing pressure distributions differ from the steady suction case aft of the slot location. The steady suction pressure distribution has a sharp increase in pressure whereas the steady blowing and AFC possess a more gradual change in pressure toward the TE. Judging from the pressure distribution, the flow is fully attached for the steady blowing at C, = 39%, but it is separated downstream of the slot when comparable suction is used. Active flow control manages to attach the flow over 95% of the chord at C, = 9%. The integrated force coefficient CF was measured and plotted in Fig. 45a as a function of slot angle relative to the LE for different slot velocities and heights. In general, a higher C, generates a higher force. An attempt to normalize this data with respect to the jet momentum (as was done in the absence of an external stream) is plotted in Fig. 45b. The data nearly collapse to a single curve for small jet injection angles relative to the ideal LE, but the scatter is still large downstream of the slot location corresponding to .,,,C , The pressure distributions along the cylinder for three different slot velocities are plotted in Fig. 46. In the first case the pressure distribution is normalized by the freestream dynamic pressure q, whereas in the second it is normalized by the cylinder diameter and by the jet momentum J . Again, the trend is a proportional relationship between slot velocity and the freestream, but neither q nor J normalize the pressure distributions correctly. In the first case the higher the ratio between the jet and the freestream velocity, the higher the negative C ,, while
CIRCULATION AND SEPARATION CONTROL EXPERIMENTS
155
a)
Fig. 45 Total force generated on cylinder for different blowing conditions: a) C,, b) CFIC,.u
in the second the roles are reversed. Some other parameters are required for better correlation of the data, although self-similarity in C, may not be possible for the entire range of velocity ratios. A possible parameter would include C, and a correction to that term that would be flow configuration dependent. One such approach was
156
D. CERCHIE ET AL.
Fig. 46 Pressure distributions on a cylinder for different blowing velocities and normalization.
proposed by Englar26:
where Us is the local velocity measured just upstream of the slot. This local velocity is not always available from past tests and should scale with freestream velocity. Therefore, using a simplified assumption, the equation can be written in
CIRCULATION AND SEPARATION CONTROL EXPERIMENTS
157
the following form for discussion purposes:
In fact, when a jet of kinematic momentum J and volumetric flow q is injected parallel to an infinite stream of velocity U,, in a way that retains constant pressure on the surfaces bounding a control volume, the application of the momentum theorem yields an invariant
which is proportional to Englar’s simplified version of C , . The two derivations lead to the same result when J is calibrated in the absence of an external stream, which is normally the case. The last equation is applied to some early data in boundary layer flow and circulation control to determine its applicability. The blown flap data from Poisson-Quinton and Lepage2 is adjusted using the modified flow coefficient and the results are replotted in Fig. 47 together with the original results that use just C,. Similar reprocessing was carried out to the win data of Williams21 (Fig. 48) and to the NACA 84-M airfoil data of Attinello (Fig. 49). In all these cases the data scatter is no greater using the modified parameter, and in some of the more critical, low-momentum blowing tests the inclusion of CQ seems to fit the data better. Large C, values in testing are typically the result of large slot velocities. For these conditions, the impact of the correction term diminishes, allowing for reasonable performance predictions to be made for a given configuration. In an effort to minimize the momentum input required for effective flow control, this modified parameter provides a better insight into the measured data for smaller flow coefficients. In Fig. 49 the data collapse better using the modified parameter for the small corrected flow coefficient conditions, even matching the measured data while it shows a negative lift increment when the flow coefficient is negative. However, it is not suggested that these data justify a new term for design purposes. What this does show is that in order to optimize flow control using a minimum input, a flow-related correction parameter would have to be added to C,. It is unknown whether a correction factor can be introduced that would collapse all flow geometries and control strategies to a unique line without the introduction of terms that account for the pressure gradient and boundary layer characteristics in the region of the slot. An interesting characteristic of flow control is shown in Fig. 50. The cylinder exhibits a lift hysteresis around its maximum lift condition due to changes in slot
9
158
D. CERCHIE ET AL. CL vs. Cp or (0.5*CpCQ) (from Polsson-Qulnlon& Lepage, 1961)
Fig. 47 Blowing over a wing section using data from Ref. 2 (Poisson-Quintonand Lepage). CL vs. Cp or (0.5*Cp-CQ)
Fig. 48 Blowing over a wing section using data from Ref. 2 (Williams).
CIRCULATION AND SEPARATION CONTROL EXPERIMENTS
159
CL vs. Cp or (0.5*CpCQ) (from John Attinello, 1961)
Fig. 49 Blowing over a wing section using data from Ref. 2 (Attinello).
location, under steady blowing. This phenomenon is similar to an airfoil when the Therefore, the initial flow condition is angle of incidence is altered around astall. also important to the performance of BLC and CC, as was discussed by Nishri and Wygnan~ki.'~ Previous tests on a Wortmann airfoil27 have shown that this hysteresis can be reduced or eliminated by using AFC. The pressure distribution around the cylinder with a slot located at 120 deg from the LE indicates the flow can be either attached, yielding a maximum lift, or completely detached, providing no lift at all (Fig. 50b). The latter pressure distribution is very close to the baseline data. The significance of a single slot location to the performance of the prescribed control mechanism has been amply demonstrated, but very few attempts have been made to add another slot with another array of actuators or jets. A second slot was thus added to cylinder 3 (Fig. 42) to determine its possible impact. Figure 51 shows the dual-slot configuration, where both slots are on the same side of the cylinder and they inject flow in the same direction. The performance of this configuration relative to the single-slot performance for the same blowing coefficient is compared. The dual-slot arrangement provides 30% more lift relative to the single slot by improving the downstream flow conditions. It is interesting to note that the second slot provides only a limited control of separation downstream, but it alters mostly the pressure distribution upstream, contributing to supercirculation. This emphasizes the significance of the location of the fluidic actuator and suggests that a distributed actuation may be more effective, as the designers of the NOTAR helicopter demonstrated.
160 a)
Pressure Comparison, Cp=O,57, or=190°
Fig. 50 Lift and pressure hysteresis for changing blowing location on a cylinder.
Control of lift is not the sole purpose of BLC, because drag reduction for cruise might be as important. By blowing from two symmetrical slots located at a = & 110 deg away from the lead stagnation point, the pressure or velocity over the cylinder can be changed without introducing lift (Fig. 52). The inviscid value of the pressure coefficient of - 3 is realized at 50% of the chord position. The cylinder’s drag drops from approximately CD > 1.0 for the natural case (not blown) case to a CD x 0.4 for the symmetrically blown case. Proper accounting should be made in order not to waste momentum, because too much flow control may be applied for given conditions. Figure 53 shows that for symmetric blowing on the cylinder at C, = 0.46, the total drag reduction is 0.61 and all of the injected flow is recovered as thrust. However, when C, is increased to 1.02,
CIRCULATION AND SEPARATION CONTROL EXPERIMENTS
161
Comparison of Pressure Distribution slot located at u=9O0 Cylinder # 3, Re=39000, Cp=1.31
Fig. 51 Pressure distributions on a single- and dual-slot cylinder.
the total drag reduction is only 0.69 and some of the injected flow is wasted. The latter case may provide other benefits such as enhanced stability, but the performance suffers. It is also interesting to note the increase in the wake's vortex shedding frequency, which is associated with the reduction in drag (Fig. 53) and is equivalent to the momentum thickness in the wake. This is Double slotted Cylinder 1loo, 0' sweep back, Re135000, Steady Suction, C, per slot
Fig. 52 Pressure distributions on a dual-slotted cylinder for different suction C,.
162
D. CERCHIE ET AL.
C,
0.00 0.46 1.02
C D ~1.02 0.26 0.18 Co
1.04 0.43
0.35
0.00 -0.14 0.32
Fig. 53 Cp distribution and wake profiles associated with a dual-slotted cylinder.
consistent with observations suggesting that a dominant Strouhal number associated with vortex shedding is constant, requiring a frequency increase resulting from a decrease in the momentum thickness. 111. Conclusions
Flow control tests over five two-dimensional, aerodynamic, and bluff bodies provided some new insight into the parameters governing active control of separation and circulation. Suction, blowing, and periodic forcing enable one to tailor the pressure distribution over the airfoil surface in a similar fashion to lofting or the introduction of passive devices such as flaps or control surfaces. This chapter questions some of the accepted concepts associated with separation and circulation control, although it is not able to provide definitive answers to these questions. For example, the enhancement of circulation can be achieved without attaching the flow; however, by forcing separated flow to reattach to the surface, circulation is generally enhanced. Flow reattachment and the control of circulation do not have to occur sequentially, as it is widely believed, leading to some critical value of an input parameter that separates the two.
CIRCULATION AND SEPARATION CONTROL EXPERIMENTS
163
The use of steady blowing is ineffective at low C,, but its ability to contribute to thrust while increasing the circulation make it attractive at C , > 5%. Because 0.05 represents a typical ratio of thrust to lift on a civilian airplane in cruise, steady blowing might become the technique of choice whenever the integration of propulsion and aerodynamics are considered. Suction and periodic excitation are much more effective in reattaching the flow at low C , values and enhancing the lift, but the input momentum is usually not recovered. Steady suction, in particular, contributes to drag. The low levels of C , required to attach the flow by periodic excitation make it the most attractive technique for lift enhancement. The momentum coefficient that is widely used to correlate the data is not as universal a quantity as it is believed to be. It is mostly deficient at low-level inputs that characterize AFC, requiring consideration of mass addition and perhaps other variables (e.g., changes in the Kutta condition and pressure gradient at the slot). A corrected flow coefficient term was discussed and applied successfully to some well-known flow control cases. The modified control parameter may be especially useful for correlating data using lowintensity, pulsed blowing or suction. A better understanding of the performance of slot design, its width, and velocity for various geometries is required. The drag of a two-dimensional bluff body is dominated by pressure drag because the skin friction contribution to drag is small in comparison. However, in the presence of fluidic control the reliance on CDpas being the major contributor to drag is wrong. In most cases, as C , increases, so does CDpwhile the actual drag decreases. This is most obvious when the TE is blunt and the flow is attached due to fluidic control. One wonders how to normalize the pressure distribution around an airfoil when C , is of order 1 because then qc % J . The controlled flows over the ellipse and the circular cylinder have many features in common. On the ellipse, however, one may change the pressure gradient upstream of the slot by changing the location of the slot relative to the center of the ellipse or by changing the thickness ratio of the ellipse. The TE performance becomes less predictable as the pressure gradient along the forward section of the elliptical airfoil is increased. An empirical parameter was introduced to scale these results and make a rational comparison with other geometries that are creating large circulation at the TE. The circular cylinder data demonstrated the importance of a good slot design as well as the superior efficiency of oscillatory flow control relative to steady blowing or suction. It also showed that there is a limit to blowing that should be observed if the overall system performance is to be objective. The curvature of the surface downstream of a slot is an important variable, as was demonstrated on the truncated NACA and GLAS I1 airfoils. A concave surface generates a positive pressure in the TE region that reduces the drag; however, it is susceptible to centrifugal instability and may generate streamwise vortices whose effect on separation and CC is not well understood. Periodic excitation emanating from a two-dimensional slot utilizes the inherent convective instability of the flow that is either separated or is on the verge of separation. Curvature renders another mode of instability that can be used to delay separation. It may enable the replacement of slot blowing by compact individual nozzles whose spanwise separation triggers the centrifugal instability associated with curvature.
164
D. CERCHIE ET AL.
Acknowledgments This work was supported in part by a grant from ONR that was monitored by R. Joslin. The authors are indebted to A. Hassan for his help in providing the CFD results that guided some of the experiments on the ellipse and the GLAS I1 airfoil. The authors wish to acknowledge H. Nagib for providing them with an important figure and for many helpful discussions related to AFC.
References ‘Prandtl, L., “The Generation of Vortices in Fluid of Small Viscosity,” Journal of the Royal Aeronautical Society, Vol. 31, 1927, p. 735. 2Lachmann, G. V. (ed.), Boundary Layer and Flow Control: Its Principles and Application, Pergamon Press, New York, 1961. 3Goldschmied, F. R., “Integrated Hull Design, Boundary Layer Control and Propulsion of Submerged Bodies,” Journal of Hydronautics, Vol. 1, 1967 p. 2. 4Poisson-Quinton, Ph., “Recherches Theoriques et Experimentales sur le Controle de Couche Limites,” VII International Congress of Applied Mechanics, 1948. ’Stratford, B. S., “Early Thoughts on the Jet Flap,” Aeronautical Quarterly, Vol. VII, 1956, p. 45. 6Helmbold, H. B., “The Lift of a Blowing Wing in a Parallel Stream,” Journal of the Aeronautical Sciences, Vol. 22, 1955, p. 341. 7Spence, D. A., “The Lift Coefficient of a Jet-Flapped Wing,” Proceedings of Royal Society Series A, Vol. 238, 1956, p. 46. ‘Legendre, R., Influence de 1’Emissiond’un Jet au bord de Fuite d’un Prof1 sur 1’Ecoulement autour de ce Profil, Comptes Rendus, AcadBmie des Sciences, Paris, 1956. ’Woods, L. C., “Some Contributions to Jet-Flap Theory and to the Theory of Source Flow from Aerofoils,” A.R.C. Current Paper 388, 1958. “Malavard, L., Sur une Thkorie Lineaire du Souflage au bord de Fuite d’un Profil d’Aile, Comptes Rendus, AcadBmie des Sciences, Paris, 1956. “Wygnanski, I., “The Effect of Jet Entrainment on Loss of Thrust on a TwoDimensional Jet-Flap Aerofoil,” Aeronautical Quarterly, Vol. 17, 1966, pp. 31 -51. ‘2Wygnanski, I., and Newman, B. G., “The Effect of Jet Entrainment on Lift and Moment for a Thin Airfoil with Blowing,” Aeronautical Quarterly, Vol. XV, 1964, p. 122. 13Hynes,C. S., “The Lift, Stalling and Wake Characteristics of a Jet Flapped Airfoil in a Two Dimensional Channel,” Stanford Univ., Stanford, CA, SUDAAR No. 363, 1968. 14Stratford, B. S., “An Experimental Flow with Zero Skin Friction Throughout its Region of Pressure Rise,” Journal of Fluid Mechanics, Vol. 5, 1959, pp. 17-35. ‘’Elsberry, K., Loeffler, J., Zhou, M. D., and Wygnanski, I., “An Experimental Study of a Boundary Layer that is Maintained on the Verge of Separation,” Journal of Fluid Mechanics, Vol. 423, 2000, pp. 227-261. ‘%eifert, A., and Pack, L. G., “Active Separation Control on Wall Mounted Hump at High Reynolds Numbers,” A I M Journal, Vol. 40, No. 7, 2002, pp. 1363- 1372. 17Greenblatt, D., Paschal, K. B., Yao, S. C., Harris, J., Schaeffler, N. W., and Washburn, A. E., “A Separation Control CFD Validation Test Case,” AIAA Paper 2004-2220, June 2004. 18Greenblatt, D., and Wygnanski, I., “The Control of Flow Separation by Periodic Excitation,” Progress in Aerospace Sciences, Vol. 36, 2001, pp. 487-545.
CIRCULATION AND SEPARATION CONTROL EXPERIMENTS
165
‘’Nishri, B., and Wygnanski, I., “Effects of Periodic Excitation on Turbulent Flow Separation from a Flap,” AIAA Journal, Vol. 36, No. 4, 1998, pp. 547-556. ”Darabi, A., and Wygnanski, I., “Active Management of Naturally Separated Flow Over a Solid Surface, Part 11: The Separation Process,” Journal of Fluid Mechanics, Vol. 510, 2004, pp. 131- 144. ”Williams, J. “British Research on Boundary Layer Control for High Lift,” Boundary Layer and Flow Control: Its Principles and Applications, edited by G. V. Lachmann, Pergamon Press, New York, 1961. ”Glauert, M. B., Walker, W. S., Raymer, W. G., and Gregory, N., “Wind Tunnel Tests on Thick Suction Airfoil with a Single Slot,” Aeronautical Research Council R & M, No. 2646, Oct. 1948. 23Salter,C., Miles, C. J. W., and Owen, R., “Tests on GLAS I1 Wing Without Suction in the Compressed Air Wind Tunnel,” Aeronautical Research Council R & M No. 2540, Feb. 1948. 24Seifert, A., Bachar, T., Koss, D., Shepshelovich, M., and Wygnanski, I., “Oscillatory Blowing, a Tool to Delay Boundary Layer Separation,” AZAA Journal, Vol. 31, 1993, pp. 2052. 25Neuendorf,R., Lourenco, L., and Wygnanski, I., “On Large Streamwise Structures in a Wall Jet Flowing over a Circular Cylinder,” Physics of Fluids, Vol. 16, No. 6, 2004, pp. 2158-2169. 26Englar, R. J., “Test Techniques for High Lift Two-Dimensional Airfoils with Boundary Layer and Circulation Control for Application to Rotary Wing Aircraft,” Canadian Aeronautics and Space Inst. Annual General Meeting, “Practical Aspects of V/STOL Wind Tunnel Testing,” May 1972, pp. 1-50. ”Neuburger, D., and Wygnanski, I., “The Use of a Vibrating Ribbon to Delay Separation on Two Dimensional Airfoils,” Proceedings of Air Force Academy Workshop on Separated Flow, F.J. Seiler Research Labs. Rept. TR-88-0004, 1987.
Chapter 6
Noise Reduction Through Circulation Control Scott E. Munro,* Krishan K. Ahuja,+ and Robert J. Englar' Georgia Institute of Technology, Atlanta, Georgia
Nomenclature a = speed of sound c = chord c1 = airfoil lift coefficient h = slot height riz = mass flow rate p = pressure q = dynamic pressure, $pV2 R = radial distance from jet exit to measurement location r = radius of CCW surface Re = Reynolds number T = temperature V = velocity a = angle of attack Af = frequency bandwidth for narrowband acoustic spectra 0 = polar angle (with respect to the flow axis) p = density Subscripts s = associated with slot T = associated with tunnel freestream *Graduate Student, School of Aerospace Engineering; currently at Naval Air Warfare Center, Weapons Division, China Lake, California. Student Member A I M . 'Regents Researcher and Professor, Georgia Tech Research Institute, and School of Aerospace Engineering. Fellow, AIAA. *Principal Research Engineer, Georgia Tech Research Institute, ATAS Lab. Associate Fellow AIAA. Copyright 0 2005 by the authors. Published by the American Institute of Aeronautics and Astronautics, Inc., with permission.
167
168
S. E. MUNRO, K. K. AHUJA, AND R. J. ENGLAR
j = associated with jet o = ambient condition
I. Introduction NE of the major environmental dilemmas facing today's aircraft industry is noise pollution from aircraft, especially around the airport. There is a large emphasis on minimizing community noise due to operation of aircraft at and around airports. Thus, airlines, aircraft manufacturers, The National Aeronautics and Space Administration (NASA), and the Federal Aviation Administration (FAA) have made reducing aircraft noise a priority. NASA has proposed a goal of lowering total aircraft noise emissions by 20 (effective pressure noise level) EPNdB by 2020. In order to meet this goal, NASA and other organizations have been encouraging innovative research to help reduce aircraft noise. Because a major contributor to aircraft noise on approach is airframe noise (or perhaps even on takeoff if the engine noise is eliminated), reducing this noise would be helpful in reaching the industry goals. The major contributors to airframe noise are the landing gear, the slats, and the flaps. Much work has been done in these areas in the last five years in an effort to reduce their noise emissions. Of course, the best solution would be to have an aircraft without these protrusions into the flowfield. Obviously, an aircraft without landing gear would have serious drawbacks, but there are alternative high-lift systems that could replace conventional wing flaps and slats, which have shown great promise in maintaining and even surpassing the lifting benefits of conventional flaps. Circulation control wings (CCW) have been researched and developed extensively, primarily for the purpose of increasing performance and reducing or replacing the conventional flap system of an aircraft.' Over the years, CCW systems have gone through many configuration designs for many different applications, including versions for rotorcraft, fighter aircraft, and short haul transports.' However, there has been limited research investigating the possible acoustic benefits provided by such a system, other than occasional references to smaller noise footprints due to shorter takeoff and landing distances. The only known work on the acoustics of CCW is that of Salikuddin et a1.,2 who evaluated the noise field of an upper surface blown wing with circulation control (CC). That study, however, did not provide an indication of the acoustic benefits of a CCW compared with a conventional wing for the same lift. Because CCW systems have already been shown to be an adequate replacement for conventional flap systems in the aerodynamic realm,' they are immediately a candidate for reducing airframe noise as they eliminate much of the structure of the conventional flap system that protrudes into the flow. However, there are many issues that need to be resolved before the claims of lower noise are validated. The CCW system has never been evaluated on an acoustics basis, so it must be optimized for this, while maintaining, at a minimum, the lift characteristics of a conventional system. The acoustic
0
NOISE REDUCTION THROUGH CC
169
impact of several parameters must be investigated, such as the blowing slot height, slot velocity, and CCW geometric configuration (i.e., flap type and deflection angle). In order to correctly define the best combination, new areas of research will have to be investigated, including jet noise of extremely high aspect ratio (AR) nozzles, and the effects of jet turning on its noise propagation. These many issues are the motivation of the present study. The current work involves both experimental and computational efforts. Only experimental results are presented in this chapter. Computational results are presented in Chapter 22 of this volume and in Ref. 3. 11. Background
The CCW concept has been researched since the 1960s. The CCW uses a rounded trailing edge (Fig. l).' Air is blown tangentially along the upper surface from a plenum supply inside the wing through a slot just upstream of the rounded trailing edge (TE). Blowing moves the upper surface separation point around the TE, thus changing the TE stagnation point location, and hence the circulation for the entire wing. The higher-speed air moving along the surface also causes a suction peak in this region and contributes to increased lift. The slot flow remains attached to the surface due to the so-called Coanda e f f e ~ tAt . ~ low blowing velocities, the tangential blowing behaves similarly to a boundary layer control (BLC) device by adding energy to the slow-moving flow near the surface. At higher blowing rates, the lift is increased by the change in circulation already described. A CCW can be designed without any mechanical moving elements if desired. This is achieved using a rounded TE, where the amount of lift is controlled by the pressure valve to the supply
TANQENTIAL BLOWING OVER ROUNDED TFWILINQ EDGE SURFACE
Fig. 1 Schematic circulation control wing concept.
170
S. E. MUNRO, K. K. AHUJA, AND R. J. ENGLAR
plenum. This eliminates the need for flaps with hinges, tracks, screw drives, and hydraulics. The increment in lift generated is controlled by the nondimensional parameter C,, defined using slot and freestream properties:
With a wing, the nondimensionalizing area is the wing surface S. For an airfoil, C, is typically given in C,/ft, because the chord is the only available reference length. In general, a given C, will provide a given increment in the lift coefficient over the entire range of angles of attack below stall. The exception to this is when the slot jet velocities or slot heights are large enough to cause the jet to separate prematurely. Thus, C, is used extensively in the literature when discussing CC. The large, circular trailing edges used in many of the early experiments evolved into a dual-radius hinged flap, mainly because the nonsharp TE greatly increased drag.195-8 The hinged flap was a compromise of several desired features. The flap had a curved upper surface, like the cylindrical TE, but a flat lower surface. This overcame the problem of high drag in cruise associated with the nonsharp TE of the early designs. Overall, the hinged-flap, dual-radius design still maintained most of the CC lift advantages, but greatly reduced the drag problem associated with the circular TE system. The flap itself has several mechanical advantages compared to conventional Fowler flap systems. The flap is about one-fourth to one-third the size of a conventional flap. This means lower flap weight, and so fewer structural components are required to hold it in place.8 The flap is also a simple hinged flap, rather than a complex Fowler-type flap that requires complex gearing, tracks, and through gaps, which most likely contribute to airframe noise on their own. The reduced size and simplicity of the CCW system, even with a small flap, clearly offers some advantage over a conventional system. There are many potential uses for circulation control. However, the two applications that have received the most research attention have been CC rotors (CCR) and CCW applied to an aircraft for short takeoff and landing (STOL) capability. The reader is referred to Refs. 1 and 5 , where further details and citations on CCW research can be found. Some research pertinent to the present work is briefly mentioned below. The Navy sponsored a full-scale flight-test program on an A-6/CCW in the late 1970s. The design, tests, and results are documented in Refs. 9- 11. Research has also been carried out to investigate applying the CC system to a Boeing 737 type of aircraft. A summary of the effort is documented in Ref. 6. The only known acoustic work on CCW configurations was performed by Salikuddin et a1.2 There are other otential uses for CC, including automotive and in helicopter~,’,’~where noise reduction may also be appropriate. The acoustic benefits shown in this paper should be applicable to other areas also.
NOISE REDUCTION THROUGH CC
171
111. Facilities and Instrumentation The anechoic flight simulation facility (AFSF) was used in the experiments. It is located at Georgia Tech Research Institute (GTRI) located at its Cobb County Research Facility in Smyma, Georgia. The AFSF operates in an open-jet windtunnel configuration. It is an anechoic facility that allows acoustic measurements to be made in the presence of a freestream (Fig. 2). The tunnel inlet has a square inlet that converges down to a 28-in. round duct. The duct terminates in an anechoic room as an open jet. Protruding out from the downstream wall is the collector, which is 4 ft wide by 5 ft high. The collector duct extends outside the building and ends at a centrifugal fan powered by a diesel engine. The facility is open circuit, drawing air from outdoors. The details of the facility can be found in Refs. 14 and 15. In the current experiments, the wings are mounted via mounting brackets to the open jet. This locates the wing across the jet opening immediately downstream of the end of the duct. Figure 3 shows one of the conventional wings mounted at the exit of the open jet. The ambient pressure in the chamber, the plenum pressure for the slot, pressures in the air supply line venturi mass flow meter, and pressure in the inlet (for freestream velocity) were monitored on individual pressure transducers and manually recorded for each test point. Acoustic measurements were made with B&K, 4135, 1/4-in. microphones. One microphone was mounted on a traverse system that translated the microphone from angles of 30 deg to 90 deg (where 0 deg is the freestream direction). This system was arranged to make all measurements in the flyover plane. The
Collector
Fig. 2 Schematic of anechoic flight simulation facility (AFSF).
172
S. E. MUNRO, K. K. AHUJA, AND R. J. ENGLAR
Fig. 3 Photo of a conventional wing mounted in AFSF.
microphone was connected to a multichannel digital frequency analyzer, which is run by software on a PC. Figure 4 shows a schematic of the blowing system for the CCW. It consists of high-pressure 3/4-in. tubing, a mass flow venturi, pressure gauges, and a muffler. On the upstream end, the tubing is connected to an existing high-pressure line with a control valve upstream. The flow passes through a mass flow venturi, and then goes through more tubing to an in-house built muffler, which absorbs the upstream valve noise. Downstream of the muffler, the air passes through more tubing to inlets for the CCW plenum. The test model wing used in Ref. 6 was used as the test model for this study. This CCW model, shown in Fig. 5 , has a supercritical baseline airfoil shape, but has many different detachable CCW TE configurations. These included different sized flaps and cylindrical trailing edges. Based on past aerodynamic studies, the best overall aerodynamic characteristics were obtained with the small CCW flap configurations. The small deflectable flap allowed for low drag during cruise, but by blowing over the curved upper surface with the flap deflected, significant flow turning could still be achieved when desired. The highest lift configuration was found to be with the flap deflected 90 deg. This was used as the starting configuration for the current acoustic tests. The conventional wing had the same general shape as the CCW over most of the chord. However, its trailing edge was altered with a cutout for a stowed flap. A single-slotted Fowler flap was attached. Two different flaps were tested. The flap was deflected 30 or 40 deg from the chord line to simulate a landing configuration. Both flaps spanned the entire wing, but one flap had a cutout in at the
NOISE REDUCTION THROUGH CC
173
1" High Pressure
Piping
Fig. 4 CCW blowing system configuration.
midspan point. Figure 6 shows the airfoil profile of the model and a drawing depicting the flap cutout. The photo in Fig. 3 is of the model installed in the AFSF. The cutout is to simulate the cutouts on a real aircraft. Cutouts are often present for structural reasons or to prevent engine exhaust from impinging on a lowered flap.
IV. Technical Approach The current work focused on optimizing a CCW system for low noise impact while maintaining aerodynamic performance sufficient for direct comparison to a conventional flapped-wing configuration. The first step was to determine if and how a CCW configuration can have lower noise than a conventional system. This step involved side-by-side comparison of representative configurations
Supercritlcsl Contour
Fig. 5 Schematic of CCW flap-wing configuration, generic supercritical airfoil shape.
174
S. E. MUNRO, K. K. AHUJA, AND R. J. ENGLAR
a)
Fig. 6 a) Schematic of conventional flap-wing configuration, generic supercritical airfoil shape; b) drawing of conventional wing with flap with cutout.
under the same conditions, that is, the same freestream flow and lift conditions. Because there are several variations of CCW systems that have been researched, a basic study of different CCW configurations was carried out. The test models were used in other aerodynamic experiments, so this also allowed the use of these data when making the acoustic comparisons. The optimized blowing configuration was compared with a conventional wing system. Basic noise spectra of the CCW and conventional wing configurations were acquired at several mean flow velocities and angles of attack. Specific cases where the different configurations had the same lift coefficient were then compared directly. Lift data from previous studies were used for this comparison.
V. Results and Discussion A. Acoustic Optimization of Existing CCW State-of-the-Art Configurations The CCW concept has been around for nearly 40 years, and there have been many advances, changes, and modifications to the basic concept to improve its overall performance. To attempt to test acoustically all the different configurations would be unreasonable, because many of the changes were made to improve the system. There is little reason to test acoustically a system that
NOISE REDUCTION THROUGH CC
175
is technologically surpassed by a better version. Thus, the goal of the current study is to investigate two or three of the best performing CCW configurations. Based on previous aerodynamic work, the CCW with its flap deflected 90 deg was chosen as the starting point for the study (a possible high-lift configuration for landing approach). This had the best overall high-lift aerodynamic performance of several configurations tested in previous studies. The flap was eventually adjusted to 30-deg deflection to prevent flap-edge vortex shedding noise that was present in the 90-deg arrangement. Six slot heights were chosen for the optimization study, ranging from 0.003 to 0.020 in. These dimensions were chosen because they were typical slot heights used in earlier aerodynamic studies.6 A wide range of slot Mach numbers was evaluated, ranging from 0.3 to 1.2. The acoustically optimized CCW test configuration was compared with a conventional flap configuration. The conventional model had the same generic airfoil shape as the CCW, except near the trailing edge to accommodate the conventional flap. The flap chord was about 30% of the wing chord and deflected 40 deg to simulate a landing configuration. Data were acquired for each test configuration at freestream speeds of 100, 150, 200, and 250 ft/s (nominal) and at geometric angles of attack of 0,7, and 14 deg. The majority of the data presented in this section was acquired at a geometric angle of attack of 0 deg and at the highest freestream velocity of about 240 ft/s unless otherwise noted. Figure 7 shows acoustic spectra for several slot velocities with no freestream flow for the CCW with the 90-deg flap configuration. It shows a similar trend to the basic jet velocity scaling property develo ed for round jets. For the measured velocities, V8 scaling of jet noise theory' predicts about a 19 dB increase between the two most extreme cases, which is similar to that
F
Frequency, kHz (Af = 32 Hz)
Fig. 7 CCW blowing system noise spectra with no freestream flow; VT = 0 ft/s, h = 0.006 in.
176
S. E. MUNRO, K. K. AHUJA, AND R. J. ENGLAR
measured (about 16 dB) above 2 kHz. Some noise due to scrubbing of the slot jet over the flap surface is likely to be present as well. It appears that the majority of the noise is associated with the jet noise from the slot and not due to internal model and facility noise associated with the blowing system above 2 kHz. However, below 2 kHz the scaling is not followed in the data. This is most likely due to internal noise that is generated from the flow into the wing on its way to the slot. This contaminates the signal, making the noise higher for the lower slot velocities, but not affecting the higher velocities where the jet mixing noise is expected to be dominant. Thus, the difference between the data is less than predicted by the theory. This is supported by Fig. 8. Figure 8 shows the spectra out to a frequency of 60 kHz. These figures show two slot heights, and hence two slot areas, at the same slot velocity. However, inside the wing the areas in the flow path remain the same. Because the mass flow into the wing must be the same as the mass flow out, the doubling of the exit area roughly causes a doubling of the mass flow at the exit, and hence a doubling of the mass flow inside the wing. However, because all the areas inside the wing are constant, the velocity must double inside the wing in order to double the mass flow. Thus, if noise is dominated by the internal noise, it should follow a sixth-power law of the internal velocity, as this noise is expected to be dipolelike in nature. If so, the data should reflect an 18 dB increase. However, if the noise is dominated by externally produced jet mixing noise, then it will change only to the extent that the exit area has changed. Based upon the available
Frequency, Hz (Af = 32 Hz)
Fig. 8 Sound pressure level (SPL) of CCW with different h and the same V,; 0 = 90 deg, V, = 660 ft/s, VT = 0 ft/s.
NOISE REDUCTION THROUGH CC
177
experience/theory on round jets16 this will provide for the jet mixing noise intensity proportional to slot exit area. This translates into a 3 dB increase in noise after shifting the spectrum for h = 0.006 in. to the left over the spectrum for h = 0.012 in. by a factor of one octave to allow for the shift in the noise frequencies proportional to a characteristic length. This number is somewhat smaller than the observed difference in the sound pressure levels (SPLs) of the two spectra in Fig. 8. All of these arguments assume that we can apply the lessons learned from round jets to very high AR jets. Yet, since the noise increase is of the order of 3 dB, it can be said that internal noise is not significant in this case. The fact that the observed difference in spectral SPLs is more than the expected 3 dB could also be associated with the scrubbing noise of the CCW slot jet moving over the rounded edge. If so, it is genuinely produced outside and is not contaminated by any internal noise. We believe that the data may be contaminated by noise generated internal to the wing below about 2 kHz. A muffler was built and installed in the supply line downstream of all valves to eliminate as much upstream noise as possible. However, due to the small thickness of the wing, inlets into the wing plenum are smaller than desired. This results in a relatively high velocity flow entering into the plenum, with no space to absorb the noise generated. It is believed that these noise sources may be causing a majority of the noise below 2 kHz where the noise is not following the typical V8jet noise scaling. For the time being, this will be noted and data below 2 kHz will be disregarded as either somewhat corrupted by internal noise or not understood. Figure 9 shows the noise spectra for several slot jet velocities at a constant freestream velocity and constant slot height of 0.003 in. There are several things to note. First, with no blowing there is a large-amplitude, well-defined tone. It is also important to note that, in general, the very low frequency noise (f< 4 kHz, approx.) is much greater compared to the data in Fig. 7. Some of this is from the tunnel noise itself (below about 500 Hz), but most of it is flow noise associated with the freestream flow around the wing. The tone is believed to be due to the shedding of vortices off the bluff trailing edge of the deflected flap (Stshedding = 0.2 would produce a shedding frequency of approximately 600Hz). Notice that blowing, even at low slot jet velocities, significantly reduces the magnitude of the tone. However, in this case it is not completely eliminated; in fact, it dominates the spectra at all blowing velocities. The aforementioned tone was unexpected. This presented a problem, because the tone dominated the spectrum at all blowing conditions; thus, any acoustic benefit derived from using the CCW over a conventional wing would be lost if the flap were deflected to 90 deg. Because of this, it was decided that reducing the flap deflection might produce a less dominant tone, but still provide enough lift with the right amount of blowing to equal that of a conventional wing. Figure 10 shows two curves with the flap set to 30 deg. In this case note that the tone is completely eliminated with a small amount of blowing. The computational study also produced the same result, and is presented in Ref. 3 and also in Chapter 22 of this volume. Not only is this advantageous for the current study, but this result could be used in other applications where similar shedding produces a distinct tone.
S. E. MUNRO, K. K. AHUJA, AND R. J. ENGLAR
178
h
2
'fo 7
X
w II
L
p!
n
Y
J
n v)
Frequency, kHz (Af = 32 Hz)
h
2
'p
z
X
w II
b)
Frequency, kHz (Af = 32 Hz)
Fig. 9 CCW with 90-deg flap and freestream velocity, 0 = 90 deg, V , = 220 ft/s, a ) f = 0-60 kHz; b)f= 0-5 kHz.
NOISE REDUCTION THROUGH CC
179
Frequency, kHz (Af = 32 Hz)
Fig. 10 CCW with 30-deg flap and freestream velocity, 0 = 90 deg, VT = 220 ft/s,
f = 0-5 kHz.
Data for test conditions similar to those for the 90 deg deflection are shown in Fig. 11. Again, with no blowing the tone is present. However, with small amounts of blowing, the separation is eliminated, and hence the tone is completely eliminated. Because this configuration showed more promise, the remaining parameters were optimized using the 30-deg flap configuration. Both slot height and slot jet velocity were examined. The effect of slot height was investigated next. Figure 1 2 shows data with similar freestream conditions but different slot heights. It is important to note that this figure compares different CCW configurations with the same lift. For the same C, at different h, the slot velocity will be different, because C, is dependent on mass flow from the slot. The goal is to compare the same lift, so it is best to look at the data where C, is constant, because the same C, will give the same lift in most cases. There is some variation of lift with h for high C,, but in the C, range of interest here, h does not have an independent effect on the results. Thus, the data in Fig. 1 2 show that there is a lower noise from the larger slot heights for a given lifting condition. This makes sense, because C, is proportional to mass flow through the slot. By increasing the slot height but maintaining the same mass flow (and hence same C,) the jet velocity of the slot is lower. At this point it appeared that the most appropriate conditions for comparing a CCW system to a conventional system had been found: maximize the slot height so that jet velocity is minimized. Unfortunately it was found that above a slot height of about 0.012 in. the noise began to increase (for constant C,). This was contrary to the logical trend associated with what should be happening, so some attention was given to
S. E. MUNRO, K. K. AHUJA, AND R. J. ENGLAR
180
h
2
& X
R
Frequency, kHz (Af = 32 Hz)
Fig. 11 CCW with 30-deg flap and freestream velocity, 0 = 90 deg, V , = 220 ft/s, f = 0-60 kHz.
Frequency, kHz (Af= 32 Hz)
Fig. 12 CCW with 30-deg flap at three different h, C p = 0.04, 0 = 90 deg, VT = 220 ft/s,f= 0-60 kHz.
NOISE REDUCTION THROUGH CC
181
why this was happening. If one looks more closely at C,, it contains a mass flow term. Initial results indicated that reducing the slot velocity reduced the noise. In the equation this means that V, would decrease. If one defines the mass flow term based on the mass flow “in” rather than “out” the problem becomes evident:
Density will vary with the pressure in the plenum ( p = P/RT), but it varies proportionally to slot velocity (as V, decreases, P decreases, and hence p decreases). Area is constant in the plenum regardless of slot height. Thus, in order to offset the decrease in V, and p, Vi, must increase. When this occurs, the internal noise associated with internal velocities will also increase. Figure 13 shows overall sound pressure level (OASPL) plotted against h for constant C,. If it is assumed that the highest slot velocity is dominated by external jet noise, the decrease in noise due to falling V, can also be plotted. In the figure the highest V, occurs at the smallest h. The drop in OASPL should follow the V8 scaling law. However, in this case keep in mind that the slot velocity drops due to an increase in slot area. Thus, the final estimated curve shows dropping OASPL due to slot velocity, but at a lower rate than V8 because of an increase in slot area. Note that the experimental data follow V8 scaling for some time but eventually increase away from the estimated dropoff. It is believed that this increase is due to
Fig. 13 OASPL for various C p = constant.
S. E. MUNRO, K. K. AHUJA, AND R. J. ENGLAR
182
the increasing dominance of internal noise as the slot velocity is reduced while the internal velocity is increased. Although this finding was unfortunate, it was not terribly detrimental to the study as long as one keeps in mind that proper design of the internal system will decrease the CCW noise further (in essence it should continue to drop along the estimated slot velocity dotted curve in Fig. 13 as the slot velocity is decreased). Thus, any benefit found will be enhanced with careful design of the internal system.
B. Determining an “Equal Lift” Condition The next step was figuring out how to compare the two lift augmentation systems. Aerodynamic data from previous studies were used for this (specifically from Ref. 6 ) .Aerodynamic data were available for both conventional wing configurations and the CCW in the form of lift curves (cl vs a curves). This was convenient, because for a CCW, a given C, will generally provide a Acl over the entire angle of attack range (not including the extreme high jet velocities and large slots where the jet separates from the surface). Thus, once the lift for the unblown CCW was found, this could be compared to the c1 for the conventional airfoil and the needed Acl could be calculated by subtracting the two values. This Acl was then used to determine the C, needed to match lift provided by the conventional wing flap system. Essentially, each C, is analogous to a flap setting that shifts the baseline lift curve by a given amount. For the particular CCW configuration (CCW with flap at 30 deg), a C, of about 0.04 produced about the same amount of lift as the conventional wings used in the experiments.
C. CCW vs Conventional Wings Two conventional wing configurations were tested: one configuration with a 30-deg flap spanning the entire span of the wing, and one with a flap deflected 40 deg spanning the entire wing except for a cutout region in the center span (see Fig. 6 for a drawing; Fig. 3 for a photo of it installed in the AFSF). These wings are the same basic airfoil shape as the CCW. The wings were tested at the same flow conditions as the CCW. Initially, the conventional wing with the 30-deg flap was tested. Figure 14 shows a comparison between the conventional wing with the 30-deg flap and the CCW configuration with lowest noise for the equivalent lift case. Because the h 0.012 in. data were the minimum CCW noise condition, they are presented in the figure. In the range between 1 and 10 kHz, the CCW has noise levels similar to those of the conventional system. Unfortunately, this was not the desired result, although it does provide assurance that using the CCW system does not increase the noise to the environment in its minimum noise configuration. However, many aircraft have a cutout in flaps across the span. This difference contributes a fair share of noise to a conventional wing system, because flap edge noise has been identified as a major contributor to airframe noise. Thus, this wing N
NOISE REDUCTION THROUGH CC
183
h
h
n v,
Frequency, kHz (Af = 32 Hz)
Fig. 14 CCW and conventional wing 2-D flap at similar lift condition, 0 = 90 deg, V T = 220 ft/s.
was missing a noise source that would most likely be greatly reduced in a CCW system. Because the CCW flap is much smaller, there is no need for a gap in the flap to avoid engine exhaust. Its small size would also in many cases reduce the need for gaps due to structural concerns. Thus the CCW system with a full span flap is not unreasonable. Acoustic tests were performed on the new configuration, similar to the previous tests. Figure 15 shows the comparison of the wing with the cutout flap with the CCW. As expected, the cutout in the flap increased the noise on the conventional system significantly and shows a significant advantage to using a CCW system in the region below 10 kHz and some advantage up to 40 kHz. Beyond 40 kHz, the two systems have similar noise levels. The data in this figure and following figures have different frequency ranges to emphasize the areas in the frequency spectrum where there are differences between the two systems. Similar results can be seen at other freestream velocities and angles of attack; however, the magnitude of the difference varies some depending on the conditions. Up to this point, only data from a microphone at 0 = 90 deg have been shown. This is only part of the noise picture; the changes in directivity of the noise between the two systems must be compared as well. Data were acquired at 30, 60, and 90 deg. It should be noted that there are some differences depending on the angle. Note that the 60 and 90 deg positions do not actually have a lineof-sight path to the slot exit, which is located on the top surface of the wing. It is also worth noting that the jet from the slot leaves the trailing edge of the
184
S. E. MUNRO, K. K. AHUJA, AND R. J. ENGLAR
Frequency, kHz (Af = 32 Hz)
Fig. 15 CCW and conventional wing with cutout at similar lift condition, 0 = 90 deg, VT = 220 ft/s.
wing at about 0 = 56 deg. Even with freestream velocity, the jet stays relatively close to that angle for some time beyond the trailing edge of the wing. Figure 16 compares the data for the two wing systems at 0 = 30 deg and 60 deg. At 30 deg the CCW system produces no real advantage over a conventional system. However, there is still some noise reduction in favor of the CCW system at 60 deg, similar to the 90 deg data shown earlier. These results indicate that a CCW system certainly has potential for reducing airframe noise. The results also show some trends of high-AR jets; however, there is still much left to study and resolve before all the aspects of the CC wing noise issues are solved and helpful to the design of a practical low-noise CCW system. To resolve some of the questions brought up by the CCW and to eliminate the possibility of internal noise contamination, a high-AR nozzle has been designed and fabricated. This nozzle is presently being tested by the authors in an anechoic facility and the intent is to produce a database of quality high-AR jet noise data that can be used to verify the speculations about internal noise in the experiments presented here. In addition, these data will be used to augment the present results by demonstrating the even greater benefits possible for a CCW high-lift configuration in reducing airframe noise.
VI. Conclusions Following on from the great interest in reducing aircraft noise, an innovative concept for eliminating a conventional flap system has been tested for its possible
NOISE REDUCTION THROUGH CC
a)
185
Frequency, kHz (Af = 32 Hz)
Fig. 16 CCW and conventional wing with cutout at similar lift condition, VT = 220 ft/s: a) 0 = 30 deg, b) 0 = 60 deg.
186
S. E. MUNRO, K. K. AHUJA, AND R. J. ENGLAR
acoustic advantages. Previous studies have shown that the CC wing is an aerodynamically viable alternative for conventional mechanical flaps. This study shows that there is also a substantial advantage in the acoustic realm. The results presented showed a lower noise spectrum for a CCW system compared to a conventional system for the same lifting condition. It should be noted that even if the CCW produces noise comparable to that of a conventional wing it is an advantage. This is because a CCW is expected to be much lighter than a conventional wing. It was also noted that the internal noise of the CCW blowing system of the model inhibited finding the full possible advantage a CCW system can offer. It is believed that careful design of a CCW blowing system, including internal details, could further improve the results shown here.
Acknowledgments This work was sponsored by NASA Grant NAG1-2146 through NASA Research Center Langley, under its Breakthrough Innovative Technology Program. The authors are grateful to L. Sankar of the AE school for many helpful discussions. Thanks are also due to C. Jameson for designing the HARN nozzle and to Rick Gaeta for his assistance in the experiments and many useful discussions. References ‘Englar, R. J., “Circulation Control Pneumatic Aerodynamics: Blown Force and Moment Augmentation and Modification; Past, Present & Future,” AIAA Paper 20002541, June 2000. *Salikuddin, M., Brown, W. H., and Ahuja, K. K., “Noise from a Circulation Control Wing with Upper Surface Blowing,” Journal of Aircraft, Vol. 24, No. 1, 1987. 3Liu, Y., Sankar, L. N., Englar, R. J., and Ahuja, K. K., “Numerical Simulations of the Steady and Unsteady Aerodynamic Characteristics of a Circulation Control Wing,” AIAA Paper 2001-0704, Jan. 2001. Also, see Chapter 22 of this volume. 4Dunham, J., “A Theory of Circulation Control by Slot-Blowing Applied to a Circular Cylinder,” Journal of Fluid Mechanics, Vol. 33, No. 3, 1968, pp. 495-514. 5Englar, R. J., and Applegate, C. A., “Circulation control-A Bibliography of DTNSRDC Research and Selected Outside References: January 1969 through December 1983,” David W. Taylor Naval Ship Research and Development Center, DTNSRDC-84/ 052, 1984. 6Englar, R. J., Smith, M. J., Kelley, S. M., and Rover, R. C., 111, “Development of Circulation Control Technology for Application to Advanced Subsonic Transport Aircraft,” AIAA Aerospace Sciences Meeting, AIAA Paper 93-0644, Jan. 1993. ’Englar, R. J., and Huson, G. G., “Development of Advanced Circulation Control Wing High Lift Airfoils,” AIAA Applied Aerodynamics Conference, AIAA Paper 831847, July 1983. 8 Englar, R. J., “Low-Speed Aerodynamic Characteristics of a Small, Fixed Trailing-Edge Circulation Control Wing Configuration Fitted to a Supercritical Airfoil,” David W. Taylor Naval Ship Research and Development Center, DTNSRDC/ASED81/08, March 1981.
NOISE REDUCTION THROUGH CC
187
’Nichols, J. H., Jr., Englar, R. J., Harris, M. J., and Huson, G. G., “Experimental Development of an Advanced Circulation Control Wing System for Navy STOL Aircraft,” AIAA Aerospace Sciences Meeting, AIAA Paper 81-0151, Jan. 1981. “Pugliese, A. J., and Englar, R. J., “Flight Testing the Circulation Control Wing,” AIAA Aircraft Systems and Technology Meeting, AIAA Paper 79-1791, Aug. 1979. “Nichols, J. H., Jr. et al., “Development of High Lift Devices for Application to Advanced Navy Aircraft,” DTNSRDC, Rept. DTNSRDC-80/058, AD A084-226, April 1980. ”Lane, P. Jr., “Ground Controls,” Racecar Engineering, Vol. 9, No. 8, 1999, pp. 20-23. 13Reader, K. R., “Hover Evaluation of the Circulation Control High Speed Rotor,” David W. Taylor Naval Ship Research and Development Center, Rept. 77-0034, June 1977. 14Ahuja, K. K., Tanna, H. K., and Tester, B. J., “An Experimental Study of Transmission, Reflection and Scattering of Sound in a Free Jet Flight Simulation Facility and Comparison with Theory,” Journal of Sound and Vibration, Vol. 75, No. 1, 1981, pp. 51-85. 15Ahuja, K. K., Tester, B. J., and Tanna, H. K., “The Free Jet as a Simulator of Forward Velocity Effects on Jet Noise,” NASA Contractor Rept. No. 3056, 1978. 16Ahuja,K. K., and Bushell, K. W., “An Experimental Study of Subsonic Jet Noise and Comparison with Theory,” Journal of Sound and Vibration, Vol. 30, No. 3, 1973, pp. 317-341. ”Tarn, C. K. W., and Zaman, K. B. M. Q., “Subsonic Jet Noise from Non-Axisymmetric and Tabbed Nozzles,” AIAA Paper 99-0077, 1999. 18Tam, C. K. W., and Auriault, L., “Jet Mixing Noise from Fine Scale Turbulence,” AIAA Paper 98-2354, 1998. ‘’Tarn, C. K. W., Golebiowski, M., and Seiner, J. M., “On the Two Components of Turbulent Mixing Noise from Supersonic Jets,” AIAA Paper 96-1716, 1996. ”Tam, C. K. W., “Influence of Nozzle Geometry on the Noise of High Speed Jets,” AIAA Paper 98-2255, 1998. ’lAhuja, K. K., “Correlation and Prediction of Jet Noise,” Journal of Sound and Vibration, Vol. 29 No. 2, 1973, pp. 155-168.
1I.B. Experiments and Applications: Aerospace
Chapter 7
Pneumatic Flap Performance for a Two-Dimensional Circulation Control Airfoil Gregory S . Jones* NASA Langley Research Center, Hampton, Virginia
Nomenclature A, = effective cross-sectional area of two-dimensional model b = airfoil two-dimensional span, in. C, = pressure coefficient c = airfoil chord, in. Cd = section profile-drag coefficient Cl = section lift coefficient (C, cos a - C, sin a ) C , = moment coefficient C, = normal force coefficient , C = fluidic power coefficient CT = thrust coefficient = C, C, = momentum coefficient (= rizuj/q(wc)) D = drag, lbf h = slot height of Coanda jet, in. H = tunnel height, in. Z,J,K = pressure tare coefficients for balance L = lift, lbf M = mach number riz = mass flow, lbm/s
*Research Scientist, Flow Physics and Control Branch. Senior Member AIM. Copyright 02005 by the American Institute of Aeronautics and Astronautics, Inc. No copyright is asserted in the United States under Title 17, U.S. Code. The U.S. Government has a royalty-free license to exercise all rights under the copyright claimed herein for Governmental purposes. All other rights are reserved by the copyright owner.
191
192
GREGORY
S. JONES
NPR = nozzle pressure ratio (= P,/P,) Pf = fluid power, ft.lb/s P = pressure, lbf/in.2 or lbf/ft2 p’ = fluctuating pressure, lbf/in.2 or lbf/ft2 q = dynamic pressure, lbf/ft2 ( = ;pU2) r = trailing edge radius, in. s = airfoil reference area, ft2 T = static temperature, OR t = airfoil thickness, in. U = velocity, ft/s u’ = fluctuating velocity, ft/s w = slot width, in. a = angle of attack, deg Sjet = reactionary force angle, deg p = Prandtl-Glauert compressibility ( d Ojet = Coanda jet separation angle, deg E = blockage interference ratio, u / U p = density, lbm/ft3 = circulation
m )
r
Subscripts jet, j = conditions at slot exit rake = conditions at rake location ram = conditions at engine inlet AOA = angle of attack, deg o = stagnation or total conditions 00 = freestream conditions
I. Introduction IRCULATION control (CC) technologies have been around since the early 1930s, and have been successfully demonstrated in laboratories and flight vehicles alike, yet there are few production aircraft flying today that implement these advances. These technologies are generally related to pneumatic devices falling into categories including jet flaps, blown flaps, and Coanda surfaces. Recent interest in CC aerodynamics has increased for both military and civil applications, with emphasis on providing better vehicle performance and prediction capability.’ The history of Coanda-driven CC has met with varying degrees of enthusiasm as the requirements for improved high-lift systems continue to increase. Current lift coefficient goals for extremely short take-off and landin (ESTOL) vehicles are approaching 10 and lift-to-drag ratios greater than 25. Personal air vehicles (PAV) have a field length goal of 250 ft.3 To achieve these goals will require more than what a conventional high-lift system can provide. In addition to high-lift and cruise drag requirements, the next generation of aircraft will need to address other issues, including weight4 and noise.5 Conventional high-lift systems that use flaps and leading edge (LE) slats can be associated with significant weight and volume penalties of a typical wing assembly. These assemblies are also complex (up to 3 and 4 subelements) and very
C
5
PNEUMATIC FLAP PERFORMANCE OF CC AIRFOIL
193
sensitive to location relative to the main element of the wing. The need to simplify and reduce the weight of these systems without sacrificing performance is the focus of this effort. Coanda-driven CC techniques generally offer high levels of lift for small amounts of These systems are perceived to be simpler and less weighty than conventional high-lift systems. However, advanced system studies of CC systems being applied to modem aircraft have been limited or nonexistent, and so the ability to buy its way onto an aircraft is generally unproven. Nevertheless, several blocks to real aircraft applications reappear in every discussion of CC. These include, source of air (typically bleed or bypass air from the engine or added auxiliary power unit), unknown weight penalties related to the internal air delivery system, engine out conditions, drag penalty associated with blunt trailing edge (TE), and large pitching moments associated with aircraft trim. Although this is not a comprehensive list, these issues will be used as a guide in developing a CC wing for general aviation applications. A primary objective of this effort is to evaluate the benefits of pulsed CC and to reduce the mass flow requirements for a given lift performance as well as to reduce the cruise drag penalty associated with a large CC trailing edge. Secondary objectives of this study were to evaluate the dual blown pneumatic concept as a control device and to determine potential benefits of returned thrust (i.e., thrust is lost at the engine due to bleeding mass from the engine, so how much thrust is returned to the aircraft through the wing). 11. NASA CC Requirements
Application of CC to different aircraft platforms is driven by requirements that are dictated by mission.’ The National Aeronautics and Space Administration (NASA) Vehicle Integration, Strategy and Technology Assessment (VISTA) office describe many of these missions. Each of the vehicle sectors within the VISTA program could benefit from CC technologies, but personal air vehicles (PAV) and ESTOL vehicles seem to benefit the most. The personal air vehicles shown in Fig. 1 have characteristics that resemble general aviation vehicles but meet stiffer requirements for field length (i.e., high lift), noise signatures, and cruise efficiency (L/D). With a fresh look at point-to-point travel, NASA’s PAV program will address airport infrastructure, ease of use, and reductions in the cost of travel. Today’s small aircraft utilize significantly oversized wings for cruise and simple hinged flaps for high lift. These systems are adequate for the current
Fig. 1 Notional concepts of NASA personal air vehicles.
GREGORY
194
S. JONES
airport infrastructure. However, as these airport requirements become more stringent, high lift and cruise efficiency must be improved. The PAV goals used for this effort included a 250 ft field length that will require resizing the wing with , , C = 4.0, yielding an L/D,,, of 20. In the near term, reduced approach a, speeds enable a 1000ft field length and can improve safety in addition to reducing community noise signatures. If equivalent control margins and gust sensitivity are achieved, safety (in terms of accident avoidance reaction time and survivability) is proportional to the approach speed. These reduced speeds require more efficient high-lift systems. Circulation control technologies have been identified as a candidate simplified high-lift system. It may be necessary to integrate this system with other active flow control technologies (combining higher altitude cruise, gust alleviation, limited powered-lift, and so on). Air sources for CC systems for small aircraft may have a low penalty. Current high-performance small aircraft are turbocharged for altitude compensation. At landing and takeoff conditions, compressed air is thrown out the wastegate of the turbocharger (approx. 2’ lbm/s). This is a potential source for air augmentation to a CC system. Because engine out conditions are an issue for CC applications, another air source alternative is using the wake vortex energy to power a wingtip-turbine. Regardless of the air source, it is important to optimize the efficiency of the CC system for minimizing mass flow at a given lift requirement. The NASA ESTOL vehicle sector requirements are directed to a 100-passenger class vehicle that would include the following elements: 1) 52000 ft balanced field length (related goal of C , = lo); 2 ) cruise at M = 0.8; 3) noise footprint contained within the airport boundary; and 4) landing speed -50 kt. The current state-of-the-art aircraft systems can only achieve two or three of these elements simultaneously. Circulation control has the potential of enabling the achievement of all the elements of the desired capability set and could be integrated to the high-lift, flight controls, and propulsion systems as shown in a notional aircraft in Fig. 2. It is recognized that the integration of the propulsion system and the wing is paramount to the success
LEADING EDGE Active Flow Control (High Lfl)
Fig. 2 Notional concept of NASA ESTOL 100-passenger vehicle showing potential CC vehicle.
PNEUMATIC FLAP PERFORMANCE OF CC AIRFOIL
195
of either of these vehicle concepts. The focus of this chapter will be targeted at a two-dimensional baseline CC airfoil proposal that could be applied to the outer wing panel of either concept. 111. Theoretical Considerations
The two-dimensional aerodynamic performance is traditionally categorized into lift, drag, and pitching moment elements. Most fluid mechanics devices that alter the forces on a body are characterized into two force categories: 1) induced forces resulting from circulation and 2 ) reaction forces caused by jet momentum. This section will focus on lift and drag forces associated with active flow control systems that utilize pneumatic flow control. Pneumatic or blown active flow control systems can be related to boundary layer control (BLC) and/or supercirculation modes. These modes are often characterized by the fluidic power required to achieve the performance augmentation. To achieve the maximum performance on a body, it is desired to drive the stagnation streamlines toward the equivalent inviscid ~ o l u t i o nPractically, .~ this is achieved by moving the boundary layer separation to the TE. This is the performance limit for BLC techniques. To achieve supercirculation it is necessary to extend the effective TE beyond the physical TE location with a virtual or pneumatic flap, as simulated in Fig. 3. To understand the limits of airfoil performance, it in necessary to be aware of the inviscid lift characteristics. The influence of the airfoil thickness on the maximum theoretical inviscid lift coefficient (not including jet thrust or camber effects) can be described as
c,=
= 2 41
+ ):
For a limiting case of t / c of 100% (i.e., circular cylinder), the maximum lift coefficient is 4.rr and can be related to classic unblown circulation around the body lo:
rc
L = pure
(2)
Fig. 3 CFD simulation of pneumatic flap and streamline tuning using a Coanda jet.
196
S. JONES
GREGORY
r,
The magnitude of the circulation is a function of geometry alone and will be referred to as induced lift and can be related to the modified pressure on the integrated boundary of the body:
S,
257
L=-
prsin e d e
(3)
Recall that for an inviscid solution (circular cylinder), the normal force is solely directed in the vertical plane and that drag is zero. As seen in Fig. 4, the streamlines are significantly influenced by the magnitude of the circulation In practice, the inviscid limit is never reached because of flow separation. However, for an airfoil employing a BLC or a CC device, the maximum inviscid lift is possible. When a pneumatic system that adds mass is used, an additional circulation term is added to the induced circulation to account for the reactionary forces produced by the jet, as described in Eq. (4)":
r,.
+
L = ~ u ( r cq e t )
(4)
where
r,,, = !EE ( a + 8jet) PUW
and can be related to lift and drag as
CYLINDER MAPPED INTO AIRFOlL
LE STAGNATION = TE STAGNATION LE STAQNATDN
TE STAGNATION
Fig. 4 Classic lift resulting from circulation for a circular cylinder and mapped into airfoil profile.
PNEUMATIC FLAP PERFORMANCE OF CC AIRFOIL
197
This reactionary force term can affect lift or drag depending on the orientation of the jet exit angle (Sj,,) at the boundary of the body. For pneumatic systems this reactionary force should not be confused with the thrust vectoring that an articulating nozzle generates on an engine nacelle. The reactionary force that is characteristic of a pure jet flap is at a fixed jet angle, as shown in Fig. 5 The efficiency of a pure jet flap (typically vectored normal to surface), compared to typical CC airfoils (vectored tangential to the upper surface), is realized in the differences in the induced effects that accompany the pressure field. It is recognized that both of these airfoil techniques benefit from induced forces and reaction forces that can be correlated to jet position and orientation. Nominally, jet flap airfoils depend largely on the reaction force of the jet momentum. Coanda-type CC systems capture the induced forces more efficiently and typically deliver larger lift gains than a pure jet flap. The combined induced circulation and reactionary forces are generally captured experimentally with a balance, integrated surface pressures, and/or wind-tunnel wall pressure signatures combined with wake rake pressures. The force balance is a direct measure of both induced circulation and reaction forces. Because these forces are integrated and summed at the balance, the ability to decompose the induced and reactionary components are dependent on knowing the vectored force associated with the jet. Integrated surface pressures are representative of induced circulation forces alone. To obtain the total forces along the boundary of the body, reactionary forces must be added at the appropriate S,j angle. The integrated wind-tunnel wall signature and wake rake must also account for the reaction forces generated by the jet. For typical CC systems, the jet exit is nominally directed aft, resulting in a reactionary thrust force that contributes very little to lift (except when an aft camber causes a small S,)j as shown in Fig. 6. It should be recognized that the benefit of turning the flow with the wall bounded jet along the Coanda surface is reflected in the two-dimensional induced circulation found in the modified surface pressure field. The reactionary force of the CC system augments the thrust produced by the primary propulsion system (Fig. 7). Returning a portion of the thrust that was bled from the engine to supply the CC subsystem reduces the overall system penalty associated with CC. The recovery of this thrust will be dependent on the efficiency of the Coanda nozzle and internal losses of the CC air delivery system, and so on.
PURE JET FLAP
Fig. 5 Thrust vectoring using a classic pure jet flap.
198
GREGORY
S. JONES
Fig. 6 Schematic of flow angles associated with typical Coanda-drivenflow.
It is known that nozzle efficiency is very dependent on nozzle aspect ratio (AR). Propulsion system studies of rectangular nozzle losses are generally limited to ARs less than 10. Because there is no database for large-AR nozzles ( h / w > 1300, similar to those used in CC airfoils), it would not be practical to extrapolate to obtain thrust recovery. However, for this two-dimensional study (where nozzle AR is meaningless), it is appropriate to neglect the nozzle efficiency and assume no losses. For two-dimensional CC studies the thrust can be described at the jet exit of the airfoil by the momentum or thrust coefficient:
Fig. 7 Block diagram of reactionary forces for an integrated wing and propulsion system.
PNEUMATIC FLAP PERFORMANCE OF CC AIRFOIL
199
m = pjet U.,J thw
(9)
where
and
The tradeoffs of engine thrust against reduced engine thrust augmented with CC thrust will involve detailed specifications of the geometry of the airfoil, the intake lip, internal diffusers, ducting, compressor, and jet-nozzle designs. Obviously the results would be applicable for that design only. In the absence of these details, some general estimates of the benefits or penalties of CC systems can be formulated by estimating the power requirements of CC. For a crude estimate of fluid power (Pf),it is assumed that the jet is taken from a large reservoir. The total power expended will then be at least equal to the power required to supply the jet velocity head Pjetplus the power lost at the intake as the fluid is drawn into the large reservoir Pram.This ideal power can be described as1*
where 1
m
Pjet N -pu2 2 J P and Pram= mu:
(13)
Hence, the power (ft . lb/s) required supplying a flow with a total momentum coefficient C, is
[+
Pf = c,u” 2uc.a 1
2($)2](qcoucos)
(14)
and nondimensionally
If the jet slot height h is constant and is known for a rectangular wing, the fluid power can be expressed in terms of just the parameters C, and height-to-chord
GREGORY
200
S. JONES
ratio hlc:
Figure 8 shows the nondimensional ideal power for a typical CC jet orifice.
A. Two-Dimensional Drag with Blown Systems Two-dimensional drag characteristics for blown airfoils are often complicated by the juncture flow created by the wind tunnel and airfoil model. To avoid these issues, the most reliable measurement technique for experimentally determining the drag of a blown airfoil is the momentum-loss method that employs a wake rake and is described in detail by Betz13 and Jones.14 The profile drag can be determined by integrating the wake measured one to three chords downstream of the TE
For blown airfoils, it is important to note that the measured profile drag from a wake rake must be corrected by subtracting the momentum that was added by the CC system.” The total horizontal forces on a two-dimensional model do indeed
1.2 -
Power
cw
1.0
-
0.8
-
0.4 0.2 0.6
0.00
0.05
0.10
0.15
0.20
CP
Fig. 8 Ideal power requirements for typical Coanda jets having different jet exit heights.
PNEUMATIC FLAP PERFORMANCE OF CC AIRFOIL
201
exceed that indicated by conventional wake rake calculations by the quantity riz U,. Considering a frictionless hypothetical case where the jet is exhausted at a total head equal to freestream total head easily confirms this principle. Here, the wake will indicate zero drag, but the model will experience a thrust of mu,. The way the net forces are book kept results in
This is equivalent to what a force balance would measure, assuming that the air source is considered to be internal to the model.
B. Equivalent Drag To make direct comparisons of different blown systems such as traditional CC airfoils, jet flaps, blown flaps, engine augmented powered lift systems, and so on, it is necessary to define an equivalent lift-to-drag ratio. For powered airfoil systems, the system efficiency should contain the effects of the energy that is required to obtain the airfoil performance. This also avoids the infinite efficiency that would occur when the drag goes to zero as a result of blowing. A correction can be made through an equivalent “kinetic energy” drag coefficient that is related to the power described previously. This equivalent drag can be described as Dequiv
= Dprofile -k Dpower
+ D r a m + Dinduced
where Dprofile is the profile drag, Dpower is fluid power, Dramis momentum drag force required to ingest the blowing flow rate at the engine inlet, and Dinduced is induced drag (equal to zero for two-dimension). For two dimensional flows, the equivalent drag becomes Dequiv= drag
mq?+ pU, m +2uc.a P
(19)
and
The practical implementation of the Betz and Jones wake integration techniques for blown systems is described in Ref. 18. When the rake drag coefficient is applied to the equivalent drag, it becomes
It should be noted that the kinetic energy or power that is added to the equivalent drag, dominates the equation and leads to drag values that are not practical, and masks the thrust generated by a typical CC airfoil.
GREGORY
202
S. JONES
C. Mass Flow Requirements To optimize the performance of a CC system at the lowest mass flow, it is necessary to recognize the relationships between mass flow, C,, and slot geometry. Figure 9 highlights this relationship for a given freestream condition and geometry, which is consistent with experiments described in this report. Assuming that the performance is dominated by the jet velocity ratio, reducing the slot height would result in a lower mass flow requirement.
IV. GACC Airfoil Design The General Aviation Circulation Control (GACC) wing concept was initially developed for PAV19 and is now being considered for the ESTOL concept described previously. To address the requirements of PAV, the airfoil design and initial performance goals of this wing concept were as follows: 1) To achieve two-dimensional Cl = 3 using a simplified Coanda-driven CC trailing edge. 2) To provide a pneumatic flap capability that will minimize cruise drag and provide potential roll and yaw control (dual blowing is defined as upper and lower Coanda surface blowing). This is based on closing the wake of the bluff TE associated with typical blunt Coanda surfaces. 3) To provide the capability to change the Coanda surface shape (e.g., circular, elliptical, and biconvex). 4) To provide pulsed pneumatic control to minimize the mass flow requirements for high lift. 5 ) To provide distributed flow control to customize the spanwise loading on the airfoil. To establish a relevant CC airfoil geometry that is readily available to the aerodynamic community (not restricte2 due to proprietary issues) and that has the 2bO
200
3
I150 100
50
0
0.0
25.0
50.0
75.0
100.0
125.0
(Ujet/uop Fig. 9 CC mass flow requirements, chord = 9.4 in., q = 10 psf, To = 75°F.
PNEUMATIC FLAP PERFORMANCE OF CC AIRFOIL
203
potential to be modified for the flight applications described above, several geometries were considered. From the late 1950s to the 1970s, NASA was engaged in designing supercritical airfoils for transonic transport and fighter applications. These 6-series supercritical airfoils were developed to improve the cruise performance by increasing the drag rise to Mach numbers that approached 0.8.20 The selection of the airfoil profile for this study was largely driven by the highlift requirements and with a secondary influence of cruise drag requirements. The baseline airfoil shape was initially based on unblown wing performance. Nominally, the thickness ratio has a direct effect on maximum lift, drag, stall characteristics, and structural weight.21 The effect of airfoil thickness on lift and drag are typically counterdemanding and result in tradeoffs. For both unblown and typical CC wings, the thickness ratio primarily affects the maximum lift and stall characteristics by its effect on the nose shape. For a wing of fairly high AR and moderate sweep, a larger nose radius provides a higher stall angle and a greater maximum lift coefficient. However, without blowing or active flow control, the drag increases with increasing thickness due to increased separation. Wing thickness also affects the structural weight of the wing. Statistical equations for wing weight show that the wing structural weight varies approximately inversely with the square root of the thickness ratio. Halving the thickness ratio will increase wing weight by about 41%. The wing is typically 15% of the total empt weight, so halving the thickness would increase empty weight by about 6%!’ Another benefit of a thick airfoil is the increased volume for fuel. The tradeoffs of thickness ratios will not be discussed in this paper, but the larger thickness ratio will be pursued based on the trends of maximum lift and the ability of the CC system to manage the separation issues related to large streamline turning at high-lift conditions. It was therefore desired to combine a typical supercritical section with Coanda-type CC trailing edges. Several key design issues for a CC airfoil are given in the following: 1) A large LE radius is used to alleviate the large negative peak pressure coefficients and can be used as a substitute for a mechanical LE device by delaying LE separation and airfoil stall to high angles of attack. 2) The airfoil was contoured to provide an approximately uniform chordwise load distribution near the design lift coefficient of 0.4. 3) A blunt TE was provided with the upper and lower surface slopes approximately equal to moderate the upper surface boundary layer separation and pressure recovery and thus postpone stall. The NASA LS(1)-0417 airfoil is popularly known as the GA(W)-1 airfoil. for this type of airfoil is approxiTest results for the GA(W)-1 show that Cl mately 30% greater than a typical NACA 6-series airfoil and L I D at Cl = 0.9 was about 50% greater. This 17-percent-thick supercritical airfoilz3 was chosen as a baseline geometry for the GACC airfoilz4 because of its blunt LE, large thickness ratio, and ease of application for active flow control to transonic speeds, as shown in Fig. 10. It is recognized that LE separation will become a problem as the LE stagnation moves aft. For large LE radius airfoils, this
204
GREGORY S. JONES GACC AIRFOIL PROFILE
BLUNT LEADING EDGE RADIUS-1.93%
Fig. 10 Seventeen-percentthick GACC profile with circular trailing edge.
problem occurs beyond the target lift coefficients of 3, so LE control will not be addressed for this study. It was decided to modify the GA(W)-1 with Coanda-type TEs by altering only the aft lower section of the original airfoil. The original GA(W)-1 chord line was used as the reference for angle of attack (AOA) on the GACC airfoil design, as shown in Fig. 10. The tradeoffs of sizing the Coanda surface can be related to optimizing the lift and drag for high lift or cruise conditions.25926Nominally, a larger TE Coanda radius of curvature would lead to a higher CC lift coefficient, as well as a higher cruise drag as a result of an increase in the TE diameter. The shaded area shown in Fig. 11 highlights the region of effective Coanda turning and proven lift performance highlighted by the A-G/CCW flight dem~nstrator.~’ The A-6/CCW airfoil2’ was a 6% thick supercritical wing section that incorporated a state-of-the-art large circular TE radius of 3.67% chord. This large TE functioned to guarantee a successful flight demonstration of the high-lift
-
0.000 0.000
I 0.001
0.002
0.003
0.004
0.005
h/C Fig. 11 Effective Coanda performance for different radius and jet slot heights.
PNEUMATIC FLAP PERFORMANCE OF CC AIRFOIL
205
system2* only. Any operational use of this design would require a mechanical retraction of the CC system into the wing to avoid a large cruise drag penalty. To minimize the GACC airfoil drag performance without the use of a mechanical system a dual-blowing pneumatic concept with a small radius TE was designed. A baseline circular r / c of 2% was chosen for the GACC. Three different TE shapes were designed to be interchangeable and integrate with the GACC model, as shown in Fig 12.The distance between the slots remained fixed and used the circular shape as a baseline. Both the elliptic and biconvex shapes extended the chord by 1% (0.174in.). The 2:l elliptic shape reduced the Y/C to 1% and the biconvex shape had an Y/C of 0. To compare steady, pulsed, and dual blowing using a common model required careful design of the internal flow path, as shown in Fig. 13. The ability to independently control the upper and lower slot flow enables the investigation of both positive and negative lift as well as drag and thrust for both high-lift and cruise conditions. A pulsed actuator system was integrated into the upper plenum of the model for investigation of unsteady circulation control. To obtain a uniform flow path and create a two-dimensional flow environment at the Coanda surface it was necessary to carefully design the internal flow path of all three air sources in the model, as shown in Fig. 14.Twenty actuators were distributed in the upper plenum along the span to optimize the pulsed authority to the
VARIABLE UPPER SLOT
2:l BI-CONVEX
Fig. 12 Sketch of three interchangeable TE shapes for the GACC airfoil.
GREGORY
206 UPPER STEADY MANIFOLD
S. JONES
ACTUATOR DIFFUSER
UPPER SLOT
LOWER STEADY MANIFOLD
Fig. 13 Sketch of internal flow path of the GACC airfoil.
upper Coanda jet for the high-lift mode. Air for all three sources was fed from one end of the model and was expanded into large plenums then channeled to the trailing edge jet exit. Both the upper and lower slots were adjustable (0.005 < h < 0.025) and were fed from a smooth contraction that had a minimum area ratio of 10. It is difficult to create an infinite or two-dimensional environment with a fixed-wall wind tunnel for blown airfoil systems. One must consider the relative size of the model to the size of the test section and the expected trajectory of the jet created by the blown system. To minimize the impact of the windtunnel interference for CC systems, several experimental design considerations were considered: Solid Blockage (physical chord and span related to windtunnel cross-section), Wake Blockage (how much streamline turning will be
20 ACTUATORS wlDlFFUSERS
INSTRUMENTED TRAILING EDGE COANDA SURFACE
Fig. 14 Sketch of GACC model with upper skin removed to highlight the flow path and instrumentation of the upper plenum.
PNEUMATIC FLAP PERFORMANCE OF CC AIRFOIL
207
achieved with blown system), and Juncture flow regions (aspect ratio of model). The GACC model was sized and built for the NASA Largley Research Center (LaRC) Basic Aerodynamic Research Tunnel (BART) and had a chord-to-testsection height ratio of 0.23, an aspect ratio of 3 based on a chord of 9.4 in. and a two-dimensional wall-to-wall span of 28 in. These values are conservative for the unblown c~nfiguration,~~ however, once blowing is applied, the influence of the Coanda jet on streamline turning could be significant. A two-dimensional RANS code (FUN2D) was used to evaluate the streamline turning related to Coanda blowing and supercirculation high-lift condition^.^' The free air results of this preliminary CFD evaluation indicated streamline turning and wake deflection would not impact the tunnel walls for the BART test conditions but would be influenced by the presence of the solid tunnel walls. The study of wall interference is ongoing for this experiment. V. Experimental Setup Experimental results have been obtained for a GACC airfoil in the open return Langley BART, as seen in Fig. 15. The tests were conducted over a Mach number range of 0.082 to 0.1 16 corresponding to dynamic pressures of 10 and 20 psf, respectively. Lift, drag, pitching moment, yawing moment, and rolling moment measurements were obtained from a five-component strain gauge balance. Drag data were also obtained from a wake rake. Airfoil surface pressure measurements (steady and unsteady) were used to highlight boundary layer transition and separation. A block diagram of the BART data acquisition is shown in Fig. 16. To capture the transients and time-dependent characteristics of the pulsed flowfield two approaches were developed: arrayed thin films and miniature pressure transducers. This report will focus only on the miniature pressure transducers. The small scale of the model did not lend itself to using off-the-shelf pressure transducers. Custom differential pressure gauges were designed and fabricated using GACC CHORD 9.4"
Fig. 15 Sketch of the GACC setup in the Basic Aerodynamic Research Tunnel.
GREGORY
208
S. JONES
UNSTEADY P+p’
Fig. 16 Block diagram of BART data acquisition for GACC setup.
MEMS sensors attached directly to the skins of the model leading and trailing edges. These transducers were not temperature compensated, making real-time calibration necessary. To keep the measured errors from exceeding 0.05% of the full scale (2 psid), a reference pressure was monitored and calibrations were performed when necessary. This was also the case with the ESP system for ten independent 32-port modules with ranges of 10 in. H20,1 psid, and 2.5 psid. The five-component strain gauge balance was also custom designed and fabricated for the GACC model. Normal, axial, pitching moment (ref. 50% chord), rolling moment, and yawing moment limits are shown in Table 1. A drawback to the GACC balance was that the axial resonance of the balance/model system was too close to the dynamics of the loaded airfoil, resulting in vibration of the model.
Table 1 GACC balance limits Normal, lbf
Axial, lbf
Pitching moment, in. lbf
Rolling moment, in. lbf
Yawing moment, in. lbf
100
10
1600
400
40
PNEUMATIC FLAP PERFORMANCE OF CC AIRFOIL
209
This vibration did not always exist, but led to larger than expected errors in the axial force measurement. Therefore the drag data will be reported only from the wake rake results. The GACC model has three plenums, which are required for use in different modes of operations (e.g., high-lift, cruise, pulsed, and so on). Each plenum is supplied with air that is independently regulated, as shown in Fig. 17. To achieve the potential mass flow requirements for the largest slot area, a 2000 psia high-pressure external air source (3000 psia max) was used. The air is preheated to compensate for Joule Thompson effects and temperatures are maintained to within 1" R. The mass flow was measured with three independent turbine meters. These flow meters are precalibrated and compensated for density variation at the point of measurement (accuracy = 1% reading). The high-pressure plenum that supplies the pulsed actuation system is buffered with a 7.1 ft3 air tank to eliminate the pulsed backpressure flow at the control and flow measurement station. The pressure limits of each of these systems were driven by the pressure ratio at the slot exit. As a result of pressure losses in the system the upper and lower plenums were limited to 50 psid and the actuator pressure was limited to 200 psid. These limits enabled sonic capability at the slot exit. A trapeze system was used to couple the air delivery system to the model as shown in Fig. 18. Special attention was given to the calibration of the balance due to the number of airlines that cross the balance. Unpressurized calibration results are applied to a 6 x 21 calibration matrix and account for the linear interactions (first order) and the second-degree nonlinear interactions of the balance.30931 Each pressure line was then independently loaded and characterized with no flow (see appendix to this chapter). With the model mounted vertically in the tunnel, the only loads experienced by the model as a result of the air delivery system were thrust loads along the span of the model. This is the same as the side-force, which is not gauged or measured. The flexible hoses maintain a vertical orientation to the model and eliminate horizontal forces being applied to the balance.
VOLUME BOOSTER
Fig. 17 GACC air delivery system having three independent air supply lines.
210
GREGORY
S. JONES
LOWER JET AIR SUPPLY
Fig. 18 GACC balance and model interface with air delivery through trapeze system.
Measurement of the drag was initially obtained with the balance and reported in Ref. 19. However, upon careful inspection of the issues related to juncture flow interference and balance vibration, it was determined that the drag information from the balance was unreliable. A total head wake rake was designed and fabricated for the BART. The streamwise location of the rake was determined based on a balance of streamline turning (flow angle at the rake face) and the sensitivity of the pressure transducers. CFD and wind-tunnel wall-pressure signatures were used to identify that the jet wake was aligned with the freestream streamlines at x / c greater than 3.5 from the TE of the model. An example of the wall-pressure signature is shown in Fig. 19 for typical high-lift conditions. The magnitude of the wall-pressure signatures shown in Fig. 19 indicates that a correction may be warranted for the dynamic pressure and angle of attack.
PNEUMATIC FLAP PERFORMANCE OF CC AIRFOIL
21 1
-0.50
-0.25
ACp O.O0
0.25
0.50 -4.0 -3.0 -2.0 -1.0 0.0
1.0
2.0
3.0
4.0 5.0
XIC (LE REF)
Fig. 19 Wind-tunnel wall-pressure signatures for different lift coefficients (solid symbols for upper wall, open symbols for lower wall), h = 0.020 in., q = 10 psf, circular trailing edge.
Several wall correction techniques are described in the 1998 AGARD “Wind Tunnel Wall Corrections” report.32Corrections of two-dimensional experiments for wall effects are compounded by the two-dimensional AR and the juncture flow of the model and wind-tunnel wall interface. As a first approximation of the wall interference characteristics, corrections for two-dimensional lift interference are made using a classic approach described in the appendix. It is recognized that these corrections are inadequate and that the wall signature method may be more appropriate. evaluation^^^ of the wall-signature method are ongoing and are not applied to the data presented in this report. The wall-signature pressure distribution is also used to locate the streamwise wake rake position for this experiment. The criteria for the rake measurements are based on a tradeoff of transducer sensitivity and flow angularity of the flow at the probe tip. Based on these criteria, the wake rake was located 3.6 chords downstream of the TE of the model at an AOA of Odeg. The wake profiles shown in Fig. 20 are representative of the effectiveness of the streamline turning created by the circular CC airfoil configuration. The errors associated with the integration of the wake to determine measured drag are related to the nonzero pressures outside the wake region. Although the rake spans the entire test section, only 86% is used for the wake integration, thus eliminating the influence of the floor and ceiling boundary layers. The measured drag was determined to have a repeatability of Cd = f0.0005. For the momentum sweep at AOA = 0, the wake moved approximately one chord below the centerline. An example of an AOA sweep at a fixed blowing rate is shown in Fig. 21. The wake moved approximately 1.5 chords below the centerline prior to stalling.
GREGORY
212
S. JONES
0.0125
0.0075
0.0025
CP -0.0025
-0.0075
I
NO BLOWING I
-0.0125 I -2
-1
0 1 U C (WAKE POSITION)
2
Fig. 20 Wake profile of GACC with circular trailing edge, AOA = 0, x/c = 4.64.
The measurement of the nondimensional momentum coefficient can be obtained from parameters described in Eq. (8). Using mass flow and measured pressure ratios to obtain Ujet, the momentum coefficient can be calculated without any knowledge of slot height. This is the preferred method because of the potential errors in measuring the slot height of the small-scale model used
AOA
-- -- -
--10.0 -6.0
-
10.0
CP
-0.10
-0.15
-0.20 -2
*C,o,,
-1
0
1
2
U C (WAKE POSITION)
Fig. 21 Wake profile of GACC with circular trailing edge, -10 c AOA c 10, C, = 0.075, x/c = 4.64.
PNEUMATIC FLAP PERFORMANCE OF CC AIRFOIL
213
in this test. However, post-test evaluation of the mass flow data revealed problems with the turbine meters that can be associated with the turbine meters being located in high-pressure legs of the flow path. This resulted in the use of slot height to determine the momentum coefficient. Slot height is a critical parameter for correlation to airfoil performance and was given careful attention. Nominally, the slot height was set with a digital height gauge (accuracy = 0.0001 in.) under no flow conditions. The height was then readjusted to obtain a uniform velocity along the span of the slot. The slot height was locked into place with a push-pull set of screws located approximately 1 in. from the slot exit inside the settling region of the jet plenum. The 0.010 in. TE of the stainless steel skin was observed under load with a microtelescope and did not appear to move. However, post-test span wise jet velocities measured at the slot exit with a hot wire probe, shown in Fig. 22, indicate variations of 20% relative to the reference jet velocity determined from pressure ratio. Most of these variations can be identified with the wake of the internal push-pull screws used for setting slot height. The variations of the low jet velocities are larger than the higher jet velocities. It was also discovered that the extreme inboard and outboard slot velocity (not shown) was significantly lower than the core region of the span. This is attributed to internal flow separation at the inlet and exit of the flow manifold internal to the model. Although affecting only the extreme 0.5 in. sections of the span, it does effectively reduce the length of the blowing section of the jet. The large-scale span-wise variation is thought to be due to internal flow variations and/or errors in setting the slot height under loaded conditions. Setting the final slot height was done on site with the model mounted in the tunnel and mass flow being added. The confined space of the small wind tunnel made setting the slot height difficult because of issue of accessibility and noise. Pressurizing the
0
0.2
0.4
0.6
0.8
1
SPANISPAN,,,
Fig. 22 Example of spanwise velocity deviation for different jet exit Mach numbers (biconvex TE configuration, h = 0.020 in.).
214
GREGORY
S. JONES
model for maximum conditions created a jet noise and flow environment that was uncomfortable for the operator setting the slot height. Therefore, a low jet velocity was chosen for the slot height adjustment process. As seen in Fig. 22, there is a large scatter in the low-speed jet data. This gives rise to a greater sensitivity and data scatter to the location of the measurement while setting the slot height. To compound this problem, a hand-held 0.010 in. OD (outside diameter) flattened pitot-probe sized to fit just inside the slot was used to make the spanwise velocity profile of the jet exit. The errors in probe location and angularity led to additional data scatter, which contributed to the errors in setting slot height. A post-test average slot height was determined using two methods: 1) a direct velocity profile and 2) a conservation of mass method. During the post-test evaluation of the spanwise velocity distribution, it was discovered that the large-scale Mach number variation along the span was consistent from low to high Mach numbers. Post-test hot-wire measurements of the slot jet profile for the biconvex configuration are shown in Fig. 23. The slot height was nominally set to 0.020 in. Normalizing these profiles with the velocity measured via the pressure ratio used throughout the experiment revealed that the hot-wire maximum velocity results were 20% higher, as shown in Fig. 24. This is consistent with the span location chosen for the velocity profiles. The conservation of mass method for determining slot height utilizes the integrated jet velocity determined from the pressure ratio and the measured mass flow:
Each TE configuration had two targeted slot heights to be tested, h,,, = 0.010 in. and h,,, = 0.020 in. Post-test analysis revealed that the slot heights were 5-30%
Fig. 23 Example hot-wire velocity profiles at the slot exit plane. Measurements are between adjustment screws at Span/Span,,, = 0.1 (biconvex TE configuration, h = 0.020 in.).
PNEUMATIC FLAP PERFORMANCE OF CC AIRFOIL
215
2.000
1.500
0.000
0.2
0
0.4
0.6
0.8
1
1.2
1.4
UJET-HW/"JET-REF
Fig. 24 Normalized velocity profiles at the upper surface exit plane of the biconvex TE.
higher than was thought at the time of setup, as shown in Fig. 25 for the circular TE. The calculated slot height also varied up to 18% with increasing nozzle pressure ratio. An average of slot height for the varying mass flow was used for reporting purposes. Extrapolating the biconvex calculated profile to the unblown condition results in a 0.021 in. setup. This is consistent with the slot height measured in the post-test slot profile hot-wire measurements shown in Fig. 24.
c
1.oo
1.10
1.20
1.30
NPR
Fig. 25 Slot height variations as internal plenum pressure increases for circular TE.
Next Page GREGORY
216
S. JONES
VI. Airfoil Performance Airfoil performance will be discussed for two modes of the GACC airfoil: the high lift mode with upper slot blowing and the cruise mode with upper and lower slot (dual) blowing. The efficiency of pulsed blowing will be discussed as part of the high-lift mode.
A. High Lift Mode 1. Baseline (No Blowing) Lift, drag, and pitching moment will be used to establish the two-dimensional baseline performance of the GACC airfoil with different TEs. The original GACC airfoil was designed around the circular TE having an r / c of 2%. Therefore, the circular TE will be used as the reference for the elliptic and biconvex trailing edges. Comparing the lift performance of the three TEs with no blowing in Fig. 26, the circular TE has a lift enhancement of ACl = 0.16 at a zero degree AOA relative to the biconvex and elliptic TEs. This is also reflected in the TE pressures shown in Fig. 27. Comparisons of the drag performance for the three TE are shown in Fig. 28. There are few differences in the indicated drag. This can be related to boundary layer transition fixed at 5% chord and the fixed trailing height established by the steps created by the upper and lower slots. Minimum drag occurs at zero lift and AOA = - 6. The airfoil efficiency shown in Fig. 29 indicates that the circular TE is more efficient than the elliptic or biconvex TEs with no blowing. The peak efficiency occurs at AOA = 6 deg and is consistent with the differences in lift. The
2.0
1.5 1.o
C, 0.5
0.0 -0.5 -1 .o -20
-1 0
0
10
AOA
Fig. 26 Baseline lift coefficient with no blowing (balance data).
20
Previous Page PNEUMATIC FLAP PERFORMANCE OF CC AIRFOIL
217
-5.0 -4.0
3.0 -2.0
CP -1
.o
0.0
1.o 2.0
0.00
0.25
0.50
0.75
1.oo
XIC
Fig. 27 Pressure distribution for GACC airfoil, no blowing, AOA = 0.
drag polar shown in Fig. 30 illustrates a relatively flat drag characteristic for the region of lift, which is consistent with cruise conditions (e.g., Cl = 0.5).
2. Circular TE The circular Coanda TE will be used as a reference for comparisons of performance throughout the rest of this paper. This section will highlight the circular TE performance for high-lift conditions. Although somewhat arbitrary, the initial goal of this effort was to generate a lift coefficient of 3 at an AOA 0.10 0.08 0.06 CD
0.04 0.02 0.00 -20
-10
0
10
20
AOA
Fig. 28 Baseline drag coefficient with no blowing (wake rake).
218
GREGORY
S. JONES
-1 0
0 AOA
50
40
30 20
UD 10 0
-1 0 -20 -20
10
20
Fig. 29 Baseline GACC airfoil efficiency with no blowing.
of 0 deg. Figure 31 illustrates that using upper Coanda blowing, the target lift coefficient of 3.0 was achieved. The maximum lift that this airfoil can achieve is still undetermined, but will be limited by the LE performance of the airfoil for a given blowing condition. As the blowing is increased, the LE stagnation point moves aft on the lower surface, creating a condition conducive for boundary layer separation and LE stall. The maximum lift and stall characteristics of this CC airfoil are reduced with increasing C,, as highlighted in Fig. 31. These data are consistent with other supercritical CC airfoils with large LEs. Lower Coanda blowing gives this airfoil configuration a unique ability to manage lift and drag by generating a negative lift capability. The combination 0.10 0.08 0.06 CD
0.04 0.02 0.00 -1.0
-0.5
0.0
0.5
1.o
1.5
2.0
Cl
Fig. 30 Baseline drag polar for GACC airfoil with no blowing.
PNEUMATIC FLAP PERFORMANCE OF CC AIRFOIL
219
4.0
3.0 2.0 CL
1.o 0.0
-1 .o -20
-1 0
0 AOA
10
20
Fig. 31 Airfoil lift performance with circular TE and hlc = 0.0022 (open symbols represent lower blowing).
of upper and lower blowing with variations in AOA enables the designer to customize lift and drag for either approach or takeoff conditions. The open symbols shown in Fig. 31 highlight the lower Coanda blowing. The pneumatic flap effect of lower blowing compensates for the TE camber as demonstrated by zero lift at AOA = 0 (Cp,o,,= 0.024). These effects are more related to cruise drag and will be discussed later in this chapter. The efficiency of the Coanda blowing can be related to the slot height and the radius of the Coanda surface. For a fixed Coanda surface radius of Y/C = 2%, an h/C of 1.4% performed better than an h / C of 2.2%, as shown in Fig. 32. 3.5 3.0
∆CL = 60.3 ∆Cµ
2.5 2.0 C Cll
∆CL = 45.3 ∆Cµ
1.5 1.0 1.o
h/C 0.0014 0.0022
0.5
BOUNDARY LAYER CONTROL
0.0 0.00 0.00
0.02
SUPERCIRCULATION CONTROL
0.04
0.06
0.08 0.08
Cµ CP
Fig. 32 Lift performance of circular TE, AOA = 0.
0.10
GREGORY
220
S. JONES
The lift augmentation for the small slot was 60.3 in the separation control regime compared to the 45.3 augmentation for the larger slot. To extend into the supercirculation regime it is necessary to push the rear stagnation beyond the physical TE, forming a pneumatic flap. A shift in the lift augmentation efficiency highlights this effect, as shown in Fig. 32. The limit of the separation region for this airfoil occurs at a C, of approximately 0.03 and a lift coefficient of 1.8. To predict the mass flow requirements and lift performance in the supercirculation region, it is possible to extend the supercirculation lift augmentation line. The drag characteristics corresponding to Eq. (18) are shown in Fig. 33. Thrust is generated for the low blowing rates that are characteristic of most CC airfoils including GACC. Combinations of Coanda blowing and AOA allow for variable drag at a fixed lift condition. As an example, the drag can be varied by ACd = 0.060 at a lift coefficient of 2.0. This would include both a thrust and drag capability. The limitations of this capability are related to the LE stall characteristics and may be augmented with LE active flow control. To gain a greater understanding of drag characteristics for this airfoil, the total drag measured in the wake can be decomposed into a two-dimensional circulation induced force represented by the pressure distribution on the airfoil (shown in Fig. 34) and the reactionary force created by the Coanda jet evaluated at the jet exit. The reactionary force and the induced force can be combined to create the total force measured. Because the total drag force is known from the wake rake data and the reactionary force C, is equivalent to C ,, then the twodimensional circulation induced force will become
0.10
0.05
-0.05
-0.1 0
-1 .o
0.0
1.o
2.0
3.0
4.0
Cl
Fig. 33 Airfoil drag polar for circular TE, hlc = 0.0022, wake rake data (open symbols represent lower blowing).
PNEUMATIC FLAP PERFORMANCE OF CC AIRFOIL
221
WC
b)
COANDA SURFACE -(I(Degrees)
Fig. 34 GACC pressure distribution with circular TE, AOA = 0,h / c = 0.00106: a) airfoil pressure distribution; b) expanded view of circular TE pressure distribution.
An example of the two-dimensional circulation induced drag force is shown in Fig. 35. These data corresponds to the lift data in Fig. 32. It can be observed that the slope change related to the supercirculation region in the lift data is also evident in the drag data, occurring at a momentum coefficient of approximately 0.03. The efficiency of a blown airfoil has traditionally been related to an equivalent drag as described earlier in the text. The equivalent drag shown in Fig. 36 highlights the conversion of measured thrust to equivalent drag for two slot configurations. Although this enables the comparison of one blown system with another, it is dangerous for the designer to use these values, as seen by comparing Figs. 35 and 36. The efficiency of the airfoil can be represented by
GREGORY
222
0.00
0.02
S. JONES
0.04
0.06
0.08
0.10
CP Fig. 35 Drag performance of circular TE, AOA = 0.
the lift-to-equivalent-drag ratio shown in Fig. 37. Comparison of the two slot configurations indicates a greater efficiency of the larger slot. This is a result of the drag benefits of the larger slot and is related to the turbulence characteristics of the Coanda jet. The peak efficiency occurs in the vicinity of the transition from boundary layer control to supercirculation (refer to Fig. 32). The two-dimensional LID equivalent efficiency of the airfoil can also be related to the fluidic power required of the high-lift system, as shown in Fig. 38. The corresponding equivalent drag data are shown in Fig. 39. The fluidic power can be related to the reactionary thrust component described in Fig. 35. The dashed line in Fig. 35 represents the contribution of the fluidic
'd
EQUlV
CP Fig. 36 Equivalent drag of circular TE, AOA = 0.
PNEUMATIC FLAP PERFORMANCE OF CC AIRFOIL
223
UD EQUlV
0.00
0.02
0.04
0.06
0.08
0.10
CP Fig. 37 Efficiency of circular TE, A O A = 0.
power to the equivalent drag. Any values that deviate above or below this line can be related to the two-dimensional circulation induced effects described above and highlight the magnitude of the dominating contribution of the fluidic power to the equivalent drag. Evaluating the measured drag per fluidic power reveals that the most efficient use of the fluidic power occurs in the boundary control region. This is shown in Fig. 40, where ACd/Cp,is a minimum. The magnitude of the incremental thrust
LID EQUIV
0.00
0.10
0.20
0.30
CPf (FLUIDIC POWER)
Fig. 38 Pumping power required to achieve equivalent G A C C airfoil efficiency for circular TE, AOA = 0.
GREGORY
224
S. JONES
‘d
EQUlV
0.00
0.10
0.30
0.20
0.40
Cpf (FLUIDIC POWER)
Fig. 39 Fluidic power required to achieve equivalent drag for circular TE, AOA = 0.
for the larger slot height is 0.9324 at a fluidic power of 0.03873 shown in Fig. 41. This corresponds to a thrust of 0.0295 (see Fig. 35). This also illustrates a benefit of a blown system compared to other active flow control techniques such as syntheticjets and suction systems. Without the benefit of the reactionary force of the jet, the best performance a traditional active flow control system could achieve would be related to moving or attaching the
0.00
0.05
0.10
0.15
0.20
0.25
Cpf (FLUIDIC POWER)
Fig. 40 Drag efficiency per fluidic power for GACC airfoil with circular TE, AOA = 0.
PNEUMATIC FLAP PERFORMANCE OF CC AIRFOIL
225
"d
Fig. 41 Drag per power ratio for GACC airfoil with circular TE, AOA = 0.
boundary layer to the most aft portion of the airfoil. This would result in a minimum drag associated with skin friction alone. For a tangentially blown system typical of CC airfoils, the reactionary forces enable penetration into the outer flowfield that is not available to unblown systems. To make a direct comparison of these different active flow control systems it would be necessary to equate the relevant power (watts, horsepower, and so on) to achieve a comparable drag performance. Another performance parameter of interest is the lift-increment-per-power ratio, ACl/Cp,shown in Fig. 42. This parameter is occasionally used for direct
0
1
ACI
2
3
Fig. 42 Lift per power ratio for GACC airfoil with circular TE, AOA = 0.
S. JONES
GREGORY
226
Table 2 Comparison of GACC lift increment-per-power to similar powered systemsI2 Item GACC (h/c = 0.0014) Elliptic CC43 TE blown flap3' Flap knee44 (BLC mode)
ACL/Cp, (ACi = 0.5)
ACLICp, (AC1 =1.0)
44.3 40.4 42.6 26.8
31 28.6 33.2 7.48
comparisons of similar power-augmented devices. The comparisons are made at ACl values of 0.5 and 1.0, which are consistent with the boundary control region, and the initial stage of supercirculation. For the GACC airfoil, the smaller slot develops more lift for a given power setting than the larger slot in the boundary layer control region. As the power (or momentum) is increased into the supercirculation region, the influence of slot height on lift-to-power augmentation decreases. Comparisons of the power requirements for the GACC and other similar airfoils are shown in Table 2. The GACC airfoil performance is comparable to that of a similar CC airfoil and blown flaps with active flow control. The pitching moment characteristics of the GACC airfoil are shown in Fig. 43. These values are consistent with other CC airfoils. 3. Per$ormance Comparisons of TE The following section will focus on comparisons of the different shape TEs with a fixed slot height of h / c = 0.0022. The shapes include circular, elliptic, and biconvex profiles, having effective TE rad of r / c = 2, 1, and 0%,
-1
.o
0.0
1.o
2.0
3.0
4.0
Cl
Fig. 43 Twenty-five percent chord pitching moment characteristics of GACC, hlc = 0.0022.
PNEUMATIC FLAP PERFORMANCE OF CC AIRFOIL
0.00
0.02
0.04
0.06
0.08
227
0.10
CP Fig. 44 Comparison of lift performance for the GACC airfoil for different TE shapes, hlc = 0.0022.
respectively. The lift performance of the larger radius configuration is higher than the other configurations, as seen in Fig. 44. A comparison of the drag performance, shown in Fig. 45, highlights the improvement of the drag as a function of the smaller r / c . The elliptic TE ( r / c = 1%)has less drag than the circular TE ( r / c = 2%) throughout the boundary layer and supercirculation region. Transitioning from the boundary layer region to the supercirculation region, the total thrust of the elliptic TE exceeds
0.00
0.02
0.04
0.06
0.08
0.10
CF Fig. 45 Comparison of the thrust performance of the GACC having three different TE shapes.
GREGORY
228
S. JONES
the reactionary thrust, implying a net two-dimensional circulation induced thrust. The drag performance of the biconvex shape mimics the circular TE performance in the BLC region. The thrust for the biconvex configuration extends beyond the reactionary thrust throughout the supercirculation region. Comparisons of drag polars for the three different TEs are shown in Fig. 46. The effectiveness of the sharp TE is reflected in the increased thrust for the biconvex TE. Comparisons of pitching moments for the three TEs are shown in Fig. 47. The biconvex TE has the lowest pitching moment for any given lift. The benefits of high thrust and low pitching moment come at the price of momentum coefficient; for example, for a lift coefficient of 2, the thrust of the biconvex is 110 counts larger and the moment is 50 counts smaller than the circular TE performance. However, the momentum coefficient increased by a factor of 2.
B. Cruise Configuration To address the issue of a blunt TE for typical CC configurations at cruise, the GACC was designed with a dual blowing capability, that is, upper and/or lower blowing on the Coanda This enables the operator to augment the system thrust while providing roll and/or yaw control. The following section will address only the dual-blown circular TE performance. 1. Dual Blowing for Circular Coanda Sur$ace It should be recognized that the cruise condition for this airfoil would be operated at a substantially higher Mach number and higher dynamic pressure, thereby reducing the momentum coefficient. These low-speed data do not account for the airfoil compressibility and potential shock manipulation that typical CC configurations may provide. For cruise conditions, the CC performance characteristics are
0.00
0.50
1.00
1.50
2.00
2.50
3.00
Cl
Fig. 46 Comparison of drag polars for three different TE shapes, h / c = 0.0022.
PNEUMATIC FLAP PERFORMANCE OF CC AIRFOIL
0.00
0.50
1.00
1.50
2.00
229
3.00
2.50
Cl
Fig. 47 Comparison of pitching moments (referenced to 50% chord) for three different TE shapes, h / c = 0.0022.
limited to the boundary layer control region. Nominally, lift coefficients that are of the order 0.5 are desired during cruise operations. To characterize the lift performance of the dual-blown configuration of the GACC airfoil, the upper blowing condition was fixed and the lower blowing was swept, as shown in Fig. 48. As expected, the upper blowing performance remains proportional to the lift. Combining this upper blowing with lower blowing will result in a lift reduction. However, this reduction does not occur until the initial stages of thrust.
0.001
0.010
0.100
1.000
CI+UPPER + LOWER)
Fig. 48 Lift performance for dual blowing, h / c = 0.0022.
GREGORY
230
0.001
S. JONES
0.010
0.100
1.000
CP(UPPER t LOWER)
Fig. 49 Drag characteristicsof the circular dual blown configuration, h / c = 0.0022.
The effectiveness of the dual blown configuration is realized in the drag performance. The drag characteristics associated with Fig. 48 are shown in Fig. 49. The drag performance seems to be independent of upper blowing in the boundary layer control region. The drag polar, shown in Fig. 50, indicates that thrust can be adjusted for a given lift (e.g., for a fixed Cl = 0.5, a ACd = -0.043 can be adjusted using dual blowing).
-0.5
0.0
1.o
0.5
1.5
2.0
Cl
Fig. 50 Drag polar for the dual blowing cruise configuration of the GACC airfoil, circular TE, h / c = 0.0022 (upper and lower).
PNEUMATIC FLAP PERFORMANCE OF CC AIRFOIL
231
0.20 0.15 0.10 0.05
cp
0.00 -0.05 -0.10 -0.15 -0.20
-2
-1
0
1
2
WAKE RAKE POSITION (UC)
Fig. 51 Wake profiles for the dual blowing cruise configuration of the GACC airfoil, circular TE, reference Cpupper = 0.003, h / c = 0.0022 (upper and lower).
The wake profile shown in Fig. 51 corresponds to the fixed upper blowing of C, = 0.003. As the lower blowing rate increases, the profile goes from a single
peak to a double peak, then returns to a single peak. This indicates that the upper and lower jets are independent and do not mix efficiently for the blunt circular TE. The equivalent drag for the circular dual-blown configuration is shown in Fig. 52. The minimum equivalent drag occurs at a combined momentum
CD EQUlV
0.000
0.025
0.050 0.075 CI+UPPER t LOWER)
0.100
Fig. 52 Equivalent drag for the GACC dual blown circular TE.
232
GREGORY
S. JONES
0.025
0.050
UD EQUlV
0.000
CP(UPPER
0.075
0.100
+ LOWER)
Fig. 53 Airfoil efficiency for the GACC dual blown circular TE.
coefficient of 0.03 and a fixed upper momentum coefficient of 0.003. This is consistent with a measured total drag of -0.012. The peak efficiency, shown in Fig. 53, occurs at a total momentum coefficient of 0.021. This is consistent with the measured drag transitioning from drag to thrust. 2. Pulsed Blowing As will be shown in this section, pulsed blowing from the upper slot is intended to reduce the mass flow requirements for a com arable steady blowing performance?6937 The GACC pulsed blowing systemZBis based on a high-speed valve that delivers a high volumetric flow to the upper jet exit. The actuator is close coupled (internally located x/c = 0.90) to the jet exit through a rapid diffuser to deliver a pulse of air that can be varied in magnitude, frequency, and duty cycle. An example of the pulse train is shown in Fig. 54. The quality of the rise time and decay of the pulse train is related to the overall actuator authority. The rise and decay time of the pulse train is dependent on the internal volume located internally just upstream of the jet exit. This includes the 10:1 contraction and the settling area downstream of the rapid diffuser exits. The time-dependent pulse train is referenced to the jet exit or = 0 deg of the Coanda surface. The averaged pressure field is compared to a comparable steady blowing condition, shown in Fig. 55. The separation associated with this condition was identified to occur in the range 75 < < 90 deg, whereas steady blowing was in because 60 < < 75 deg. This corresponds to the lift performance shown in Fig. 56. The mass flow reduction of 55% corresponds to the 40% duty cycle shown in Fig. 54. It should be emphasized that this reduction is limited to the BLC region because of current limits in actuator authority. The turbulence magnitude and frequency of the steady jet increases just downstream of the jet exit, then increases along the Coanda surface to peak at
+
+
+
PNEUMATIC FLAP PERFORMANCE OF CC AIRFOIL
233
Time (sec)
Fig. 54 Time record of circular Coanda surface pressures with pulsed upper blowing, 35 Hz, 40% duty cycle, circular TE, h / c = 0.00106.
6 = 30 deg, shown in Fig. 57. The magnitude and frequency then decays until the jet separates from the Coanda surface in the range 60 < 6 < 75 deg. For the pulsed jet, the turbulence magnitude and frequency of the jet-on portion of the pulse train increases just downstream of the jet exit, then increases along the Coanda surface to peak at 6 = 60 deg, as shown in Fig. 58. The magnitude and frequency then decay until the jet separates from the Coanda surface in the case 75 < 4 < 90 deg.
4 PEG) Fig. 55 Comparison of steady and pulsed pressure distribution for the circular TE, h / c = 0.00106.
GREGORY
234
S. JONES
2.0 DC=80%
DC=6O%
I
\
DC=40% DC=20%
c,
1.0
0.5 0.0
1 I 0
55%
I
STEADY I
0.005
I
I
0.01
0.015
I
0.02
0.025
CP
Fig. 56 Comparison of lift performance for steady and pulsed blowing on the circular TE, h / c = 0.00106.
Fig. 57 Frequency content of the pressure field on Coanda surface, steady jet, circular TE, h / c = 0.00106: a) nondimensional spectra for steady jet; b) expanded view of frequency content for the influence of the shear and entrained flow.
PNEUMATIC FLAP PERFORMANCE OF CC AIRFOIL
0.01
8
b)
0.10
1.00
10.00
10
12 14 16 FUU (LREF:TE DIAMETER)
235
100.00
18
Fig. 58 Frequency content of the pressure field on Coanda surface for the pulsed jet, actuator drive; 35 Hz, 40% duty cycle, circular TE, h / c = 0.00106: a) nondimensional spectra for pulsed jet; b) expanded view of frequency content for the pulse-on portion of pulse train.
DC=40%
DC=S?%
1.5
c, 1.0 0.5
0.000
0.005
0.010
0.015
0.020
0.025
Ck
Fig. 59 Mass flow reduction for pulsed elliptic TE, h / c = 0.0022, BLC region.
236
GREGORY
S. JONES
The performance benefit of the pulsed elliptic TE is significantly less than that of the circular TE, shown in Fig. 59. For a lift coefficient of 1.0 there is a 29% reduction of mass flow for the pulsed elliptic TE compared to the 55% reduction of the circular TE. There was no measurable benefit in mass flow reduction for the pulsed biconvex TE. The effectiveness of the pulsed blowing can be related to radius of curvature of the Coanda surface and jet separation. The pulsed effectiveness for larger Y/C that is represented by the 2% circular TE, moved the time-averaged separation beyond the maximum TE location of x/c = 1.0, that is, from the upper Coanda surface to the lower Coanda surface. Several factors contribute to the effectiveness of the pulsed jet, including a larger instantaneous velocity, the increased turbulence (for mixing), pulse frequency, pulse duty cycle, and the limitation of a steady jet to remain attached to a small radius of curvature. Further research is needed to isolate these parameters.
VII. Conclusions The results of this study have addressed two of the major hurdles that limit the application of CC to aircraft: 1) reducing the CC mass flow bleed requirements from the engine and 2 ) Conversion of the cruise drag associated with a blunt TE to thrust through a pneumatic cruise flap. The efficiency of the GACC airfoil is compared to other CC airfoils in Fig. 60. The details of the other CC airfoil data are described in Ref. 17 and shown here to capture the range of possibilities for the GACC configuration.
UD (EQUIV)
Fig. 60 Comparison of GACC efficiency with similar CC airfoils, AOA=O unless otherwise noted (curves do not necessarily represent the envelope of maximum efficiency cI/cdequiva,em,h
PNEUMATIC FLAP PERFORMANCE OF CC AIRFOIL
237
The improved efficiency of the cambered rounded ellipse airfoil' is believed to be a function of the larger radius of the circular TE used on the elliptical airfoil. The increased efficiency of the camber for the elliptical airfoil is also shown for the t / c = 0.20 configuration.18 The camber effects of the GACC airfoil are demonstrated in the generation of higher lift for comparable momentum coefficients. Comparing the GACC efficiency to a typical blown flap38 reveals the lift benefit of attaching the jet through Coanda turning. It is speculated that the blown flap prematurely separates, limiting its lift performance to Cl < 2. Reshaping the blown flap to the dual-radius CC flap profile enables the jet to remain attached to the TE of the flap, extending its lift performance to Cl RZ 5 . It should be noted that LE blowing was required to extend the lift coefficient beyond Cl RZ 5 for the dual radius flap.39 The poor efficiency of the jet flap is generally related to the large blowing requirements associated with the reactionary force:' and the minimal effect on the two-dimensional induced pressure field. The efficiency of the GACC's dual blown configuration highlights the lowspeed cruise conditions. Nominally, the lift requirements for cruise are Cl x 0.5. Recall from Fig. 50 that most of the real drag is in the form of thrust. It is also unclear what jet U to use in the C, equation, because the upper and lower blowing were controlled independently. The general performance of the GACC airfoil is good, but has not been tested to its limits. It is recommended that LE active flow control be added to extend the limits of lift. It is also important to extend the pulsed performance benefits into the supercirculation region as well as evaluating the Three-dimensional (induced drag) effects. Selecting the GACC airfoil section for use on an ESTOL or PAV vehicle will require a system study that accounts for integration of the engine and CC system. A trade study of thrust from the engine alone or a coupled system of engine and the CC airfoil thrust should highlight the benefit of the CC system. The GACC airfoil does seem to be an excellent candidate for the outboard portion of the ESTOL wing, having good lift augmentation capability and good roll and yaw potential.
Appendix A. Wall Interference As a first approximation of the wall interference characteristics, corrections for two-dimensional lift interference can be made using a classic approach described by Krynytzky and Hackett41 and Allan and V i n ~ e n t i For . ~ ~a small model centrally located between two closed parallel walls, corrections for angle of attack, lift, and pitching moment can be estimated using the following equations:
GREGORY
238
S. JONES
-*(L) CL 2
ACm - 192 PH 4cotT = [ 1
+ (2 - M2)&]quncom
where
and
and
Examples of the wall interference corrections described by Eqs. (A.l-A.4) are small, as seen in Figs. A.l-A.4.
0.0000 -0.0025
t ...**.
-0.0050 -20
Fig. A.l
.a.+*
-10
0 AOA
10
0
Angle of attack correction from wall interference (circular TE).
PNEUMATIC FLAP PERFORMANCE OF CC AIRFOIL
239
0.025
I
-0.0501 -20
M
I
-10
0 AOA
Q
10
Fig. A.2 Lift corrections from wall interference (circular TE). 0.010
0.008
0.006
A0.177 0.134 0.093
ACm 0.004 0.002
o -0.002 -20
.
o
o
o
~
10
20
'
m e * * *
-10
0 AOA
Fig. A.3 Moment corrections from wall interference (circular TE).
re A0.177 0.134 + 0.093
-20
-10
0 AOA
10
20
Fig. A.4 Dynamic pressure corrections from wall interference (circular TE).
GREGORY
240
S. JONES
B. Balance Corrections Data reduction equations and tare corrections for pressure lines across balance are given by NF = ~F(NFSC)-
c c c c c
Winteractions)
YM = &M(YMsc) -
+ Pressurecorrection) (PMinteractions + PressureCorrection) (YMinteractions + PressureCorrection)
RM = ~M(RMSC)-
(RMinteractions)
AF = 6AF(AFSC) -
(AFinteractions
PM = &M(PMsc) -
Pressure tare correction for axial, pitching moment, and yawing moment forces are given by
where
where Pressure Tarecorrection = I l P a c t
+ I2Pupper + I P l o w e r + I4PactPupper + IsPactPlower
The accuracy of the balance is highlighted in Table A. 1. The rolling moment and yawing moments are meaningless for two-dimensional testing and will be Table A.l
GACC strain gauge balance accuracy (95% confidence level)
Normal (%FS)
Axial (%FS)
Pitching moment (% FS)
Rolling moment (% FS)
Yawing moment (% FS)
0.04
0.39
0.12
0.07
1.64
24 1
PNEUMATIC FLAP PERFORMANCE OF CC AIRFOIL
Table A.2 Summary of GACC pressure resolutions Po freestream
AP freestream
AP model static
PSIA 15
PSID 1
PSID 2.5
(PSF) 0.1080
Amin (PSF) 0.0072
Amin (PSF) 0.0360
Amin
AP rake
AP wall signature
in. H 2 0 10 Amin (PSF) 0.0052
in. H 2 0 10 Amin (PSF) 0.0052
AP MEMS
PSID 5 Amin (PSF) 0.0720
ignored except when calculating the interactions to obtain corrected normal, axial, and pitching moments.
C. Pressure Measurement Limits The errors associated with the pressure data described above are related to the resolution of the pressure instrumentation. Nominally the pressure instrumentation errors are characterized by percent of full scale. See Table A.2 for a summary of the GACC pressure instrumentation. References ‘Jones, G. S., and J o s h , R. D., Proceedings of the 2004 NASAIONR Circulation Control Workshop, NASA/CP-2005-2 13509, June 2005. ’Rich, P., McKinley, R. J., and Jones, G. S., “Circulation Control in NASA’s Vehicle Systems Program,” NASA/CP-2005-213509/PTl, June 2005, pp. 1-36. 3Moore, M. D., “Wake Vortex Wingtip-Turbine Powered Circulation Control High-Lift System,” NASA/CP-2005-213509/PT2, June 2005, pp. 641 -656. 4McLean, J. D., Crouch, J. D., Stoner, R. C., Sakurai, S., Seidel, G. E., Feifel, W. M., and Rush, H. M., “Study of the Application of Separation Control by Unsteady Excitation to Civil Transport Aircraft,” NASA/CR-1999-209338, June 1999. ’Streett, C. L., “Numerical Simulation of a Flap-Edge Flow Field,” Fourth AIAA/CEAS Aeroacoustics Conference, AIAA Paper 98-2226, June 1998. 6Wood, N., and Nielson, J., “Circulation Control Airfoils Past, Present, and Future,” AIAA Paper 85-0204, Jan. 1985. ’Englar, R. J., “Circulation Control Pneumatic Aerodynamics: Blown Force and Moment Augmentation and Modifications: Past, Present, and Future,” AIAA Paper 2000-2541, June 2000. 8Jones, G. S., Bangert, L. S., Garber, D. P., Huebner, L. D., McKinley, R. E., Sutton, K., Swanson, R. C., and Weinstein, L., “Research Opportunities in Advanced Aerospace Concepts,” NASA, TM-2000-210547, Dec. 2000. ’Davenport, F. J., “A Further Discussion of the Limiting Circulatory Lift of a Finite-Span Wing,” Journal of the Aerospace Sciences, Vol. 27, Dec. 1960, pp. 959-960. “Smith, A. M. O., “High-Lift Aerodynamics,” Journal ofAircraf, Vol. 12, No. 6,1975, pp. 501-530. ‘‘McCormick, B. W., Jr., Aerodynamics of V/STOL Flight, Dover Publications, Mineola, New York, 1999. ”Wilson, M. B., and von Kerczek, C., “An Inventory of Some Force Producers for Use in Marine Vehicle Control,” DTNSRDC-79/097, Nov. 1979.
242
GREGORY
S. JONES
13Betz, A., “Ein Verfahren zur direkten Ermittlug des Profilwiderstandes,” ZFM, Vol. 16, 1925, pp. 42-44. 14Jones, B. M., “The Measurement of Profile Drag by the Pitot Traverse Method,” Reports and Memorandum, No. 1688, Brit. A.R.C., Jan. 1936. ”Schlichting, H., Boundary Layer Theory, 6th ed., McGraw-Hill, New York, 1968. 16Rae,W. H., and Pope, A., Low-Speed Wind Tunnel Testing, 2nd ed., Wiley, New York, 1984. 17Kind, R. J., “A Proposed Method of Circulation Control,” Ph.D. Thesis, Univ. of Cambridge, June 1967. “Englar, R. J., and Williams, R. M., “Test Techniques for High Lift, Two-Dimensional Airfoils with Boundary Layer and Circulation Control for Application to Rotary Wing Aircraft,’’ Canadian Aeronautics and Space Journal, Vol. 19, No. 3, 1973, pp. 93-108. ”Jones, G.S., Viken, S. A., Washbum, A. E., Jenkins, L. N., and Cagle, C. M., “An Active Flow Circulation Controlled Flap Concept for General Aviation Applications,” AIAA Paper 2002-3157, June 2002. ”Lan, C. E., and Roskam, J., Airplane Aerodynamics and Perjormance, Roskam Aviation and Engineering, 1981. ”Homer, S. F., and Borst, H. V., Fluid-Dynamic Lift, Homer Publishing, 1985. ”Raymer, D. P., Aircraft Design: A Conceptual Approach, AIAA Education Series, 3rd ed., AIAA, Reston, VA, 1999. 23McGhee, R. H., and Bingham, G.H., “Low-Speed Aerodynamic Characteristics of a 17-Percent Thick Supercritical Airfoil Section, Including a Comparison Between WindTunnel and Flight Data,” NASA, TM X-2571, July 1972. 24Cagle, C. M., and Jones, G. S., “A Wind Tunnel Model to Explore Unsteady Circulation Control for General Aviation Applications,” AIAA Paper 2002-3240, June 2002. 25Englar, R. J., and Williams, R. M., “Design of Circulation Controlled Stem Plane for Submarine Applications,” David Taylor Naval Ship R&D Center, Rep. NSRDC/AL-200 (AD901-198), March 1971. 26Englar,R. J., “Low Speed Aerodynamic Characteristics of a Small Fixed Trailing Edge Circulation Control Wing Configuration Fitted to a Supercritical Airfoil,” DTNSRDC, Rep. DTNSRDC/ASED-81/08, March 1981. 27Englar, R. J., Hemmerly, R. A., Taylor, D. W., Moore, U. H., Seredinsky, V., Valckenaere, W. G., and Jackson, J. A., “Design of the Circulation Control Wing STOL Demonstrator Aircraft,” AIAA Paper 79- 1842, AIAA Aircraft Systems and Technology Meeting, Aug. 1979; also Journal ofAircraft, Vol. 18, No. 1, 1981, pp. 51-58. ”Pugliese, A. J., and Englar, R. J., “Flight Testing the Circulation Control Wing,” AIAA Paper 79-1791, presented at AIAA Aircraft Systems and Technology Meeting, Aug. 1979. 29Kuethe, A. M., and Chow, C.-Y., Foundations of Aerodynamics, Bases of Aerodynamic Design, 5th ed., Wiley, New York, 1998. 30Preller, R. F., and Rose, 0. J., “Langley Wind Tunnel Force Reduction Program,” NASA, CR-165650, NOV.1980. 31Smith, D. L., “An Efficient Algorithm using Matrix Methods to Solve Wind Tunnel Force-Balance Equations,” NASA, TN-D-6860, Aug. 1972. 32Ewald,B. F. R. (ed.), “Wind Tunnel Wall Correction,” AGARDograph 336, Oct. 1998. 331yer, V., Kuhl, D. D., and Walker, E. L., “Improvements to Wall Corrections at the NASA Langley 14x22-FT Subsonic Tunnel,” AIAA Paper 2003-3950, June 2003.
PNEUMATIC FLAP PERFORMANCE OF CC AIRFOIL
243
34Rose, R. E., Hammer, J. M., and Kizilos, A. P., “Feasibility Study of a Bi-Directional Jet Flap Device for Application to Helicopter Rotor Blades,” Honeywell, Doc. No. 12081FR1, July 1971. 35 Rogers, E. O., and Donnelly, M. J., “Characteristics of a Dual-Slotted Circulation Control Wing of Low Aspect Ratio Intended for Naval Hydrodynamic Applications,” AIAA 42nd Aerospace Sciences Meeting, AIAA Paper 2004-1244, Jan. 2004. 360yler, T. E., and Palmer, W. E., “Exploratory Investigation of Pulse Blowing for Boundary Layer Control,” North American Rockwell, Rept. NR72H-12, Jan. 15, 1972. 37Walters,R. E., Myer, D. P., and Holt, D. J., “Circulation Control by Steady and Pulsed Blowing for a Cambered Elliptical Airfoil,” West Virginia Univ., Aerospace Engineering TR-32, Morgantown, WV, July 1972. 38Lawford,J. A., and Foster, D. N., “Low-Speed Wind Tunnel Tests on a Wing Section with Plain Leading- and Trailing-Edge Flaps Having Boundary-Layer Control by Blowing,” British Aeronautical Research Council R&M 3639, 1970. 39Englar,R. J., and Huson, G. G., “Development of Advanced Circulation Control Using High-Lift Airfoils,” AIAA Paper 83-1847, July 1983. 40Williams, J., and Alexander, A. J., “Some Exploratory Three-Dimensional Jet-Flap Experiments,” Aeronautical Quarterly, Vol. 8, 1957, pp 21 -30. 41Krynytzky, A., and Hackett, J. E., “Choice of Correction Method,” AGARDograph 336, Section 1.4, Oct. 1998. 42Allan, H. J., and Vincenti, W. G., “Wall Interference in a Two-Dimensional-Flow Wind Tunnel with Consideration of the Effect of Compressibility,” NACA Rept. 782, 1944. 43Englar,R. J., “Two-Dimensional Subsonic Wind Tunnel Test of Two 15-percent Thick Circulation Control Airfoils,” NSRDC, Technical Note AL-211, Aug. 1971. 44Alvarez-Calderon, A., and Arnold, F. R., “A Study of the Aerodynamic Characteristics of a High-Lift Device Based on a Rotating Cylinder and Flap,” Stanford Univ., Dept. of Mechanical Engineering Technical Rept. RCF-1, Stanford, CA, 1961.
Chapter 8
Trailing Edge Circulation Control of an Airfoil at Transonic Mach Numbers Michael G. Alexander,* Scott G. Andeqt and Stuart K. Johnsont NASA Langley Research Center, Hampton, Virginia
Nomenclature b = model span, in. c = chord, in. cref= reference chord, 30 in. CD = discharge coefficient Cl = sectional lift coefficient C, = sectional pitching moment coefficient C p = pressure coefficient C, = momentum coefficient g , = gravitation constant = 32.174 lbm-ft/lbf-s h = average measured slot height, in. h / c = nondimensional slot height riz = mass flow, lbm/s P, = freestream static pressure, psia Po = total pressure, psia q = dynamic pressure, psi r = radius s = reference area, ft2 t = airfoil thickness To = total temperature, O R V = velocity, ft/s
*Aerospace Engineer. Associate Member AIM. 'Aerospace Engineer. Copyright 02005 by the American Institute of Aeronautics and Astronautics, Inc. No copyright is asserted in the United States under Title 17, U.S. Code. The U.S. Government has a royalty-free license to exercise all rights under the copyright claimed herein for Governmental purposes. All other rights are reserved by the copyright owner.
245
246
M. G. ALEXANDER, S. G. ANDERS, AND S. K. JOHNSON
x = chordwise distance, in. y = span distance, in. y / b = nondimensional span location a = angle of attack, deg p = density, lbm/ft3 y = ratio of specific heat ACl/C, = lift augmentation ratio
Subscripts jet = air flow that exits nozzle 1 = lower and lift plenum = airfoil plenum s = slot TE = trailing edge u = upper 0.25 = quarter chord co = infinity
I. Introduction IRCULATION control (CC) is considered one of the most efficient methods for lift augmentation at low Mach numbers.' The device augments an airfoil's lifting capability by tangentially ejecting a thin jet of high-momentum air over a rounded trailing edge (TE).* The jet will remain attached to the surface as along as the low static pressures created by the jet are large enough to balance the centrifugal forces acting to detach the jet (Fig. l).3 The jet moves the separation point around the TE toward the lower surface of the wing and entrains the external flowfield. This entrainment and separation point movement produces a net increase in the circulation of the wing, resulting in lift a~gmentation.~
C
Tangential Blowing Over a Rounded Coanda Surface
Fig. 1 Tangential blowing over a Coanda surface.
CC OF AIRFOIL AT TRANSONIC MACH NUMBERS
247
Numerous experimental CC tests using the Coanda effect to enhance lift have been conducted at subsonic velocities on relatively thick (15%) airfoil section^.^ The focus of this experiment is to evaluate the effectiveness of TE CC on a thin airfoil section at transonic Mach numbers. A wind-tunnel test was conducted on a 6% thick slightly cambered elliptical airfoil with both upper- and lower-surface slot blowing. Parametric evaluations of jet slot heights and Coanda surface shapes were conducted at momentum coefficients Ce from 0.0 to 0.12. The data were acquired in the NASA Langley Transonic Dynamics Tunnel at Mach = 0.8 at a = 3 deg and Mach = 0.3 at a = 6 deg, at Reynolds numbers per foot of 1.0 x lo6 and 3.6 x lo5,respectively.
11. Model Description The configuration tested in this experimental investigation is a semispan rectangular circulation control airfoil (CCA) with zero leading edge (LE) and TE sweep, having a circular end plate at the tip. The model, as shown in Fig. 2a, was mounted in the wind tunnel on a splitter plate located 3 ft off the tunnel wall. The model incorporated CC by blowing tangentially from spanwise rectangular slots located upstream of a trailing edge “Coanda surface”. The model has two separate and isolated internal plenums that feed air to either the upper or lower rectangular slot nozzle. The rectangular slot exits are located at x/cref= 0.9 and extend the full span (60 in.) of the
Fig. 2 a) CCA model (view from right rear quarter, looking upstream): b) CCA airfoil section.
248
M. G. ALEXANDER, S. G. ANDERS, AND S. K. JOHNSON xicref = 0.9
x/cref = 0.9
x/cref = 0.9
Fig. 3 Coanda surfaces.
model. The model is instrumented with a total of 157 static and total pressure taps, one accelerometer, and a type-J thermocouple located in each plenum. The model has a surface finish of 32 pin., and the Coanda surface external finish from upper slot exit to lower slot exit was of 16 pin. thickness.
A. Circulation Control Airfoil The circulation control airfoil (CCA) section is a simple 6% thick elliptical airfoil having 0.75% camber (Fig. 2b). The model span b was 60 in. with zero LE and TE sweep. A reference chord (c,f) of 30 in. gave the model an aspect ratio of two and a taper ratio of one. Common practice for testing semispan models on a reflective plane is to refer to this as an aspect ratio (AR) four wing. The CCA model tip is capable of accommodating either a 30-in.-diam circular end plate to promote two-dimensional flow or a “t/2” tip used to evaluate threedimensional effects. The model was tested with the end plate as shown in Fig. 2a. B. Coanda Surface Definition Three elliptical TE surfaces (referred to as Coanda surfaces) were manufactured with length-to-height ratios of 1.78:1, 2.38:1, and 2.98: 1, as illustrated in Fig. 3. The 2.38:l Coanda surface installed on the CCA model with the end plate removed is shown in Fig. 4. The minor axis of the Coanda surface
Fig. 4 End view of a Coanda surface and aft surfaces.
CC OF AIRFOIL AT TRANSONIC MACH NUMBERS
249
Table 1 Coanda radius and slot height dimensions Coanda
Chord, in. rs, in. rTE,in. rsIC bE/C
1.78: 1
2.38: 1
2.98: 1
27.82 1.44 0.25 0.052 0.009
28.09 2.57 0.19 0.091 0.007
28.36 4.02 0.15 0.142 0.005
Guidelines: r/c
0.02 to 0.06 0.024 0.039 0.051 0.14 0.22 0.29
hllrs Wrs h3vS hl/rTE h2/rTE
~~IYTE Guidelines: h l r
0.014 0.022 0.028 0.18 0.30 0.38
0.009 0.014 0.01 8 0.23 0.37 0.48
0.01 to 0.08
was aligned with the slot exit to ensure the minimum exit area occurred at = 0.9. The horizontal axis of the ellipse was then mapped to the camber line of the elliptical airfoil that formed a 5-deg converging nozzle at the slot exit. The Coanda surface spanned the TE of the model (60 in.). Reference 6 provided guidelines for Coanda surface radii of curvature as listed in Table 1. It is not possible to meet the entire guideline radii of curvature on a 6% thick airfoil. It was therefore decided that preference would be given to the slot radius of curvature in an effort to achieve initial jet attachment. As a result, a family of elliptical Coanda surfaces was chosen that have large slot radii of curvature and small TE radii of curvature.
x/c,f
C. Slot Definitions Three upper and lower slot heights for each Coanda surface were possible for this wind-tunnel investigation. The slot heights are given in Table 2. A fourth slot height (h4) was constructed during the test using the upper surface small slot ( h / c = 0.0012) aft skin by applying four layers of tape at 0.0035 in. per layer
Table 2 Slot and chord measurements Slot
c, in.
h, in.
hlc
hl h2 h3 h4
27.82 28.09 28.36 28.36
0.0350 0.0560 0.0730 0.0210
0.0012 0.0020 0.0026 0.0007
250
M. G. ALEXANDER, S. G. ANDERS, AND S. K. JOHNSON
for a total thickness of 0.014 in. The resulting “half height” slot was used with the 2.98: 1 Coanda, resulting in an exit h = 0.021 in. or h/c = 0.0007. The aft upper and lower removable surfaces were designed to set the slot heights by varying the internal mold line while not disturbing the outer mold line of the model. Average measured slot height h and chord lengths were used to determine the heightto-chord ratio ( h / c ) of each slot. Table 2 lists the measured height and chords and the resulting h/c. Slot height against Coanda radius information is shown in Table 1.
D. Aft Surface Three sets of aft surfaces were manufactured and attached to the main airfoil body, which formed the upper and lower external airfoil contour as well as the internal 5-deg convergent nozzle contour (Fig. 5 ) . The aft skins also contained chordwise surface static pressure taps at y / b = 0.5. Any aft surface in combination with any Coanda surface ensured the minimum nozzle area was located at the nozzle exit. Each aft surface also established a discrete slot height above the Coanda surface.
E. End Plate The CCA model used a circular end plate to promote two-dimensional flow conditions. The end plate was a 30-in.-diam circular plate constructed from a 0.25-in.-thick aluminum plate with the outside edge beveled. The design of the end plate was based on sizing criteria found in Ref. 7. A removable cutout located at its TE allowed for Coanda surface removal and replacement. F. Internal Plenum As seen in Fig. 2b, the airfoil section is divided into contiguous, separate, and isolated upper and lower plenums. The ratio of the slot height to plenum height ranged from 3.8 to 12.8 depending on the slot height. This ensured low flow velocities in the plenum that helped maintain uniform plenum flow.
Aft Lower Surface
Fig. 5 Aft surface identification.
CC OF AIRFOIL AT TRANSONIC MACH NUMBERS
251
The model has the capability of holding six removable, 0.050-in.-thick, highpressure-loss screens. The screens were fastened to the plenum floor and extended to the plenum ceiling. Each screen has a porosity of 30% and is capable of being placed in both upper and lower plenums at the three locations. The screen’s porosity was sized using the method described in Ref. 8. A small parametric test was performed using the CCA airfoil and the plenum screens to determine which screen combination created the optimum uniformed flow in the spanwise direction. From those data, it was determined to use one screen in each plenum in the aftmost position. The aft screen was located at x/cref= 0.72 and ran full spanwise and parallel to the slot nozzle.
G. Boundary Layer Trip A boundary layer trip strip’ was located 1.5 in. (measured along the surface) aft of the LE on the upper and lower surface. The trip strip used epoxy dots having a diameter of 0.038 in., a thickness of 0.015 in., and an edge-to-edge spacing distance between the epoxy dots of 0.098 in.
111. Instrumentation All pressures were obtained using miniature electronic pressure scanners. CCA Surface Static Pressures A total of 83 external static surface pressure taps was located at y / b = 0.5 on the upper and lower airfoil surface (42 upper and 41 lower taps). There are two spanwise rows of ten static pressures taps located at x/cref= 0.5 and 0.8 on each upper and lower airfoil surface. A.
B. Coanda Surface Static Pressures Each Coanda surface had a total of 19 static surface pressure taps located at y / b = 0.5 every 10 deg radially from 0 deg to 180 deg with 0 deg and 180 deg at the nozzle exit (Fig. 6 ) . Table 3 Internal plenum pressure tap locations Taps
vlb
X1Cre-f
0.2 0.2
0.3 0.8 0.8 0.8 0.8 0.8
0.45
0.5 0.55 0.8
252
M. G. ALEXANDER, S. G. ANDERS, AND S. K. JOHNSON 0
180
Fig. 6 Coanda surface tap placement.
C. Total Pressures Each plenum had six total pressure taps. Their locations are given in Table 3. Pressure taps at x/c,f = 0.8 are located aft of the high-loss screen and pressure taps x/c,f = 0.3 are used to determine the total pressure entering the plenum from the intake nozzle. The total pressure for the plenum was averaged using taps 2, 3, 4,5 , and 6 to obtain the nozzle exit total pressure. D. Thermocouples The plenum has two iron-constantan, type-J thermocouples located in each plenum aft of the aft plenum screen that were used to measure plenum total temperature. IV. Facility
A. Model Support The Transonic Dynamics Tunnel” (TDT) model support systems used for this test were a sidewall turntable and splitter plate, as depicted in Fig. 7. The splitter plate was located 3 ft from the tunnel wall using wall standoffs. The rigid support and model instrumentation was placed inside an aerodynamic shape or “canoe” located between the splitter plate and the tunnel sidewall. B. Air Supply Air was supplied to the test section via two 1-in. high-pressure flex lines delivering a maximum of 1 lbm/s at 200 psia. Total temperature of the supply air was uncontrolled and ranged from - 13°F to 70°F. Each supply line was attached to a
CC OF AIRFOIL AT TRANSONIC MACH NUMBERS
253
Topview
Fig. 7 CCA model installation in the TDT.
control valve that regulated total pressure to the CCA model. A manually operated crossover line located upstream of the control valve allowed mass flow to be diverted from one line to another. After the control valve, each line of the supply air went through its dedicated critical flow venturi and then entered the model plenum.
V. Test Procedures and Conditions A. Lift and Pitching Moment The sectional lift coefficient [Eq. (l)] and quarter chord pitching moment coefficient [Eq. (2)] were obtained by numerically integrating (with the trapezoidal method) the local pressure coefficient at each y / b = 0.5 chordwise orifice from the upper and lower surface of the model:
254
M. G. ALEXANDER, S. G. ANDERS, AND S. K. JOHNSON
B. Mass Flow The momentum coefficient is calculated using
The ideal jet velocity (ft/s) was calculated" based on the assumption that the slot jet flow expands isentropically to the freestream static pressure [Eq. (4)]:
Mass flow was determined using Eq. ( 5 )
The discharge coefficient was obtained from critical flow venturi calibrations conducted in the NASA Jet Exit facility. The conditions at the critical flow venturi were calculated from a static pressure measurement taken at the throat and a total pressure and temperature near the venturi throat.
VI. Test Conditions The test conditions and ranges are outlined in Table 4. No corrections were applied to account for tunnel flow angularity, wall interference effects, or endplate effects. VII. Discussion of Results The higher Reynolds number data will be presented first, because the experimental investigation at transonic conditions was the main testing objective.
A. Mach = 0.8, a = 3 deg 1. Coanda Su8ace Effect In Figs. 8 and 9, Coanda surface effects are presented for the upper and lower slot blowing, respectively. At Mach = 0.8 at a = 3 deg, each Coanda Table 4 CCA test range of conditions Mach
Po,psia
P,,psia
To,"F
0.3 0.8
2.7-4.1 3.0-4.1
2.6-3.8 2.0-2.7
67-94 95-125
Relft 3.6 7.8
lo5 to 5.5 lo5 to 1.0
lo5 lo5
CC OF AIRFOIL AT TRANSONIC MACH NUMBERS
Coanda Surface
255
Slot Height
Fig. 8 Coanda surface effect, upper slot blowing, Mach = 0.8, a = 3 deg.
surface was capable of generating incremental lift and pitching moment at each blowing condition. Upper slot blowing generated positive lift and negative pitching moment increments, whereas the lower slot blowing generated negative lift and positive pitching moment increments. Generally, the data in Fig. 8 display three distinct regions. The first region is characterized by an increasing lift increment with increasing C, followed by a plateau region in most cases and then, finally, a region of negative lift increment with further increasing C,. As the Coanda surfaces lengthened, increasing C, stretched the regions further. The Coanda surface effect observed in these data indicates the longer Coanda surface is more effective over the mid- to high-C, range, whereas all three Coanda surfaces are equivocal at the low end of C,. The data suggest the jet on the longer Coanda surface remains attached longer over a larger range of momentum coefficients, but, conversely, the jet separates much sooner on the smaller Coanda surfaces. This data trend is generally followed in Fig. 9 for lower surface blowing. However, the lower surface blowing is not as effective in producing lift increment as the upper surface blowing over
256
M. G. ALEXANDER, S. G. ANDERS, AND S. K. JOHNSON
Coanda Surface
Slot Height
Fig. 9 Coanda surface effect, lower slot blowing, Mach = 0.8, a = 3 deg.
the same range of momentum coefficients. Differences in upper and lower slot blowing are probably due to angle of attack (AOA), camber, and jet exit angle. Also, as seen in Fig. 9, none of the Coanda surfaces tested on the lower surface was capable of generating incremental lift or pitching moment for h / c = 0.0026. The lift augmentation ratio (ACJC,) for upper and lower slot blowing is presented in Figs. 10 and 11, respectively. The upper and lower slot blowing data indicated that the larger the Coanda surfaces, the greater the magnitudes of lift augmentation. It was observed that as C, increased, lift augmentation decreased in magnitude with the exception of the data obtained at h / c = 0.0026 (Fig. 11) which, as previously noted, generated insignificant lift increment. Maximum augmentation was typically achieved on each Coanda surface at momentum coefficients less than 0.005. It appeared that the larger Coanda surface was more effective over a larger range of C, at any given h/c.
CC OF AIRFOIL AT TRANSONIC MACH NUMBERS
Coanda Surface
257
Slot Height
Fig. 10 Lift augmentation, Coanda surface effect, upper slot blowing, Mach = 0.8, a = 3 deg
2. Slot Height Effect In Figs. 12 and 13, slot height effects are presented for the upper and lower slot blowing. The data are the same as previously presented, but replotted to better evaluate slot height effect. At Mach = 0.8 at a = 3 deg, the smallest slots were most capable of generating incremental lift and pitching moment at each blowing condition. The lift augmentation ratio for the upper surface slot blowing slot height effect is presented in Figs. 14 and 15. It is observed that the smaller the slot h / c on any given Coanda surface, the greater the lift augmentation. As stated earlier, as C, increased, the augmentation diminished.
B. Mach = 0.3 and (Y = 6 deg 1. Coanda Su$ace Effect
In Figs. 16 and 17, Coanda surface effects are presented for the upper and lower slot blowing, respectively. At Mach = 0.3 at a = 6 deg, each Coanda
258
M. G. ALEXANDER, S. G. ANDERS, AND S. K. JOHNSON
Coanda Surface
Slot Height
Fig. 11 Lift augmentation, Coanda surface effect, lower slot blowing, Mach = 0.8, a = 3 deg.
surface was capable of generating incremental lift and pitching moment at each blowing condition. Increasing incremental lift and moments are observed with increasing blowing rate with upper slot blowing, creating positive lift increments and negative pitching moment increments, whereas lower slot blowing created negative lift and positive pitching moment increments. Upper and lower slot blowing incremental lift and moment data trends for each Coanda surface displayed a marked decrease in effectiveness at higher blowing rates. Also observed is an apparent “pinch down” in the h/c = 0.0012 and 0.0020 slot data from C, = 0.06 to 0.08 that diminished as the Coanda surface increased. This may indicate a reattachment effect (in the immediate region of the slot) followed by a lull where there is little flow turning with C, increment. The lull is then followed by a period of ., On the flow turning around the Coanda bulb as a result of the increased C upper surface blowing (Fig. 16), as the slot size h / c was increased, the
CC OF AIRFOIL AT TRANSONIC MACH NUMBERS
Coanda Surface
259
Slot Height
Fig. 12 Slot height effect, upper slot blowing, Mach = 0.8, (Y = 3 deg.
preferred Coanda surface went from 1.78:l at h / c = 0.0012 to 2.98:l at h / c = 0.0026. It is observed in Fig. 17 that the lower slot blowing force and moment increments followed the same trend as the upper slot blowing, but had reduced absolute values of force and moment increments than that of the upper surface blowing (Fig. 16). Differences in upper and lower slot blowing are probably caused by AOA, camber, and jet exit angle. At Mach = 0.3 at a = 6 deg, the smaller slot ( h / c = 0.0012) on the smaller Coanda surface (1.78: 1) generated the largest increments over the largest C, range, making it the preferred surface at this test condition. The lift augmentation ratio for upper and lower slot blowing is presented in Figs. 18 and 19, respectively. As was seen in the M = 0.8 data, the lift augmentation decreased with increasing C ., Unlike the M = 0.8 data, the smallest Coanda surface generated the largest augmentation ratio from all of the data shown. However, the smallest Coanda surface did not achieve the largest augmentation ratio for all slot heights. At h / c = 0.0012 the 1.78:1 Coanda surface achieves the
260
M. G. ALEXANDER, S. G. ANDERS, AND S. K. JOHNSON
Coanda Surface
Slot Height
Fig. 13 Slot height effect, lower slot blowing, Mach = 0.8, cu = 3 deg.
largest augmentation ratio. At h/c = 0.0026, the 2.98: 1 Coanda surface achieves the largest augmentation ratio.
2. Slot Height Effect In Figs. 20 and 21, slot effects are presented for the upper and lower slot blowing. The data are the same data as previously presented, but replotted to better evaluate slot height effect. For each Coanda surface the data suggest that the smaller the h/c, the greater ACl and ACm generated for the upper (Fig. 20) and lower (Fig. 21) slot blowing. The lift augmentation ratio for the upper and lower slot blowing is presented in Figs. 22 and 23. In Fig. 22, at each Coanda surface tested, the smaller the slot, the greater its augmentation ratio becomes.
C. Nozzle Pressure Ratio In Fig. 24, incremental lift data are presented at Mach numbers of 0.8 and 0.3 as a function of nozzle pressure ratio (NPR). The surface and slot height noted in
CC OF AIRFOIL AT TRANSONIC MACH NUMBERS
Coanda Surface
261
Slot Height
Fig. 14 Lift augmentation, slot height effect, upper slot blowing, Mach = 0.8, a = 3 deg.
the figure was the best configuration for each Mach number. The NPR data are presented as an aid in interpreting the data. For NPR values greater than 1.893, the exit slot is choked and therefore the jet is supersonic.
D. Velocity Ratio In Fig. 25, incremental lift data are presented at Mach numbers of 0.8 and 0.3 as a function of velocity ratio for the same configurations used in the NPR figures. These data are presented for reference purposes similar to the NPR data to orient the reader to the ranges of velocity ratios tested.
E. Pressure Distributions Figure 26a presents data taken at Mach = 0.8 at a = 3 deg, for the 2.98:l Coanda surface and h / c = 0.0012 slot configuration. A C , effect was not observed on the LE of this airfoil. The data suggest a possible weakening of the upper surface shock with increasing C,. In Fig. 26, which shows the
262
M. G. ALEXANDER, S. G. ANDERS, AND S. K. JOHNSON
Coanda Surface
Slot Height
Fig. 15 Lift augmentation, slot height effect, lower slot blowing, Mach = 0.8, a = 3 deg.
Coanda surface pressures, the pressure data suggested a shock just aft of the nozzle exit with flow reattachment and pressure recovery. The surface pressure data indicated the shock moved aft with increasing C ,. Also, note at C, = 0.017 and 0.02, the jet completely detaches from the surface. Figure 27 presents data taken at Mach = 0.3 at a = 6 deg for the 1.78:l Coanda surface and h / c = 0.0012 slot configuration. A C, effect is observed on the LE at this test condition. As C, was increased, the LE suction peak broadened further downstream up to a C, = 0.046. The data indicated at C, 2 0.046 that no further enhancement of the LE suction is observed. In Fig. 27, which shows the Coanda bulb pressures, the pressure data at C, 2 0.046 suggested a shock just aft of the nozzle exit followed by flow reattachment. As C, is increasing, an increasing negative pressure field is seen over the remaining length of the Coanda bulb surface. In addition, the surface pressure data suggest that the shock may be moving aft with increasing C .,
CC OF AIRFOIL AT TRANSONIC MACH NUMBERS
Coanda Surface
263
Slot Height
Fig. 16 Coanda surface effect, upper slot blowing, Mach = 0.3, a = 6 deg.
VIII. Conclusions A wind-tunnel experiment conducted at Mach numbers 0.3 and 0.8 on a two-dimensional, 6% thick airfoil with a modified TE to enhance the Coanda effect by tangential jet slot blowing was accomplished. Incremental sectional lift and quarter-chord pitching moment and lift augmentation ratio data were presented to support any indications of slot height and Coanda surface effects. At the transonic cruise condition, Mach = 0.8 at a = 3 deg, it was found that the effectiveness increased with decreasing slot height and increasing Coanda surface elliptical ratio. The 2.98:l Coanda surface with the upper slot blowing position having a slot height of h / c = 0.0012 slightly outperformed the lower slot position with the upper slot, generating a maximum ACl of 0.25 at a C, of 0.008. At the lower speed and Reynolds number condition, Mach = 0.3 at a = 6 deg, it was found that the effectiveness increased with decreasing slot height and decreasing Coanda surface elliptical ratio. The 1.78:1 Coanda surface with the
264
M. G. ALEXANDER, S. G. ANDERS, AND S. K. JOHNSON
Coanda Surface
Slot Height
Fig. 17 Coanda surface effect, lower slot blowing, Mach = 0.3, a = 6 deg.
upper slot blowing position having a slot height of h / c = 0.0012 gave the maximum ACl generated at 0.75 at a C, of 0.085. Increasing incremental lift and moments are observed with increasing blowing rate with upper slot blowing creating positive lift increments and negative pitching moment increments, whereas lower slot blowing creates negative lift and positive pitching moment increments. Lower slot blowing was not as effective in producing lift and pitching moment increments at transonic velocities as the upper slot blowing over the same range of momentum coefficients. The pressure distribution on all Coanda bulbs at Mach 0.8 suggests the jet detached from the bulb surface at the higher blowing rates, indicating a limit to the amount of blowing that can be accomplished without losing effectiveness. Trailing edge blowing influenced the flowfield upstream of the slot.
NoI( h/c=0.0020)
I
-I
n 9 z
cn
P5 5 0
I
z
C
5rn n
cn
N
Q,
Fig. 18 Lift augmentation, Coanda surface effect, upper slot blowing, Mach = 0.3, (Y = 6 deg.
cn
Slot(h/c=0.0012)
N (3,
Q,
Fig. 19 Lift augmentation, Coanda surface effect, lower slot blowing, Mach = 0.3, a = 6 deg.
CC OF AIRFOIL AT TRANSONIC MACH NUMBERS 267
ka
a
268
M. G. ALEXANDER, S. G. ANDERS, AND S. K. JOHNSON
0
c)
v1
I
M 0
w
Coanda(l.781)
Coanda(2.38 1)
Fig. 22 Left augmentation, slot height effect, upper slot blowing Mach = 0.3, a = 6 deg.
Coanda(2.381)
N
4 0
? rn X
D
z
0
rn
n
v, 0 D
z
0
rn
n
v)
Pz v)
P Fig. 23 Left augmentation, slot height effect, lower slot blowing Mach = 0.3, a = 6 deg.
CC OF AIRFOIL AT TRANSONIC MACH NUMBERS
271
Upper surface blowing 0.8 L
" " ~ " " ~ " " ~ " " ~ " " I " " i
0.7
0.6 c ......................................................................................... 0.5 0.4
1
.
......
......
0.3 c ..... .......,... L-
4
0.2 0.1
0
0
2
1
3
4
6
5
NPR
Lower surface blowing
0 -0.1
-0.2 -0.3
i\,
_ ............................................. ~
............................
-0.4
~
/cj .
Mach AOA; CoanddS;
+ M I =+ 0.3; a = 6f 1.78:1)/0.0012 M = 0.8, a = 3f 2.98:1)/0.0012
,
.I
-
........................................
-0.5
-0.6 _............
....................................................................
i.......
-0.7
-0.8 0
1
2
3
4
5
NPR
Fig. 24 Nozzle pressure ratio vs AC, upper and lower slot blowing.
6
M. G. ALEXANDER, S. G. ANDERS, AND S. K. JOHNSON
272
Upper surface blowing 0.8 0.1 0.6
0.5 1"
0.4 0.3
0.2 0.1 0
0.5
1
1.5
2
2.5 UjetNinf
3
3.5
4
4.5
Lower surface blowing 0
I " " I " " ~ " " ! " " I " " ~ " " ! " " _
-0.1 ~
Mach AoA; CoandaiSlotG / c )
I
-0.2 -0.3 AcI
-0.4 -0.5
-0.6 -0.1 0.5
1
1.5
2
2.5 UjetNinf
3
3.5
4
Fig. 25 Velocity ratio vs AC,,upper and lower slot blowing.
4.5
CC OF AIRFOIL AT TRANSONIC MACH NUMBERS
273
Upper surface leading edge pressure distribution distribution -1.5
-1
C
P
No Blowing Cµ = 0.002 Cµ = 0.003 Cµ = 0.004
Cµ = 0.006 Cµ = 0.008 Cµ = 0.009 Cµ = 0.011
Cµ = 0.012 Cµ = 0.014 Cµ = 0.017 Cµ = 0.02
-0.5
0 Upper Surface Airfoil Leading Edge 0.5 -0.1
0
0.1
0.2
0.3
0.4
x/c Upper surface trailing edge pressure distribution -1.5
-1
C
P
-0.5
0
No Blowing Cµ = 0.002 Cµ = 0.003 Cµ = 0.004 Cµ = 0.006 Cµ = 0.008 Cµ = 0.009 Cµ = 0.011 Cµ = 0.012 Cµ = 0.014 Cµ = 0.017 Cµ = 0.02 Aft upper surface
Coanda Bulb Upper Surface 0.5 0.92
0.94
0.96
0.98
1
x/c Fig. 26 Pressure distribution, C, effect, upper slot blowing, Coanda (2.98:1), slot ( h / c = 0.0012), Mach = 0.8, (Y = 3 deg.
274
M. G. ALEXANDER, S. G. ANDERS, AND S. K. JOHNSON
Upper Surface Leading Edge Pressure Distribution
Fig. 27 Pressure distribution, C, effect, upper slot blowing; Coanda (1.78:1), slot (h/c = 0.0012), Mach = 0.3, a = +6 deg.
CC OF AIRFOIL AT TRANSONIC MACH NUMBERS
275
Acknowledgments The authors would like to acknowledge the assistance of those individuals whose efforts made this test possible. From the TDT, Chuck McClish, Don Keller, Jennifer P. Florance, and Wesley Goodman, from Lockheed-Martin, Jerome Cawthorn, and from the Naval Surface Warfare Center, Carderock Divison, Ernest Rogers and Jane Abramson. References ‘Novak, C. J., Cornelius, K. C., and Road, R. K., “Experimental Investigations of Circular Wall Jet on a Circulation Control Airfoil,” AIAA Paper 87-0155, Jan. 1987. 2 Englar, R. J., “Investigations into and Application of the High Velocity Circulation Control Wall Jet for High Lift and Drag Generation on STOL Aircraft,” AIAA Paper 74-502, June 1974. 3Ahuja, K. K., Sankar, L. N., Englar, R. J., Munro, S. and Liu, Yi., “Application of Circulation Control Technology to Airframe Noise Reduction,” GTRI Rept. A5928/ 1, NASA Grant NAG-1-2146, Feb. 2000. 4Abramson, J., “The Low Speed Characteristics of a 15-Percent Quasi-Elliptical Circulation Control Airfoil with Distributed Camber,” David W. Taylor Naval Ship R&D Center, Rept. DTNSRDC/ASED-79/07 (AD-A084-176), May 1979. ’Nielsen, J. N., and Bigger, J. C., “Recent Progress in Circulation Control Aerodynamics,” AIAA Paper 87-0001, Jan. 1987. 6Rogers, E., and Abramson, J., “Selected Notes on Coanda Circulation Control Airfoils,” unpublished notes, NSWC, Ap. 2002. ’Hoerner, S. F., and Borst, H. V., Fluid-Dynamic Lift, Hoerner Fluid Dynamics, Bakersfield, CA; 2nd ed., June 1992. ‘Blevins, R. D., Applied Fluid Dynamics Handbook, Krieger Publishing Company, Melbourne, Florida. Reprint Edition, June 2002. ’Holmes, J. D., “Transition Trip Technique Study in the McAir Advanced Design Wind Tunnel,” Technical Memorandum 4395, May 1984. (The McAir name is used interchangeably with McDonnell Aircraft Company.) “Staff, Aeroelasticity Branch, “The Langley Transonic Dynamics Tunnel,” LWP-799 Sept. 1969. 11 Englar, R. J., “Two-Dimensional Transonic Wind Tunnel Test of Three 15-percent Thick Circulation Control Airfoils,” Technical Note AL-128, Dec. 1970. ”Alexander, M. G., Anders, S. G., Johnson, S. K., Florence, J. P., and Keller, D. F., “Trailing Edge Blowing on a Two-Dimensional Six-Percent Thick Elliptical Circulation Control Airfoil up to Transonic Conditions,” NASA Technical Memorandum 2005213545, March 2005.
Chapter 9
Experimental and Computational Investigation into the Use of the Coanda Effect on the Bell A821201 Airfoil Gerald Angle II,* Brian O'Hara,* Wade Huebsch,' and James Smith' West Virginia University, Morgantown, West Virginia
Nomenclature A = area b = span, ft C = coefficient of c = chord, ft D = reference lengths, ft F = download force, lbs h = height, ft L = length, ft P = pressure, lb/ft2 Re = Reynolds number V = velocity, ft/s p = density, slug/ft3
Subscripts j =jet p = blowing 00 = freestream 1, 2, 3, 4 = reference numbers
*Graduate Research Assistant, Mechanical and Aerospace Engineering. Student Member AIAA. 'Assistant Professor, Mechanical and Aerospace Engineering. Member AIAA. 'Professor, Mechanical and Aerospace Engineering. Member AIAA. Copyright 0 2005 by the American Institute of Aeronautics and Astronautics, Inc. All rights reserved.
277
278
G. ANGLE II, B. O'HARA, W. HUEBSCH, AND J. SMITH
I. Introduction HE COANDA effect can be described as the balance between the inertial and normal pressure gradient forces in a near-surface jet of a fluid. A simple case used to describe this phenomenon is a two-dimensional wall jet, which entrains the surrounding fluid. As the boundary layer is entrained, the local pressure in the boundary layer is reduced, creating a pressure gradient that pulls or entrains the jet towards the surface. From the conservation of momentum, as fluid is entrained, the jet velocity is reduced. Eventually, the jet velocity is low enough that the fluid viscosity creates an adverse pressure gradient, again separating the flow. Expanding this concept to a convexly curved surface, a pressure gradient is created, forcing the jet to bend around the surface, until the adverse pressure gradient is reached. Newman' determined that the flow in a curved wall jet is relatively insensitive to Reynolds number Re, as defined below, provided it is in excess of a threshold value of 40,000. Thus
T
Re=
[
1
( P - PcO)V,.VcO PV2
l'*
where P is the local pressure, P , is atmospheric pressure, vj and Vi, are the jet and freestream velocities, and p and v are the density and viscosity of air. An approximation of a Coanda jet is a constrained jet, where the streamlines of the freestream act as a restricting surface. Early experimentation into constrained jets determined that the inflow velocities of the jet flow do not differ from the constrained and unconstrained cases, provided that the momentum of the jet is sufficiently higher than that of the freestream. Looking in more detail at the boundary layer of the confined jet as the Reynolds number increases, the flow tends to compress slightly, which inhibits its boundary layer development. This delay in boundary layer growth hinders the entrainment of the flow, maintaining the composition of the jet and increasing the bulk jet velocity. The goal of this work is to use blowing slots to induce the Coanda effect in the leading edges (LE) and trailing edges (TE) of the airfoil. Parameters other than the freestream velocity that affect the ability for flow to remain attached to a curved surface include the four primary variables-radius of curvature, slot location, slot size (height and span), and blowing pressure-which are characterized by the coefficient C, as defined in Eq. ( 2 ) : pj vj'hb
c, = 1/2p,
V icb
where p is the density, V is velocity, h is the slot height, b is the span, c is the airfoil chord, and the subscripts j and 00 represent the jet and freestream values, respectively. General trends exist for these parameters. For instance, as the slot size is reduced, the separation of the flow is delayed because less mass flow can be added to the boundary layer, and because of higher jet velocity (at the same C,). For a given slot location, an increase in the radius of curvature, or the blown pressure, results in a delay of the onset of flow separation. This
COANDA EFFECT ON THE BELL A821201 AIRFOIL
279
delay in separation, controlled by the interaction of all three of the variables, experiences a theoretical upper limit of 245 deg, measured from the slot opening, according to Newman.’ These Coanda jets, placed on the LE and TE of the main wing of the V-22 “Osprey,” can be used to reduce the downforce caused by the rotorwash. This paper expands upon the experimental results shown by Angle et a1.,2 and compares computational methods to simulate this flow phenomena. Discussions of the experimental apparatus and computation methods are presented. The experimental results and computational fluid dynamic (CFD) predictions are shown, together with their comparison and recommendations for further testing. 11. Experimental Apparatus and Procedure
A model of the Bell A821201 airfoil coordinates provided in Felker’ and Felker and Light,g with a 19-in. chord length and an 18-in. span (Fig. 1) was constructed and tested at the West Virginia University Aerodynamic Wind Tunnel Facility. The reader is referred to Angle et a1.* for additional information on the model geometry and wind tunnel facility. This model produced a test section blockage of 15%, which is relatively high for wind tunnel testing. However, this size was needed for the desired instrumentation for the two-dimensional preliminary testing of this concept. Force coefficients can be adjusted to account for solid blockage using the formula presented by Barlow et al.3 and restated in Eq. (3):
Fig. 1 CAD drawing of the experimental model.
280
G. ANGLE II, B. O’HARA, W. HUEBSCH, AND J. SMITH
where CDis the adjusted download force coefficient, CD,,, is the measured download force coefficient, A is the model frontal area, and S is the test section crosssectional area. Surface pressure readings were taken on this model using multiple static pressure ports, as discussed by Angle et a1.2 The aerodynamic forces were measured using a three-load cell (0-25 lb each) system, two in the download direction to provide force and moment, and the third in the normal direction to measure force, as shown in Fig. 2. The structure supporting the model in the test section produced a measurable drag that had to be accounted for when calculating the drag on the wing. Because the experimental apparatus was the same, the instrumentation error is the same as that discussed by Angle et a1.,2 which was found to be 0.11 lb for the force measurements and an error of 0.19 in the pressure coefficient value. The drag on the support apparatus was determined from the standard drag coefficient for a cylinder, from Young et al.4 The resulting moments about the pivot point, above the test section, were removed from the recorded moments, resulting in Eq. (4) for the determination of download force on the model:
COANDA EFFECT ON THE BELL A821201 AIRFOIL
281
where D represents the resulting forces, L denotes the corresponding moment arms, and the subscripts are as shown in Fig. 2. This figure shows the attachment points for the two load cells used to determine the drag on the system, as in Angle et a1.,2 and a third load cell was added to measure the force normal to the drag. Surface pressure taps were also provided on the model, but not repeated for the tests associated with this phase of the project. The large test section, 4 x 6ft, of the Closed Loop Wind Tunnel at West Virginia University, was used for this testing. The maximum airspeed of this test section is just above 60 ft/s; however, because of blockage effects, only 59 ft/s could be achieved during testing. The resulting Reynolds number was 6 x lo5, based on airfoil chord length. Once the model was installed in the test section and the load cells calibrated, testing was conducted with the results shown in Fig. 3. To perform a test the wind tunnel was brought to the desired airspeed and data were collected from the load cells. Data were collected for each test point for a four-minute test sample, with repeats of the baseline after every five tests. Use of the term baseline refers to testing with zero pressure on both the LE and TE blowing slots. After collection of the data, the following procedure was used to reduce the raw voltage data from the load cells. The voltage values were taken through the calibration curves shown in Fig. 3. A baseline average was computed from the three baseline runs to be used as the reference force as well as the zero pressure force value. A simple percent reduction was calculated between the time average data for each run and the baseline average.
15 m
-0.01
-0.005
0
0.005 0.01 Signal Voltage (V)
0.01 5
Fig. 3 Calibration curves for the three load cells.
0.02
282
G. ANGLE II, B. O'HARA, W. HUEBSCH, AND J. SMITH
111. Computational Model and Procedure Because Fluent 6.1 was the computational solver used for this study, its grid generator, Gambit 2.1, was used to create the computational grid and boundaries. A two-dimensional grid was created based on a cross-section of the WVU wind tunnel. The overall dimensions of the grid can be seen in Fig. 4. The general setup used was a two-dimensional cross section of the wind-tunnel test section with a scale model of a Bell A821201 airfoil equipped with 0.0625-in. blowing slots. The LE and TE blowing slots are located at 1.61 and 70.55% of the chord length, respectively. The chord length for this model as in the experimental setup was 19 in. The displayed measurement of 16.76in., in Fig. 4, is the length from the LE of the airfoil to the end of the 67 deg deflected flap. The width of the computational test section was 48 in. and the length was set as 84 in. This length was chosen so that most of the wake profile could be captured. Gambit 2.1 allowed the creation of various types of boundaries. At the top of the grid a velocity inlet that produced a uniform airflow downward was created. The bottom of the grid was specified as a pressure outlet. Each of the blowing slots was created as velocity inlets. The rest of the boundaries were set as noslip walls. The mesh was created using unstructured triangular cells. Additional grid points were clustered around the blowing slots and immediately downstream of the wing, where large gradients and flow separation were expected. A total of 2131 grid points were created on the surface of the airfoil. This resulted in an
84"
60"
COANDA EFFECT ON THE BELL A821201 AIRFOIL
283
average y+ value of approximately 12 for cells next to the wall and very close to the blowing slot; Fluent, Inc., recommends having a mesh with y+ values between 1 and 5 . However, use of the enhanced wall treatment model in Fluent can allow for coarser meshes to be solved. The entire grid comprised 2,184,528 triangular cells and 1,093,464 nodes. Figure 5 shows the overall grid, and Fig. 6 a close-up of the grid on the LE. Attempting to match the experiment, the boundary and initial conditions needed to be correlated to accepted input values. The velocity inlet at the top of the grid was set to 59 ft/s directly downward, the mean experimental velocity. Because the experiment used varying plenum pressures, both the experimental and computational inputs were converted to the blowing slot momentum coefficient. Using Eq. (2), Tables 1 and 2 were created, which show the blowing slot momentum coefficient for the computational and experimental tests. The blowing slot velocities were then determined to be 0, 10, 60, 130, and 200 ft/s. The rest of the initial conditions were set as standard atmospheric conditions.
Fig. 5 Full computational grid.
284
G. ANGLE II, B. O’HARA, W. HUEBSCH, AND J. SMITH Table 1 Computational slot velocity and corresponding momentum coefficient Slot v,ft/s
0 10 60 130 200
CP
0 0.0002 0.0068 0.0319 0.0756
After the initial conditions were all set, the first step was to find an initial solution. The commercially available codes in Fluent 6.1 were used as the solver. For each blowing slot velocity, a laminar solution with first-order accuracy was found. This was done to help the higher order solver converge upon a solution. Each laminar solution was than solved again using second-order upwinding accuracy and Fluent 6.1’s two-equation renormalization group kinetic energy-dissipation (RNG k-e) solver. Other solver settings used in Fluent 6.1 were two-dimensional, double precision, segregated solver, cell-based solution, with enhanced wall treatment. The RNG k-e solver with enhanced wall treatment was selected because it has been found to yield good results for cases dealing with circulation control (CC) by Chang et al.5 A quick study was also carried out comparing the different solver types available in Fluent 6.1. The RNG k-e solver
Fig. 6 Computational grid near the leading edge.
COANDA EFFECT ON THE BELL A821201 AIRFOIL
285
Table 2 Experimental slot pressure and corresponding momentum coefficient
c,
Slot P,psi
0
0
5
0.0116
10 15 20 25
0.0232 0.0348 0.0464
0.0580
produced results that appeared to be very realistic, while taking considerably less time than the five-equation RSM turbulence model.
IV. Experimental Results Data from the normal load cell were found to be negligible because they were of the order of less than 1 lb. This corresponds to a deflection of less than fivethousandths of an inch, indicating an error of the order of the resolution of the load cells in the download direction. The baseline test case (nonactive blowing) experienced a total download force of 18.75 lb, measured from the two load cells, at the test Reynolds number of 5.94 x lo5. As seen in Fig. 7, which is nondimensionalized by dividing out the no-blowing download force, for lower blowing coefficients there is an increase in the download force with 1.04 1
I
1.02 u
$ 1 B -C g 0.98 y
n m
0.96
I
E
qE 0.94 C
2
0.92
-.-
,
0
0.02
0.04
Blowing Coefficient (%)
Fig. 7 Download force variation with blowing coefficient.
0.06
286
G. ANGLE II, B. O’HARA, W. HUEBSCH, AND J. SMITH
the LE slot active, and a smaller increase when the TE is activated. As the blowing coefficient is increased, the LE slot decreases the nondimensional download force, while the TE slot produces a fairly constant increase in download above the baseline value as the blowing coefficient is increased. The curve showing data for both slots active demonstrates the combined effects of the individual blowing slots. The data are summarized in Table 3, where a positive value indicates a reduction in the download on the A821201 airfoil model. These results show that, with the current configuration, the LE is more effective at reducing the download force. However, when using both slots there is still an 8% reduction in the force. It should be noted that no effort has yet been made to optimize slot placement and that the TE flap is deflected according to current V-22 operating practices. These results do show the overall viability of the blowing slot mechanism as a means of reducing the downwash force. There is also the potential to use a variant of the technique discussed in this paper to assist in the control of the pitching moment of the airfoil. By adjusting the blowing pressures separately, the pitching moment can be altered. With further testing, this potential benefit can be better defined. Additional experimental data can be found in Angle et a1.* V. Computational Results Immediately behind the separation points, both at the LE and TE, turbulent eddies formed. These turbulent eddies generally caused lower pressures that increased the download. With the blowing slots in place it was found that these eddies could be reduced in size. This reduction in size is a result of the position of the separation point. Although the separation point is a good indicator of how much the download is being reduced, it is also useful to be able to visualize the areas of lower pressure, for example, where the flow is circulating, with pathline, vorticity, and vector plots. Circulating flow is easily seen by plotting pathlines in the regions behind the blowing slots. Figures 8 and 9 show particle tracks, which are colored by particle identification, near the LE and deflected flap of the airfoil. These figures are helpful from a potential flow point of view and seem to be similar to other
Table 3 Experimental reductions in download force Percent reduction Internal pressure, psig
c,
0 5 10 15 20 25
0 0.01 0.02 0.03 0.05 0.06
LE only
TE only
Both
0
0 -0.35 -1.16 - 1.07 -1.13 -0.77
0 -3.12 - 0.59 2.29 5.08 8.68
- 2.84
0.63 3.88 6.67 9.23
COANDA EFFECT ON THE BELL A821201 AIRFOIL
287
Fig. 8 Pathlines colored by Particle ID near the leading edge.
active CC studies such as the one carried out by Swanson et a1.6 Pathlines can only tell part of the story for download reduction; they illustrate the path of air particles but do not really show turbulence or velocity gradients. Upstream of the wing, the flow appears to be mostly uniform and laminar, and immediately downstream of the wing large amounts of turbulence form. This is shown in Fig. 10, a
Fig. 9 Pathlines colored by Particle ID near the trailing edge.
288
G. ANGLE II, B. O’HARA, W. HUEBSCH, AND J. SMITH
Fig. 10 Contours of vorticity magnitude (I/s) around the airfoil.
vorticity contour plot around the wing. The highest values of vorticity are in the jets of air coming out of the blowing slots. These high values were filtered out so that the circulation downstream of the LE and flap could be seen. Figures 11 and 12 show the velocity vector plots at the LE and TE, respectively. It is shown in the close-up figures that large eddies form just underneath the LE and at the end of the flap. Figures 8 to 12 are all for the 200 ft/s blowing slot case. Given the proper reference values, including depth, characteristic length, velocity, and density, Fluent 6.1 can calculate the forces acting on the airfoil. These forces were comparable to data presented by Riba.’ Figure 13 shows that the download that was computed for the experimental and computational results
Fig. 11 Vector plot near the leading edge.
COANDA EFFECT ON THE BELL A821201 AIRFOIL
289
Fig. 12 Vector plot near the trailing edge.
shows a similar trend. The computational results were all computed at sea-level standard atmospheric conditions whereas the experiment was conducted in Morgantown, WV, at an elevation of 1240 ft. Despite this difference, which was accounted for in the use of force and pressure coefficients, the amount of download when compared to the baseline tests for each approach is very similar, as shown in Fig. 14. Figure 14 is a more appropriate indicator of how the
....................................................................................
-
...................................................................................................
-
-2
5
A
..............................................................................
uExperimental
01
0
0.01
0.02
0.03 0.04 0.05 Blowing Coefficient (Cp)
0.08
0.07
0.08
Fig. 13 Comparison of download between experimental and computational techniques.
0
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
Blowing Coefficient (Ck)
Fig. 14 Comparison of the percent download reduction between experimental and computational techniques.
computational tests compare with the experiments than Fig. 13 because of its nondimensional nature.
VI. Conclusions This chapter has presented CC as a method to reduce the force felt by a surface in the wake of a rotor. Typical applications of CC are looking at the airflow over the surface of the airfoil, where this particular application is looking at the flow approximately normal, - 85 deg angle of attack using the conventional definition. This difference in flow characteristics seems to have slightly altered the trends present in the conventional application of active CC methods. At low blowing coefficients there is a small increase in the force, followed by a decrease in the force. Some of this decrease in force is a result of the reduction in wake area, but it is not clear that this is the only aspect capable of reducing the force. Further investigation will help clarify the true force reducing mechanism(s), which could include the jet momentum conservation. The trends in the experimental and computational tests show that active CC, through the use of blowing slots on the LE and TE of the Bell A821201 airfoil, can reduce the download force felt from the rotor wash of a tilt-rotor aircraft. Experimental testing demonstrated a reduction of approximately 10% from the baseline 18.7 lb download. The baseline download of the computational tunnel simulation was found to be 241b and had a maximum reduction of around 12%. The percent reduction of the download provided a reasonable match in both the trend and magnitude.
COANDA EFFECT ON THE BELL A821201 AIRFOIL
291
Many aspects of using CC need to be investigated further. Some of these include looking into optimizing the placement of the LE and TE slots. Current testing has only studied one location for each of the slots. With decent agreement between the computational model and the experimental results, the process of finding optimum placement will be simplified. Currently, new experimental and continuing computational models are under development to address aspects of the current data. The new experimental model will be sized to fit into the small test section of the WVU Closed Loop Wind Tunnel to allow for testing at different Reynolds numbers and take test section blockage into account. A cost/benefits analysis is also being conducted to determine the practical application of using such a system on a tilt-rotor aircraft to increase the such aircrafts’ performance.
References ‘Newman, B. G., The Dejexion of Plane Jets By Adjacent Boundaries-Coanda Effect; Contained in Boundary Layer and Flow Control, Vol. 1, Pergamon Press, New York, 1961, p. 232. ’Angle, G., Riba, C., Huebsch, W., Thompson, G., and Smith, J., “Download Wake Reduction Investigation for Application on the V-22 ‘Osprey’,” Society of Automotive Engineers Technical Paper 2003-01-3020, Sept. 2003. 3Barlow, J. B., Rae, W. H., and Pope, A., Low-Speed Wind Tunnel Testing, 3rd Ed., Wiley, New York, 1999. 4Young, D. F., Munson, B. R., and Okiishi, T. H., A Brief Introduction to Fluid Mechanics, Wiley, New York, 1997. ’Chang, P. A. 111, Slomski, J., Marino, T., and Ebert, M. P., “Numerical Simulation of Two- and Three-Dimensional Circulation Control Problems,” A I M Paper 2005-0080, Jan. 2004. %wanson, R. C., Rumsey, C. L., and Anders, S. G., “Progress Towards Computational Method for Circulation Control Airfoils,” AIAA Paper 2005-0089, Jan. 2005. ’Riba, C. A., “Circulation Control for Download Wake Reduction in the V-22 Aircraft,” Masters Thesis, Department of Mechanical and Aerospace Engineering, West Virginia Univ., Morgantown, WV, 2003. ‘Felker, F. F., “Wing Download Results from a Test of a 0.658-Scale V-22 Rotor and Wing,” Journal of the American Helicopter Society, 1992, pp. 58-63. ’Felker, F. F., and Light, J. S., “Reduction of Tilt Rotor Download Using Circulation Control,” Proceedings of the Circulation-Control Workshop, 1986, pp. 429-447. 10 Englar, R. J., “Experimental Investigation of the High Velocity Coanda Wall Jet Applied to Bluff Trailing Edge Circulation Control Airfoils,” Masters Thesis, Univ. of Maryland, College Park, MD, 1973. “Felker, F. F., Shinoda, P. R., Heffernam, R. M., and Sheehy, H. F., “Wing Force and Surface Pressure Data from a Hover Test of a 0.658-Scale V-22 Rotor and Wing,” NASA TM-102244, Feb. 1990.
Chapter 10
Novel Flow Control Method for Airfoil Performance Enhancement Using Co-Flow Jet Ge-Cheng Zha* and Craig D. Paxtont University of Miami, Coral Gables, Florida
Nomenclature CL = lift coefficient C , = drag coefficient
C, = momentum coefficient c, = specific fuel consumption D = drag E = endurance F = thrust m = mass flow rate k = turbulent kinetic energy M = Mach number PR = total pressure ratio of engine compressor P , = total pressure R = range Re = Reynolds number S = wing span area ( b x chord) U = velocity V = velocity Wo = takeoff gross weight W1 = empty weight y+ = nondimensional length scale for turbulent boundary layer
*Associate Professor, Department of Mechanical and Aerospace Engineering. +Graduate Student, Department of Mechanical and Aerospace Engineering. Copyright 0 2005 by the American Institute of Aeronautics and Astronautics, Inc. All rights reserved.
293
294
G.-C.ZHA AND C. D. PAXTON
a = angle of attack p = density
7 = efficiency y = ratio of specific heats E = turbulent dissipation rate Subscripts = freestream j =jet injection
00
L = landing TO = takeoff T = touch ground
I. Introduction 0 ACHIEVE high-performance aircraft design, revolutionary technology advancement should be pursued to dramatically reduce the weight of aircraft and fuel consumption, and significantly increase aircraft mission payload and maneuverability. Both military and commercial aircraft will benefit from the technology. Flow control (FC) is the most promising route to break through the conventional aerodynamic design limit and bring dramatic performance improvement to aircraft.'-3 The National Aeronautics and Space Administration (NASA), U.S. Air Force, and aerospace industry have recently made great efforts to develop flow control t e ~ h n o l o g y . ~To - ~ enhance lift and suppress separation, various flow control techniques have been used, including a rotating cylinder at the leading edge (LE) and trailing edge (TE),3,899circulation control (CC) using tangential blowing at LE and TE,1°-16 multi-element pulsed jet separation c o n t r 0 1 , ' ~ -and ~ ~ so on. When a flow control technique is developed, there are three issues that may need to be considered: 1) effectiveness-the FC method should provide substantial improvement in aerodynamic performance, which primarily includes lift enhancement, drag reduction, and stall margin increase (suppression of separation); 2) energy efficiency-the FC method should not cause significantly more energy expenditure, otherwise the penalty may outweigh the benefit for the whole aircraft as a system, including minimal penalty to the propulsion system and minimal weight increase resulting from the FC system; 3) easy implementation-the FC technique should not be too difficult to implement. The rotating cylinder method is generally most effective when the LE or TE are thick, and hence may be more applicable to a low-speed airfoil. It also needs a system to drive the rotating system and can increase aircraft weight. The multi-element airfoil can generate high lift, but generally comes with large drag and weight penalty due to the moving parts. In addition, the highlift flap system increases noise during landing.23 A cc airfoillO,l 1,15,16 relies on a local favorable pressure gradient on a curved surface to attach the flow-the Coanda effect. Such a favorable pressure gradient
T
CONTROL METHOD USING CO-FLOW JET
295
exists at the airfoil LE as a result of the suction and at the end of the TE because of the low base pressure when the TE is blunt. To make the CC airfoil effective, the blunt TE is therefore needed. However, this will create large drag at cruise. To overcome the dependence on a large TE for the CC airfoil, a movable flap at the airfoil TE has been suggested by Englar." The moving parts will increase the weight penalty to the aircraft. At large angle of attack (AOA), because the main flow cannot resist the large adverse pressure gradient, the local TE favorable pressure gradient cannot be achieved and hence the Coanda effect is difficult to maintain. If only TE blowing is used, the CC airfoil will usually stall at a smaller AOA than the regular noncontrolled airfoil.24 To increase stall margin, LE blowing needs to be added.24 A considerably high penalty placed by the CC airfoil on the propulsion system is the dumped blowing jet mass flow. The blowing air for the wing is usually sourced from the engine compressor bleed. The mass flow rate of the engine bleed is directly proportional to the decrease of thrust; that is, an engine will suffer 1% thrust decrease for 1% bleed flow used for wing flow control, and suffer 1-3% fuel consumption increase depending on whether the bleed is from the compressor front stage or back stage. To avoid the jet mass flow rate penalty caused by blowing, the synthetic jet or pulsed jet with open or closed loop feedback control are These methods need a jet generation system, and complicated actuation and sensor systems, which may increase the degree of difficulty in implementing the FC system and increase the weight of the aircraft as well. Because the interaction of the synthetic jet with the main flow is generally weak, its effectiveness in enhancing lift and suppressing separation may not be as dramatic as desired. For example, the results shown in Ref. 19 using the periodic synthetic jet show about 35% increase , , ,C , and little increase of stall AOA, while the co-flow jet airfoil tested in of the Ref. 25 increases the , , ,C , and AOA range by 220 and 153%, respectively, with C, = 0.28. A movable flap is also used with the synthetic jet flow control airfoil studied in Ref. 19, which will increase the aircraft weight. The new airfoil flow control technique using the co-flow jet (CFJ)26suggested in this paper is aimed at considering all the three issues mentioned above, that is, effectiveness, energy efficiency, and ease of implementation. The co-flow jet airfoil opens an injection slot near the LE and a suction slot near the TE on the airfoil suction surface. The slots are opened by translating a great portion of the suction surface downward. A high-energy jet is injected tangentially near the LE, and the same amount of mass flow is sucked in near the TE. The turbulent shear layer between the main flow and the jet causes strong turbulence diffusion and mixing, which enhances lateral transport of energy from the jet to the main flow and allows the main flow to overcome a severe adverse pressure gradient and remain attached at a high AOA. The strong adverse pressure gradient enhances jet mixing,27 and the stall margin is significantly increased. The high jet velocity induces high main flow velocity on the suction surface and hence creates high circulation and lift. The energized main flow will fill the wake velocity deficit, which results in a reduced drag or thrust (negative drag). A CFJ airfoil wing does not need a high-lift flap system and can therefore reduce noise during landing. A CFJ airfoil does not rely on the Coanda effect at the LE or TE and thick LE or TE are not required. Hence, the low form
296
G.-C.ZHA AND C. D. PAXTON
drag of modem airfoils can be maintained. The CFJ technique can be applied to any type of airfoil, including low-speed thick airfoils and high-speed thin airfoils. The level of lift enhancement, drag reduction, and stall mar in increase of the CFJ airfoil is very dramatic, as proved by wind-tunnel tests.&,** Because a CFJ airfoil blows and sucks the same amount of mass flow, the jet mass flow can be recirculated through the propulsion system instead of being dumped away. This can significantly reduce the penalty of energy expenditure to the overall airframe-propulsion system when compared to the blowing-only methods. The CFJ can always be on during the entire flight mission. The lift enhancement and drag reduction can be controlled by adjusting the injection total pressure, and hence the jet mass flow rate, throughout the mission according to different needs. No moving parts are required. The CFJ airfoil concept suggested in this paper appears to have the following advantages: 1) It is very effective in enhancing lift and suppressing separation. 2) It dramatically reduces drag or creates thrust (like a bird wing generating both lift and thrust) and hence can achieve very high C,/Co at low AOA (cruise), and very high lift and drag at high AOA (takeoff and landing). 3) It can significantly increase AOA operating range and stall margin. 4) It imparts only a small penalty to the propulsion system. 5 ) It can be applied to any airfoil, thick or thin. 6) It can be used for the entire flying mission instead of only during takeoff and landing. 7) It can be used for low- and high-speed aircraft. 8) It is easy to implement, with no moving parts. The preceding advantages of the CFJ airfoil may derive the following superior aircraft performances: 1) extremely short takeoff and landing distances; 2) supersonic aircraft having small wing size matching cruise need, but also having high subsonic performance (e.g., high lift as low drag at M < 1); 3) high maneuverability, high safety, and fast acceleration military aircraft; 4) very economic fuel consumption; 5 ) small wing span for easy storage, light weight, and reduced skin friction and form drag; 6) low noise because of no high-lift flap system at landing (at takeoff, the wing thrust or reduced wing drag will rely less on the engine thrust and hence will have less nozzle jet velocity, which will result in lower noise); 7) heavy lift rotorcraft with effectively no dynamic stall; and 8) stealth aircraft with no moving control surface. The purpose of this research project is to study the working principle and demonstrate the superior performance of the CFJ airfoil based on CFD simulation and experiment. This paper presents the CFD results and analysis, which are the basis for the wind-tunnel tests of proof of concept conducted in Refs. 25 and 28. The detailed CFD data also provide a very useful qualitative physical insight into the working mechanism of the CFJ airfoil. 11. Results and Discussion
As a reference, the CFJ airfoil is compared with the baseline airfoil with no flow control. Figure 1 shows the baseline airfoil, NACA2415, and the airfoil
CONTROL METHOD USING CO-FLOW JET
297
baseline airfoil
injection slot
suction slot
Fig. 1 Baseline airfoil NACA2415 and the airfoil with CFJ slot.
with CFJ slot. The chord length is 0.3 m. The coflow jet airfoil is modified from the baseline airfoil by translating the suction surface vertically lower by 1.67% of the chord. The slot surface shape is exactly the same as the original baseline airfoil suction surface. The slot inlet and exit are located at 6.72 and 88.72% of the chord from the LE. The slot inlet and exit faces are normal to the slot surface to ensure that the jet will be tangential to the main flow. The slot inlet and exit area are 1.56 and 1.63% of the chord. The Fluent CFD software is used as the tool to simulate the airfoil flows in this study. The mean flow governing equations are the two-dimensional compressible Navier-Stokes equations. The k - E turbulence model with wall function is used to save CPU time. The solutions of two typical cases are compared with the solutions using the k--E model integrating to the wall. The results show little difference. When the wall function is used, y t is of the order of 15-100. When the turbulent boundary layer is solved by integrating to the wall, the y? is of the order of 1. The wall function method therefore requires less grid to resolve the boundary layer and significantly saves CPU time. The reason that k--E is used is because of its capability of taking into account the turbulent boundary layer history effect by solving the complete transport equations of k and &, and the k--E model is more capable than algebraic models to predict the separated flows, which occur when the airfoil stallsat high AOA. The full turbulent boundary layer assumption is used because the CFD solver does not have a transition model. The 0-mesh is generated as shown in the zoomed region around the airfoil in Fig. 2. The baseline mesh has the dimensions 240 x 100 in the direction around the airfoil and in the radial direction, respectively. In the CFJ slot, the mesh size is 80 x 12 in the streamwise and spanwise directions, respectively. A rectangular farfield boundary is used with the downstream boundary extended to 30 chord length, upstream, lower and upper boundary to 20 chord length. The y+ ranges from 15-30 on the airfoil surface. The freestream Mach number is 0.3 and the Reynolds number is 1.9 x lo6. For all the computation, the jet inlet holds a constant total pressure equal to 1.315Pt,. The static pressure at jet suction is iterated to match the jet injection mass flow rate.
298
G.-C.ZHA AND C. D. PAXTON
Fig. 2 Zoomed mesh around the airfoil with co-flow slot.
CFJ Airfoil Performance Figure 3 presents the lift coefficient against angle of attack for the baseline airfoil and the CFJ airfoil. For the baseline airfoil, the lift coefficient predicted by CFD agrees excellently with the experiment results at Re = 3 x lo6 before CFD predicts a little delayed stall and higher lift coefficient in the stall region. Figure 3 indicates that the lift of the CFJ airfoil is increased significantly. The zero-lift AOA for the baseline airfoil is - 2 deg, and is -6 deg for the CFJ airfoil. The stall AOA is increased by 2 deg. Hence the operating range of AOA is increased totally by 38%. The maximum lift value is increased by SO%, which is the minimum increase in the order of magnitude. When the AOA is decreased, the lift increase is greater in percentage terms. For example, at AOA = 2 deg, the lift increase is 250%. For the CFJ airfoil, a few selected points are recalculated using the refined mesh of dimensions 480 x 200 around the airfoil and 160 x 30 in the slot. The refined mesh lift coefficients are shown in Fig. 3 and agree excellently with the baseline mesh, which indicates that the numerical solutions are converged based on the mesh size. Figure 4 shows the streamlines at AOA = 20 deg. The baseline airfoil has a massive separation, whereas the CFJ airfoil flow is nicely attached. The attached flow is mainly a result of the turbulent mixing,30 which transfers energy from the jet to the main flow so that the main flow can overcome the severe adverse pressure gradient to stay attached. A.
CONTROL METHOD USING CO-FLOW JET
299
AOA
Fig. 3 Lift coefficient against angle of attack.
Figure 5 is the isentropic Mach number distribution on the surface of the airfoil at AOA = 20 deg. The isentropic Mach number is defined as
The isentropic Mach number is only a function of surface static pressure. Hence it indirectly gives the surface static pressure. At the same time, the isentropic Mach number also indicates the approximate Mach number outside of the wall boundary layer assuming that the total pressure loss is small. Figure 5 shows that the CFJ airfoil creates a very strong suction effect near the LE and the flow is accelerated from the freestream Mach number 0.3 to the peak Mach number 1.7. The supersonic flow is only in the LE region and smoothly transits to subsonic flow with no shock wave created. The peak Mach number of the baseline airfoil is about 0.9. However, the baseline airfoil cannot sustain the severe pressure gradient and the massive separation yields small loading on the aft portion of the airfoil. The CFJ airfoil has much higher LE acceleration and diffusion on the suction surface and stronger deceleration on the pressure surface, which results in higher lift and circulation. Figure 5 also shows that the LE stagnation point of the CFJ airfoil is located more downstream than that of the baseline airfoil because of higher circulation. The first spikes near the LE are caused by the CFJ injection, which induces the strong LE suction through turbulence mixing. The shape of the spike is not necessarily
G.-C.ZHA AND C. D. PAXTON
300
baseline airfoil
Fig. 4 Streamlines at angle of attack of 20 deg.
5 =cf
1.6
-
1.4
-
1.2
-
I
_ _ _ _ _baseline _.
-coflow
1-
8
.-P 0.8 9
E
-8
0.6
-
’.
W
0
h
J
I
I
I
0.25
I
I
I
I
I
0.5 WChord
I
I
I
I
I
0.75
I
I
I
I
I 1
Fig. 5 Surface isentropic Mach number distribution at angle of attack of 20 deg.
CONTROL METHOD USING CO-FLOW JET
301
accurate and may be created by the numerical boundary condition treatment. The second spike near the TE is a result of the low-pressure suction at the jet suction slot. Figure 6 presents, the Mach number contours in the LE region at AOA = 20 deg for the baseline and CFJ airfoils. It shows that the CFJ airfoil has a local supersonic region near the LE. The high-energy jet mixes with the mainflow through a turbulent shear layer. It should be noted that the fundamental mechanism of the CFJ airfoil is the turbulent mixing between the jet and the main flow, which transfers energy
baseline airfoil
Fig. 6 Mach number contours at 01 = 20 deg for the baseline and CFJ airfoil.
302
G.-C.ZHA AND C. D. PAXTON
from the jet to the main flow. A high mixing rate is therefore desirable. As indicated by Greitzer et al.,27 the adverse pressure gradient enhances jet mixing. Based on this principle, the injection slot of the CFJ airfoil is located downstream of the LE suction, as shown in Fig. 5 . After the LE suction, the pressure continuously increases until reaching the suction slot near the TE. This is very different from the CC airfoil technique, which places the injection right on the geometric leading position, where the LE suction starts, and a strong local favorable pressure gradient exists because of the suction. The other factor that enhances the turbulent jet mixing is having a long enough distance in which the mixing can occur; that is, the suction slot should be located as close to the TE as possible, subject to geometric constraint. However, the CFD simulation indicates that the CFJ airfoil performance is more sensitive to the injection location than to the suction slot location. In general, the closer the injection to the LE, the more effective the CFJ, but the injection must be located downstream of the LE suction peak. Unlike the CC airfoil, which blows at the LE (near the stagnation point with high pressure) and at the TE where the pressure is high, the CFJ airfoil has the injection downstream of the LE suction peak where the pressure is near its lowest, and has the suction near the TE where the pressure is nearly highest (except for the stagnation point). The CFJ airfoil therefore creates a more favorable pressure condition for injection and suction and may need less energy to pump the same amount of jet mass flow than does a CC airfoil. In this study, although the AOA varies, the CFJ injection total pressure is held constant to simulate passive flow control. At different AOA, the main flow will have different static pressure at the location of the jet injection, which determines the jet mass flow rate of the jet and the jet injection velocity. The jet momentum coefficient therefore varies with AOA. The jet momentum coefficient based on the conventional definition is given by mj vj c,- 0.5pmU&S
where mj is the injection mass flow rate, vj is the injection velocity, p, and U , are the freestream density and velocity, and S is the wing span area. Figure 7 shows the variation of C , with AOA. When AOA varies from -8 to 22 deg, C, increases from 0.15 to 0.25. Figure 8 is the drag polar for the baseline airfoil and the CFJ airfoil. When the AOA is high, both the lift and drag of the CFJ airfoil are significantly higher than for the baseline airfoil. When comparing the maximum lift points for the two airfoils, the drag of the CFJ airfoil is 160% higher than that of the baseline airfoil. However, when AOA < 4 deg, the lift of the CFJ airfoil is significantly higher and the drag is significantly lower than that of the baseline airfoil. When AOA < 0 deg, the lift coefficient is still very large (Cl = 0.862 at AOA = 0 deg; see Fig. 3), but the drag becomes negative and a thrust is generated. The thrust is primarily generated from the strong LE suction, which is the same mechanism as a flapping bird wing generating both lift and
CONTROL METHOD USING CO-FLOW JET
303
0.2
c 0)
0.05
0
Fig. 7 Jet momentum coefficient against angle of attack.
2.5
I
-
2 1.5
u
-
1 0.5
- - - - - .base line
-
coflow
-0.5
I I I
I
0
I
\
,
I
I
I
I
I
0.1
I
I
I
I
I
I
0.2
Cd
Fig. 8 Drag polar for the baseline and CFJ airfoil.
I
I
0.3
G.-C.ZHA AND C. D. PAXTON
304 0.3 0.25
-
0.2
-
0.15
-
0.1
-
0.05
-
3
- - - - - . Cdpressure - _ - _ - Cdfriction
-.-I-.---.__
0-0.05
" I
c -
L r r r r ~ ' ' ' ~ ' ' ' ' ~ ' ' ' ' ~ ' ' ' ' ~ ' ' ' ' ~ ' ' '
Fig. 9 Calculated drag coefficients against angle of attack for CFJ airfoil.
The drag of an airfoil arises from two sources, friction drag and pressure drag (form drag). The friction drag will always be in the opposite direction of the flight, that is, always positive. The negative drag must therefore be from the pressure drag. This can be seen from Fig. 9, which shows the friction drag, pressure drag, and total drag for the CFJ airfoil. Figure 9 indicates that the friction drag is fairly constant and decreases slightly near stall. However, the pressure drag varies largely. The pressure drag is the dominant contribution to the total drag near stall. When AOA is decreased, the pressure drag also decreases monotonically. When AOA < 4 deg, the pressure drag becomes negative, and the total drag is reduced negative values when AOA < 0 deg because of the strong LE suction. Figure 10 shows the drag distribution of the baseline airfoil. Similar to the CFJ airfoil, the friction drag is also fairly constant compared with the pressure drag. The pressure drag decreases when the AOA is decreased from the stall region. However, the pressure drag increases when the decreasing AOA passes the zero lift point and does not become negative. This is because there is no AOA that can create a strong enough LE suction for the baseline airfoil. The negative drag may also be explained from the control volume point of view. The high-velocity jet transfers the kinetic energy to the main flow because of turbulent mixing. When the AOA is not large, the diffusion is not severe. The main flow on the suction surface has a large streamwise velocity past the TE so that the streamwise velocity in the wake region is greater than the freestream velocity. This can be seen in Fig. 11, which shows the wake shape of the baseline and the CFJ airfoil at one chord length downstream of the TE. The wake of the baseline airfoil has the usual defect shape, whereas the wake of the CFJ airfoil has a protruding shape. Figure 12 presents
CONTROL METHOD USING CO-FLOW JET
305
/! I
I
t
k
0.05
0 AOA
Fig. 10 Calculated drag coefficients against angle of attack for baseline airfoil.
1.1
_ _ _ _baseline __. coflow
1.075 1.05
t
1.025
0.975 0.95 0.925
1
Fig. 11 Wake shape for the baseline and CFJ airfoil at (Y = 0 deg.
306
G.-C.ZHA AND C. D. PAXTON
basellne alrfoll
Fig. 12 Wake Mach number contours at a = 0.0 deg for the baseline and CFJ airfoil.
the Mach number contours at the TE region at AOA = 0 deg. It shows that the high-energy jet of the CFJ airfoil fills the wake and a thrust is created. Based on the control volume analysis, the drag is determined by3
where U is the wake velocity.
CONTROL METHOD USING CO-FLOW JET
307
When U is greater than Urn,the drag is negative and becomes thrust. When the AOA is very large, the jet energy is mostly used to diffuse the flow to make the flow attached. For the current study, with a constant jet inlet total pressure of 1.35P,, at AOA = 20 deg, the CFJ does not provide enough energy to the main flow and the wake velocity deficit is very large. The pressure drag is therefore overwhelming, which is desirable for short distance landing.
B. Energy Expenditure The CFJ airfoil achieves performance enhancement using the powered coflow jet, which will involve a certain amount of energy cost. The hypothesis is that the performance gain from increased lift, reduced drag, and increased stall margin will outweigh the cost of the energy expenditure of the jet; that is, the benefit will be realized when the airframe and propulsion are integrated as one system, because the jet is usually sourced from the engine. The analysis of this section is to provide the theoretical foundation of the quantitative analysis of mission analysis given in the next section. Assuming a jet engine is used to power the airplane, the power required to energize the CFJ can be considered as a part of the energy loss of the engine compressor. In other words, an extra amount of fuel needs to be burned to drive the compressor with the CFJ pumping system. The loss resulting from the CFJ is given as
Loss =
Powercfj Powercompressor
yiz,fj is the ratio of the CFJ mass flow rate to the engine mass flow rate, hcfj= hcfj/hengine, q&the efficiency, and PR is the total pressure ratio at the inlet and exit. The hcfjis small and PR,fj is usually also far smaller than PRcompressor. The penalty to the overall fuel consumption as a result of the loss of CFJ will therefore be small. The primary penalty to the energy expenditure of the whole aircraft is a result of the mass flow dumped by the flow control such as in the blowing-only method. The fuel consumption dramatically increases and the thrust is greatly reduced if a part of the flow is bled from the engine. This can be seen by applying a control volume to an engine and, assuming that the flow at the engine exit is expanded to ambient pressure, the thrust is given by where
F = (hinlet
+ hfue1)vnozzle - hinletvinlet = hnozzlevnozzle - h in le tv w
(5)
It is obvious that the bled mass flow from the compressor will decrease hnozzle, which will directly reduce the thrust. Assuming that the engine is running on the ground with V, = 0, the thrust becomes
308
G.-C.ZHA AND C. D. PAXTON
where Eq. (6) suggests that the thrust decrease will be directly proportional to the mass flow dumped if a blowing-only flow control is used. For a recirculating CFJ airfoil, this serious penalty is avoided. The specific fuel consumption (SFC) is defined as
It is clear from this that the penalty to the SFC as a result of the dumped flow is very high. The only penalty in the CFJ airfoil for aircraft fuel consumption is a result of the compressor loss given in Eq. (4), which is small and will be easily offset by the dramatic gain due to the high ratio of L I D and reduced wing weight. These advantages will be shown in the next section.
C. Mission Analysis of F-5E Aircraft Using CFJ Wing The purpose of this section is to conduct a preliminary mission analysis to study if a CFJ airfoil will be beneficial from the viewpoint of an integrated airframe-propulsion system. The military aircraft F-5E is selected as the example, because a detailed mission analysis has been conducted by Roth et a1.32-35and data are available. In Refs. 32 -35 a generalized vehicle thermodynamic loss model is introduced and a loss deck is created for the drag loss of each component of the aircraft, including the airframe and propulsion systems. The unification of the airframe drag loss and propulsion system loss makes it possible to identify the contribution of each component drag to the total fuel consumption, which is particularly useful for an aircraft designer in finding and optimizing crucial components to improve the efficiency of the whole aircraft. The F-5E design mission comprises a subsonic area intercept of 450 n mile range. The mission includes a maximum power takeoff, climb, subsonic cruise to the combat zone, 5 min allowance at Mach 1.3, 50,000 ft maximum power for combat, followed by a subsonic return cruise and 20min reserve loiter, plus 5% fuel reserve. The aircraft is powered by two J85-GE-21 engines. The total time for the mission from takeoff to landing is 84 min. Table 1 gives the detailed breakdown of fuel consumption for the baseline F-5E.33 We created an artificial F-5E using the recirculating CFJ airfoil, named F-5E-CFJ, to carry out the same mission with the same amount of payload and fuel. Without actually testing or calculating the F-5E geometry and flowfield, it is difficult to give the precise values of airfoil performance. Based on the wind-tunnel tests conducted in Refs. 25 and 28, a conservative estimate of the fuel consumption may be given. In the estimation, CL is assumed to be twice the baseline CL through the whole mission, with C , = 0.1. The wing lifting surface area therefore only needs to be half that of the baseline F-5E. The weight of the wing is then also assumed to be cut by half. The drag coefficient of the CFJ airfoil is assumed to be one-eighth of the baseline airfoil. Table 2 presents the parameters used to estimate the mission analysis. The integral through the mission for the CFJ F-5E is then calculated based on the integral results of the baseline aircraft mission analysis with the assumption that the variation of the penalty or benefit is
CONTROL METHOD USING CO-FLOW JET
309
Table 1 Work potential and fuel consumption of F-5E intercept mission based on Roth’s model” Component
Work potential, hp.min
Propulsion Compressor loss 585 Compressor PR 585 Vcompressor 585 mass flow rate Total engine loss Total engine thrust Total propulsion Airframe Wave skin friction Fuselage drag Tail drag Wing drag Induced drag Structure weight Propulsion weight Fixed equip. weight Stores weight Fuel misc. weight Store drag loss Total airframe loss
Fuel consumption, %
27,108 7.8 89% 53.6 kg/m 113,762 191,529 305,291
8.88
37.3 62.70 100
+
+
55,151 19,268 32,504
18.06 6.32 10.65
26,052 9,116 9,572 3,466 26,061 5,446 186,636
8.53 2.99 3.10 1.14 8.54 1.78 61.11
linear to the baseline results. Table 3 gives the results of the estimated mission analysis. The endurance and range are calculated based on the following formulation~*~:
Table 2 CFJ wing parameters for F-5E-CFJ Parameters
Value
VCtj
CFJ PR Wing span Wing weight CFJ CL wing CFJ Cd wing skin CP
+ wave
60% 4 1/ 2 baseline 1/ 2 baseline 2 baseline 1/8 baseline 0.1
G.-C.ZHA AND C. D. PAXTON
310
Table 3 Estimation of fuel consumption of F-5E intercept mission using recirculating CFJ wing Component Compressor loss Turbine loss Total engine loss Total engine thrust Total propulsion Airframe Wave skin friction Fuselage drag Tail drag Wing drag Induced drag Structure weight Propulsion weight Fixed equip. weight Stores weight Fuel misc. weight Store drag loss Total airframe loss Endurance Range
Work potential, hp.min
Fuel consumption,
CFJ benefit,
%
%
10.48 5.8 39.11 60.89 100
- 1.6 - 0.2 - 1.81
18.06 6.32 0.66
0 0 9.98
22,929 9,116 9,572 3,466 26,061
7.51 2.99 3.10 1.14 8.54
1.02 0 0 0 0
5,446 153,040
1.78 50.13
0 9.17 41.3 37.7
32,003 17,714 119,386 185,905 305,291
- 1.81
0
+
+
55,151 19,268 2,031.5
and
where WOand W1 are the takeoff gross weight and the weight with empty fuel tank, and ct is the specific fuel consumption. This conservative estimation suggests a benefit of 9.17% fuel consumption reduction and also 18% total drag reduction for the whole aircraft. Assuming that the F-5E-CFJ carries the same amount of fuel as the F-5E, the weight ratio of Wo/W1 is increased by 1.8%. Because of the fuel consumption reduction, drag reduction, and the increase in WO/WI, the endurance and range increase by 41 and 37.7%, respectively. The power required to pump the CFJ increases thecompressor loss by 18%, which generates a penalty to the total fuel consumption of 1.8%. The largest gain for the fuel consumption is from the drag reduction of the wing, at 9.98%. The gain from the wing weight reduction is 1.02%. The penalty to the propulsion system is easily offset by the benefits from the reduction of wing drag and structure weight.
CONTROL METHOD USING CO-FLOW JET
31 1
The following formulations are used to calculate the stall velocity, and takeoff (TO) and landing distancesz9:
Using the maximum performance wind-tunnel test results of the CFJOO25065-196 airfoil with C, = 0.29,25928the Vstall will be decreased by 44%, the takeoff distance by 68%, and the landing distance will also be reduced by 68% if it is assumed that the resultant force F 0 . 7 ~is~the same. 111. Conclusions
A novel airfoil flow control technique using a co-flow jet to achieve superior aerodynamic performance for subsonic aircraft has been studied numerically by CFD simulation. The CFJ airfoil opens a slot on the airfoil suction surface near the LE and TE. A high-energy jet is injected tangentially near the LE and the same amount of mass flow is sucked in near the TE. The jet can be recirculated to reduce the energy expenditure of the overall airframe-propulsion system by avoiding dumping of the jet mass flow, or achieving zero net jet mass flow. The turbulent shear layer between the main flow and the jet causes strong turbulence diffusion and mixing under a severe adverse pressure gradient, which enhances lateral transport of energy and allows the main flow to overcome the severe adverse pressure gradient and stay attached at a high angle of attack. The CFJ airfoil achieves significantly higher lift because of augmented circulation. The airfoil does not rely on the Coanda effect at the LE or TE. Hence, the technique can be applied to a modern high-speed thin airfoil, and can be combined with other flow control techniques. The CFD simulation indicates that the CFJ airfoil performance is more sensitive to injection location than to suction location. The injection location should be as close to the LE as possible, but must be downstream of the LE suction peak to make use of the adverse pressure gradient as enhance jet mixing with the main flow. For the NACA2415 airfoil studied, at low AOA with moderate momentum jet coefficient, the CFJ airfoil will not only significantly enhance the lift, but will also dramatically reduce the drag, or even generate negative drag (thrust). The mechanism for this is that the co-flow jet reduces the pressure drag by creating very strong LE suction, and can generate negative pressure drag greater than the friction drag. This may allow the wing to generate both lift and thrust, like a flapping wing, and cruise with very high aerodynamic efficiency. At high AOA, both the lift and the drag are significantly higher than the airfoil with no flow control, and may enhance the performance of takeoff or landing within short distances. Based on the wind-tunnel test results and a conservative estimate, for a subsonic area intercept mission analysis of the military aircraft F-5E, assuming use if a CFJ airfoil, the fuel consumption is reduced by 9%, and the endurance and range
312
G.-C.ZHA AND C. D. PAXTON
by 38 and 41%. Based on the maximum performance wind-tunnel test data, Vstall is reduced by 44%, and the takeoff and landing distances are reduced by 68%. The engine consumes an extra 1.8% fuel, but the whole system, comprising airframe and propulsion, sees a benefit. The CFJ airfoil concept suggested in this paper appears to have the following advantages: 1) very effective in enhancing lift and suppressing separation; 2) dramatically reduces drag and can achieve very high CL/CDat low AOA (cruise), and very high lift and drag at high AOA (takeoff and landing); 3) significantly increases AOA operating range and stall margin; 4) has small penalty regarding the propulsion system; 5 ) can be applied to any airfoil, thick or thin; 6) can be used for the entire flying mission, rather than only during takeoff and landing; 7) can be used for low- and high-speed aircraft; and 8) is easy to implement, with no moving parts. The aforementioned advantages of the CFJ airfoil may derive the following superior aircraft features: 1) requirement for extremely short distances for takeoff and landing; 2) supersonic aircraft to have small wing size matching cruise need, but also have high subsonic performance (e.g., high lift, low drag at M < 1); 3) high maneuverability, high safety, and fast acceleration military aircraft; 4) very economic fuel consumption; 5 ) small wing span for easy storage, light weight and reduced skin friction and form drag; 6) low noise because of no high-lift flap system and reduced wing drag or even wing thrust, which will require less engine nozzle jet velocity; 7) the possibility of heavy lift rotorcraft, essentially with no dynamic stall; and 8) stealth aircraft with no moving control surface.
Acknowledgments The authors would like to acknowledge NASA Langley Research Center (LaRC) for supporting the wind-tunnel tests under contract NNL04AA39C of NRA-03-LaRC-02.25’28We would also like to thank Geoffrey A. Hill at NASA LaRC for the discussion of possible applications of the CFJ airfoil to supersonic aircraft. References ‘Sellers, W. L. I., Singer, B. A., and Leavitt, L. D., “Aerodynamics for Revolutionary Air Vehicles,” AIAA Paper 2004-3785, June 2003. ’Gad-el Hak, M., “Flow Control: The Future,” Journal of Aircraft, Vol. 38, 2001, pp. 402-418. 3Gad-el Hak, M., Flow Control, Passive, Active, and Reactive Flow Management, Cambridge Univ. Press, Cambridge, UK, 2000. 4Anders, S., Sellers, W. L., and Washburn, A., “Active Flow Control Activities at NASA Langley,” AIAA Paper 2004-2623, June 2004. 5Tilmann, C. P., Kimmel, R. L., Addington, G., and Myatt, J. H., “Flow Control Research and Application at the AFRL’s Air Vehicles Directorate,” AIAA Paper 20042622, June 2004. 6Miller, D., and Addington, G., “Aerodynamic Flowfield Control Technologies for Highly Integrated Airframe Propulsion Flowpaths,” AIAA Paper 2004-2625, June 2004.
CONTROL METHOD USING CO-FLOW JET
31 3
’Kibens, V., and Bower, W. W., “An Overview of Active Flow Control Applications at The Boeing Company,” AIAA Paper 2004-2624, June 2004. ‘Modi, V., Fernando, M., and Yokomizo, T., “Drag Reduction of Bluff Bodies Through Moving Surface Boundary Layer Control,” AIAA Paper 90-0298, 1990. ’Cichy, D., Harris, J., and MacKay, J., “Flight Tests of a Rotating Cylinder Flap on a North American Rockwell YOV-1OA Aircraft,” NASA Paper CR-2135, 1972. 10 Englar, R. J., “Circulation Control Pneumatic Aerodynamics: Blown Force and Moment Augmentation and Modifications; Past, Present and Future,” AIAA Paper 2000-2541, June 2000. “Bradley, L. C., “An Experimental Investigation of a Sting-Mounted Finite Circulation Control Wing,” M.S. Thesis, Air Force Inst. of Technology, Wright-Patterson Air Force Base, OH, 1995. ”Wood, N., Robert, L., and Celik, Z., “Control of Asymmetric Vortical Flows over Delta Wings at High Angle of Attack,” Journal of Aircraft, Vol. 27, 1990, pp. 429-435. 13Wood, N., and Robert, L., “Control of Vortical Lift on Delta Wings by Tangential Leading-Edge Blowing,” Journal of Aircraf, Vol. 25, 1988, pp. 236-243. 14Wood, N., and Nielsen, J., “Circulation Control Airfoils-Past, Present, Future,” AIAA Paper 85-0204, 1985. ”Englar, R. J., Trobaugh, L. A., and Hemmersly, R., “STOL Potential of the Circulation Control Wing for High-Performance Aircraft,” Journal of Aircraft, Vol. 14, 1978, pp. 175-181. 16 Englar, R. J., “Circulation Control for High Lift and Drag Generation on STOL Aircraft,” Journal of Aircraf, Vol. 12, 1975, pp. 457-463. 17Smith,A., “High-Lift Aerodynamics,” Journal of Aircraf, Vol. 12, 1975, pp. 501 -530. “Lin, J., Robinson, S., McGhee, R., and Valarezo, W., “Separation Control on High Reynolds Number Multi-Element Airfoils,” AIAA Paper 92-2636, 1992. ”Wygnanski, I., “The Variables Affecting the Control Separation by Periodic Excitation,’’ AIAA Paper 2004-2625, June 2004. ”McManus, K., and Magill, J., “Airfoil Performance Enhancement Using Pulsed Jet Separation Control,” AIAA Paper 97-1971, 1997. ”McManus, K., and Magill, J., “Separation Control in Incompressible and Compressible Flows Using Pulsed Jets,” AIAA Paper 96-1948, 1996. ”Johari, H., and McManus, K., “Visulation of Pulsed Vortex Generator Jets for Active Control of Boundary Layer Separation,” AIAA Paper 97-2021, 1997. 23Whitfield,C., “Airframe System Noise Reduction,” Proceedings of 2nd NASA Vehicle System Program Annual Meeting, July 2005. 24Liu,Y., Sankar, L. N., Englar, R. J., Ahuja, K. K., and Gaeta, R., “Computational Evaluation of the Steady and Pulsed Jet Effects on the Performance of a Circulation Control Wing Section,” AIAA 42nd Sciences Meeting and Exhibit, AIAA Paper 2004-0056, Jan. 2004. 25Zha, G.-C., Carroll, B., Paxton, C., Conley, A., and Wells, A., “High Performance Airfoil with Co-Flow Jet Flow Control,” AIAA Paper 2005-1260, Jan. 2005; also AIAA Journal (submitted for publication). 26Zha, G.-C., and Paxton, C., “A Novel Airfoil Circulation Augment Flow Control Method Using Co-Flow Jet,” NASA/ONR 2004 Circulation Control Workshop, March 2004; also AIAA Paper 2004-2208, June 2004; also AIAA Book Series, Progress in Astronautics and Aeronautics. 27Greitzer, E. M., Tan, C. S., and Graf, M. B., Internal Flow,Cambridge Univ. Press, Cambridge, UK, 2004.
314
G.-C.ZHA AND C. D. PAXTON
”Zha, G.-C., Paxton, C., Conley, A., Wells, A., and Carroll, B., “Effect of Injection Slot Size on High Performance Co-Flow Jet Airfoil,” Journal ofAircrujl (to appear 2006). 29Anderson, J. D., Introduction to Flight, 4th ed. McGraw-Hill Higher Education, New York, 2000. 30Zha,G.-C., (team members Car, D. and Copenhaver, W.), “Super Diffusion Cascades Using Co-Flow Jet Flow Control,” National Research Council Summer Faculty Final Rept. Aug. 2002. 31DeLaurier,J., “Work on Flapping-Wing Flight,” Lecture, 23rd AIAA Applied Aerodynamics Conference, June 2005. 32Roth,B. A., “A Theoretical Treatment of Technical Risk in Modem Propulsion System Design,” Ph.D. Thesis, Department of Aerospace Engineering, Georgia Inst. of Tech., Atlanta, GA, March 2000. 33Roth, B. A., “Aerodynamic Drag Loss Chargeability and Its Implications in the Vehicle Design Process,” AIAA Paper 2001-5236, 2001. 34Roth,B. A., and Mavris, D., “A Method for Propulsion Technology Impact Evaluation Via Thermodynamic Work Potential,” AIAA Paper 2000-4854, 2000. 35Roth,B. A., and Mavris, D., “A Generalized Model for Vehicle Thermodynamic Loss Management and Technology Concept Evaluation,” 2000 World Aviation Conference, Paper 2000-01-5562, Oct. 2000.
Chapter 11
Experimental Development and Evaluation of Pneumatic Powered-Lift Super-STOL Aircraft Robert J. Englar* Georgia Institute of Technology, Atlanta, Georgia
and Bryan A. Campbellt NASA Langley Research Center, Hampton, Virginia
Nomenclature Aj = blowing slot area b = wing span, ft c = chord length, ft cg = center of gravity, ft CD = three-dimensional drag coefficient CDE= equivalent drag coefficient CL = three-dimensional lift coefficient C, = maximum lift coefficient CM25,CM = quarter chord pitching moment coefficient CT = thrust coefficient C, = j e t momentum coefficient [see Eq. ( 2 ) ] C, = jet momentum coefficient, leading-edge blowing CpChw= jet momentum coefficient, Channel-Wing blowing Elev = elevator deflection angle hslot,hj = blowing jet slot height, in. iT = tail incidence angle, deg m = j e t mass flux, slugs/s P d , PD, P , = duct total pressure q = freestream dynamic pressure (= $V2), psf *Principal Research Engineer, Aerospace & Transportation Lab., Georgia Tech Research Institute. Associate Fellow AIAA. 'Senior Aerospace Engineer, Configuration Aerodynamics Branch. Copyright 0 2005 by the authors. Published by the American Institute of Aeronautics and Astronautics, Inc., with permission.
315
316
R. J. ENGLAR AND B. A. CAMPBELL
s = wing area, ft2 S, = ground roll, ft
Td = duct total temperature, OR TR = resultant force, lb T/ W = thrust/weight ratio V = freestream velocity, ft/s v d = deflected slipstream velocity, ft/s vj = blowing jet velocity, isentropic, ft/s W / S = wing loading, psf x/c = nondimensional chordwise location ,,,x = moment center location, ft xTO = takeoff distance, ft a = angle of attack, deg &lipstream = slipstream deflection angle, deg ACL = incremental lift coefficient p = freestream density, slugs/ft3 p = blowing jet density, slugs/ft3
I. Introduction HE ABILITY to achieve Super-STOL (short takeoff and landing) or V/ STOL (vertical/short takeoff and landing) capability with fixed-wing aircraft has been an attractive goal in the aerospace community for over 50 years. The impetus toward its achievement has historically been the numerous benefits associated with very short to zero field length operations of nonrotary-wing aircraft. Although such capability has direct application for military missions such as those of a tilt-rotor or tilt-wing aircraft, there also exists an additional need for simple/reliable/effective personal and business-sized Super-STOL or VSTOL aircraft operating from remote or small sites, as well as increasingly dense urban environments. The development of simple, efficient aeropropulsive technology and corresponding low-speed control systems to make this possible is a goal that now seems achievable because of technical breakthroughs in pneumatic and powered-lift aerodynamic technologies. This chapter, originally presented at the NASA/ONR CC Workshop in March 2004 (see NASA CP 2005-213509, 2005), will discuss recent progress in the integration of high-lift, propulsive, and control systems, all employing common pneumatic techniques using circulation control (CC) blowing, into a promising Super-STOL configuration. Two promising technologies to evolve from earlier STOL/VSTOL research are the Custer Channel Wing powered-lift configuration and the circulation control wing (CCW) pneumatic high-lift concept. Through innovative use of the propeller slipstream, the Channel Wing airplane developed by Willard Custer (Fig. l)i-3 was able to achieve significant lift coefficient and efficient downward thrust deflection without varying the high-lift configuration geometry. This powered-lift technology, tunnel-tested by NACA in 1953,l and then flighttested and further developed by Custer in the mid-1960~,~ employed the Channel Wing concept shown in Fig. 2.3 In essence, the propeller located at the very trailing edge (TE) of the 180-deg arc circular channel in the wing
T
PNEUMATIC POWERED-LIFT SUPER-STOL AIRCRAFT
317
Three-view of the Cuetor Channel Wing CCW-5. (Author’r Collection)
Fig. 1 Three-view and in-flight photo of 1960s Custer Channel Wing Aircraft.’-3
further increased the velocity over the channel’s upper surface and augmented the circulation and lift there in much the same manner as a deflected flap, but perhaps to a greater extent. Lift was also augmented by the deflected slipstream behind the channel such that
In-flight lift coefficients nearing 5 were generated by thrust coefficients also nearing 5 as demonstrated by C ~ s t e r .However, ~ the flight-tested Custer Channel Wing aircraft demonstrated a number of drawbacks associated with low-speed handling, cruise drag, stability and control, high-incidence operation, and one-engine-out scenarios, including the following: 1) Much of the high CL was from redirected thrust, and less from circulation lift augmentation. 2 ) High cruise drag could result from the channel’s extra surface area. 3) Asymmetric thrust yields asymmetric moments and instability.
318
R. J. ENGLAR AND B. A. CAMPBELL
Airfoil Surface in Channel; Replace with New Pneumatic AirfoilsiAA Turning Surfaces Fig. 2 Basis of the Channel Wing concept and current pneumatic improvements.
4) Channel LE and TE separation could occur at high angle of attack a. 5 ) Poor low-speed control is available from conventional aerodynamic surfaces at low speeds. 6) There is nose-down pitch from aft propeller loading on the wing. 7) There is nonuniform flow around the prop at high a. 8) There is poor lift/drag ratio. 9) High angle-of-attack operation could cause poor visibility and control. 10) There are one-engine-out control problems. To alleviate these shortcomings, preliminary research has been carried out at Georgia Tech Research Institute (GTRI), where investigation in adapting CC pneumatic technology has been made (Fig. 3 and Refs. 4 and 5 , for example) to dramatically improve the Channel Wing configuration. As Fig. 2 shows, the new pneumatic configuration thus developed combines blowing on curved surfaces at the channel TE to greatly augment the lift and thrust deflection without using high angle of attack. It also employs blown CCW technology on the outboard wing panels to further augment lift and low-speed controllability while providing additional drag when needed for slow-speed approaches down steep glide slopes for Super-STOL. This channel thrust turning and lift augmentation are based on the CCW/ upper surface blowing (USB)concept of Fig. 4, where tangential blowing on a highly curved TE behind a jet engine augments flowfield entrainment, increases circulation, and deflects thrust to add more incremental lift. Thrust deflection angles of 165 deg produced by blowing were measured experimentally on wind-tunnel This concept provides pneumatic STOL, VSTOL, and thrust-reversing capabilities without any moving parts. Circulation Control Wing alone (Fig. 3) employs a similar tangential-blowing configuration, but without the pneumatic thrust deflection. Such CCW airfoils have generated
PNEUMATIC POWERED-LIFT SUPER-STOL AIRCRAFT
319
TANGENTIAL BLOWlNG OVER ROUNDED TRAILING EDGE SURFACE
Fig. 3 Basics of circulation control pneumatic technology.
measured two-dimensional lift augmentations of 80 times the input blowing When flight-tested on an A-6 flight demonstrator, CCW showed a 140% increase in useable high lift, employing only half of the bleed air available from the aircraft’s standard turbojet engines.* Figure 2 shows how these blown flow-entrainment devices would be arranged to enhance the effectiveness of the Pneumatic Channel Wing (PCW) configuration. In addition, the CCW lift capability can be applied differentially outboard to generate very large rolling and yawing moments, which are essential for controlled flight at the very low speeds of Super-STOL. Based on earlier CCW/USB wind-tunnel and full-scale data (Fig. 4)6,7and CCW flight-test data from the A-6 STOL-demonstrator program,8 the predicted lift and drag capabilities for the pneumatic channel wing configuration were expected to offer great Super-STOL promise. Reference 9 details these early
mu0 llvwt de?wilm
sl
OCHllisBrlnd
tumdmodd
Fig. 4 Previously developed CCW/upper surface blowing powered-lift concept:
320
R. J. ENGLAR AND B. A. CAMPBELL
predictions before the current wind-tunnel test data were available. These implied CLvalues approaching 9- 10 for a pneumatic channel wing aircraft with blowing on outboard CCW wing panels at relatively low aircraft angle of attack. Higher CLvalues were possible at higher thrust coefficients if higher a values were used because of the additional vectored thrust component. Again, for comparison, the Custer channel wing aircraft generated in-flight CLof 4.9, whereas a conventional slotted flap on this wing geometry would generate CL values at 2-3. Initial takeoff predictions’ showed that these PCW capabilities could produce very short, hot-day takeoff ground rolls for typical mission weights, and even zero ground roll under certain conditions. As part of an ongoing program for the NASA Langley Research Center to develop this PCW concept, GTRI and NASA have teamed together in an experimental development program being conducted at GTRI, which has provided aerodynamic and propulsive data input for design studies being conducted at both NASA and GTRI. The current paper will summarize these experimental results and discuss effects deriving from variations in PCW geometry, propeller thrust, and channel blowing.
11. Experimental Apparatus and Test Techniques A wind-tunnel development/evaluation program was conducted at GTRI on a generic twin-engine Super-STOL-type transport configuration (Fig. 5) using the 0.075-scale semispan model shown in Fig. 6. A variable-speed electric motor was installed in the nacelle, which could be located at various positions in the channel, and which drove interchangeable two-, three-, or four-bladed propellers of various diameters and pitch. Also variable was the height of the blowing slot located at 95% of the channel chord length, as well as the blowing momentum coefficient and portions of the slot arc length that were blown. Behind the slot, the rounded TE curved only 90deg (rather than the more conventional 180 deg of typical CCW configurations) for an anticipated maximum thrust deflection of approximately (90 deg a). It was already known (Fig. 4) that thrust deflections up to 165 deg yielded by blowing were a
+
Fig. 5 Conceptual PCW Super-STOL transport configuration.
PNEUMATIC POWERED-LIFT SUPER-STOL AIRCRAFT
32 1
Fig. 6 PCW/CCW semispan model installation in GTRI Model Test Facility research tunnel (three-bladed prop with unblown outboard CCW), plus jet flow turning in channel (black tufts).
possibility. Here, the momentum coefficient is defined as
This semispan model configuration (Fig. 6 ) was mounted on an underfloor balance with air supplies and automated pitch table in the GTRI Model Test Facility 30 x 43 x 90 in. test section. The tunnel wall boundary layer near the test section floor was eliminated by use of tangential floor blowing. In a follow-on version of this configuration, both the LE and the TE of the outboard CCW wing section were also blown for separation control. The emphasis in the following data is on the performance of the inboard blown PCW configuration, but performance of the outboard CCW sections to further augment lift is also shown. 111. Wind-Tunnel Evaluations and Results
Test techniques employed in the subsonic tunnel evaluation of this pneumatic powered-lift model are similar to those employed and described in Refs. 10 and 11 for blown airfoil and semispan models, except that special additional techniques were employed to account for the installation of the active propeller in the channel (see below). Some 980 wind-tunnel runs (including propeller calibrations) have now been conducted during three test programs at GTRI to develop these blown-configuration geometries and to evaluate their aeropropulsive, flight-trim, and control characteristics. A typical run consisted of a sweep (incremental variation) of prop thrust or blowing pressure at constant angle of attack and wind speed. Also, angle-of-attack sweeps or dynamic pressure (velocity) sweeps were run at constant thrust and blowing coefficients CTand C,. Numerous runs were made with varying tail configurations to evaluate pitch trim and control. Typical test results are presented in Secs. 1II.A-1II.C to demonstrate how these various parameters affected overall performance.
322
R. J. ENGLAR AND B. A. CAMPBELL
A. Tunnel Test Results, Outboard Wing ON In Figs. 7a and 7b are shown the effects on lift and drag coefficients of blowing the channel TE without the prop installed (i.e., CT = 0), but with the engine nacelle in place (Fig. 6). Note the ability of the blowing to more than double the CLmaxof the unblown configuration with virtually no reduction in the stall The C, values shown are comparable to or greater than those that angle, astall. would normally be generated by more complex moving mechanical flaps. Note also the ability of the blowing at a = 0 deg to increase CL by a factor of nearly 10 over the unblown value. At a = 0 deg, blowing at C, = 0.30 yields 50% more C, than the C,, of the unblown configuration. In Fig. 7b, the drag polars at constant C, are typically quadratic in CL. Earlier in a than where the stall begins, they follow essentially the same single curve, using blowing to progress to each successively higher C, region. Addition of the propeller to the channel brings into play the powered-lift characteristics of the PCW configuration. Figure 8, for a = 0 deg, shows the variations in C, and C, with thrust coefficient CT for fixed values of blowing coefficient. Here, in order to recognize the direct thrust component to lift and drag, thrust coefficient is defined as CT = T/(qS),where T is the calibrated uninstalled wind-on prop-alone (not-in-the-channel) thrust at the proper advance ratio that is, representative test dynamic pressure q. The reference area S is the wing semiplanform area. These thrust values were determined prior to installation in the channel by testing the prop alone in the tunnel at various rpm and tunnel speeds. Calibration curves of T (thrust) against rpm were input to the data reduction program at given test wind speeds. CT, C,, and C, are directly comparable on
Fig. 7 Measured blown lift and drag capabilities of the PCW model without the propeller installed: a) Lift vs a,b) Lift-drag polars.
PNEUMATIC POWERED-LIFT SUPER-STOL AIRCRAFT
323
Channel Wing CT
Fig. 8 Effects of prop thrust variation on lift and drag at constant blowing (C,) and (Y = 0 deg.
a common reference basis to determine force contributions from installed thrust. This avoids the difficulty that would be caused by using the standard helicopter thrust coefficient, based on rotor (or prop) geometry rather than wing area. Also, note that measured CDincludes the input thrust, which cannot reasonably be separated from the aerodynamic drag alone once the prop is in the channel. Measured CD can therefore be (and sometimes is) negative. After the initial low values of CT are exceeded, CLincreases nearly linearly with CT,and CDreduces nearly linearly. (This implies that, at a constant C,, the thrust deflection angle is nearly constant.) Figure 9 shows that incremental lift augmentation as a result of blowing (C,) is much greater than that resulting from CT (Fig. 8). Here, at CT = 2.2, the blown configuration generates CL of approximately 8.5 at a = 10 deg. The flight-tested Custer Channel Wing3 generated roughly one-third this CL at this CT, but also required a = 24-25 deg. Note also that increased blowing at a constant CT yields increased drag (rather than thrust recovery), which can be quite essential for Super-STOL approaches and short landings. These lift comparisons in Figs. 8 and 9 show that lift increases more efficiently by increasing blowing than by increasing thrust. In Fig. 10 a plot is shown of the variation in lift and drag with angle of attack for the blown powered-lift configuration in comparison with the unblown baseline configuration without the prop. Here, flow visualization showed that the initial stall ( a = 15-17 deg) seen for most of the lift curves
324
R. J. ENGLAR AND B. A. CAMPBELL
Channel Wing Ck
Fig. 9 Effects of blowing variation on lift and drag at constant CT and (Y = 10 deg.
corresponded to stall of the outboard unblown wing section, whereas the blown channel wing section then continued on to stall angles of 40-45 deg and C, values of 8.5-9. Note that CD (including thrust) increases from negative to positive values as incidence increases. Figure 11 shows the effect on lift and drag of increasing the circular arc length of the blown slot around the channel at a given prop longitudinal location ( x / c = 0.95), where the maximum slot arc of 160 deg was most effective. Blowing of more than 160 deg of channel arc was not appropriate on this model because the last 20 deg of inboard arc was along the channel right next to the fuselage, and blowing there would do little more than bounce off the fuselage. The effect on increased tail-off pitching moment caused by suction loading on the aft of the channel (either by blowing, prop slipstream, or both) is shown in Fig. 12 as a function of CT and C,, all at a = 0 deg. These moments are referred to the channel’s quarter-chord location (c/4), and confirm the typical trend of this type of blown configuration: large nosedown CM,which, although does make the aircraft much more stable longitudinally, causes problems with pitch trim. It is for this reason that additional experimental evaluations were conducted tail-on to investigate increased longitudinal trim capabilities. All data presented so far have been tail-off. A second investigation was conducted with LE blowing installed on the outboard wing CCW portion to provide counteracting nose-up pitch for trim, as well as for LE separation prevention.
PNEUMATIC POWERED-LIFT SUPER-STOL AIRCRAFT
Angle of Attack,
a, dcg
Angle of Allack,
325
a, deg
Fig. 10 Effects of blowing, Ch and a on lift coefficient, stall angle, and drag coefficient for the PCW model with unblown outboard wing.
Fig. 11 Effects on lift and drag of varying blown channel slot arc length at constant C, and at a = 0 deg: a) Prop and nacelle installed and b) Prop off, but nacelle installed.
326
R. J. ENGLAR AND B. A. CAMPBELL
Channel Wing Ck or CT
Fig. 12 Effects of prop/nacelle location, blowing and thrust on quarter-chord pitching moment, (Y = 0 deg.
B. Tunnel Test Results, Channel Wing Only Higher nondimensional thrust coefficient values were available when the channel-only configuration was tested (fuselage, blown channel and prop, but with no outboard CCW panels), because the reference planform area of the wing was also reduced. This allowed CT of x 3 for the channel-only vehicle, and, as Fig. 13 shows, lift coefficients nearing 11 were measured with a conventional horizontal tail installed at the midvertical location on the aft fuselage. Needless to say, not all of the lift values shown in Fig. 13 are trimmed longitudinally. Furthermore, for the CT= 3 case with blowing on, the conventional tail of the aircraft stalled experimentally over much of the lower a range. The possible inability to trim these Super STOL aircraft longitudinally has been highlighted as a problem of blown systems in Refs. 7 and 8. It is further emphasized in Fig. 13, where the large suction on the aft-loaded blown channel (and blown wing, if present) produces very large nosedown pitching moments (compared to the CT = 0, C, = 0, tail-off curve). Although this can produce improved longitudinal stability, these moments must also be trimmed. Horizontal tail investigations were conducted as part of this three-dimensional model development plan in the hope of determining tail location and configuration to provide enough nose-up pitch to trim the vehicle. Several horizontal tail configurations [one without an elevator, a second with a 20-deg up elevator (aelev= +20 deg), and a third with an inverted leading edge droop] were designed and fabricated. As Fig. 14 shows, these could be mounted on a vertical
PNEUMATIC POWERED-LIFT SUPER-STOL AIRCRAFT II
327
Reuma~cChannel WmgONLK, Outboard CCW OFF,RODON,Wd Tad
a,d q r c n
Fig. 13 Effect of thrust and/or blowing increase on lift and pitching moment variation with a for channel-wing-only configuration (no outboard wing panels) with tail at midlocation, iT = 0 deg.
center plate yielding variation in both tail incidence (iT) and vertical position in the propeller slipstream. High, midfuselage, and low-tail positions were tested. Testing of these tail-on configurations over a range of tail parameters revealed that a low-tail position immersed in the prop slipstream and dynamic pressure was more effective than the higher tail (Fig. 15), but the lower tail also experienced more LE stall for the same reason. This tail stall prevents the vehicle from being trimmed at this higher blowing condition (here with the outboard CCW wing on again). Considerable videotaping of flow visualization tufts on the tail revealed these problem areas and led to the development of the inverted-droop (drooped upward) LE modification for the tail. Keeping the tail LE attached allows positive nose-up pitch and thus trim to be generated for the vehicle over a much wider range of lower a values. For the channel-wing-only model with the modified tail, trimmed CL values greater than 9 are therefore seen (Fig. 16), but much of these data are still untrimmed, and again the low tail with no LE modifications is fully stalled. Thus, these data imply that further tail development (perhaps including LE blowing to prevent the tail stall without mechanical LE fixes) is needed to trim in this high CL range at all vehicle angles of attack.
C. Tunnel Test Results: Flow Attachment An additional series of flow visualizations was conducted to further identify means to prevent separated flowfields on the wing during high-lift generation. Figure 17 data show that the flow at the channel LE is entrained to the
328
R. J. ENGLAR AND B. A. CAMPBELL
a)
Fig. 14 Horizontal tail configurations evaluated, with outboard CCW wing panel on: a) high tail, b) low tail, and c) mid-fuselage tail.
point where LE separation is prevented until a = 35-40 deg or more, but that the outboard CCW is prone to stall there. Leading-edge blowing on this outboard CCW wing panel greatly entrained this flowfield as well. Figure 18 flow visualization shows this severe separation at a = 20 deg for the unblown case (Fig. 18a), whereas blowing the LE completely reattached the flowfield there. An additional means of trim and control was investigated for the PCW. Here, these large nosedown pitching moments (seen in Fig. 13, 15, and 16) are offset by moving the aircraft center of gravity aft to trim, with no tail installed. Aft centerof-gravity movement was previously performed for flight tests of the A-6/CC Wing aircraft, but with the tail on.8 Figure 19 shows data for the C T = 3 case of a tail-less PCW without outboard wing. At C, = 0, moving the centerof-gravity aft from x / c = 0.25 to 0.375 gives the aircraft neutral longitudinal stability but does produce trim over most of the angle of attack range. Similar reduction in pitching moment can be produced by aft center-of-gravity shift as blowing is increased (Fig. 19b), but this requires further aft center-of-gravity to trim at lower a, and the C, vs. C, curves are now unstable (dC,/dC, = positive). Some small control surface (such as a blown canard to provide noseup pitch and positive lift to trim) could perhaps be incorporated with a state-of-the-art control system and control laws to make this a feasible pitchtrim device without lift loss due to tail download.
PNEUMATIC POWERED-LIFT SUPER-STOL AIRCRAFT MTF063, ci4 Pitching Moment, Prop ON, N o Outboard Blo-g,Tsil
Power
329
CT-2.2,CmuChW-1.0 , CCW R a p 4 " Alpha sweeps
Qusncr-Chord Pitching Moment, C
~ e / 4
Fig. 15 Comparison of high and low tail position on PCW configuration with unblown outboard wing and horizontal tail.
Tad Effects. PneYmshF Channel WxngONLY, OutboardCCW OFF, Rop ON
Tad Effects. Pneumahc Channel WmgONLY. Outhard CCW OFF, Rap ON
Fig. 16 Comparative lift and quarter-chord pitching moment coefficients of PCW, no outboard CCW, with and without tail LE modifications.
330
R. J. ENGLAR AND B. A. CAMPBELL Outboard CCW Blowing EffectsCCW Flap=O', Tail OFF
Alpha, deg
Fig. 17 Leading-edge blowing and channel flow entrainment prevent flow separation over both channel and outboard CCW leading edges.
Fig. 18 Flow attachment caused by LE blowing on outboard CCW and channel flow entrainmentat (Y = 20 deg, channel LE not blown: a) Outboard LE slot unblown and b) outboard LE slot blown.
PNEUMATIC POWERED-LIFT SUPER-STOL AIRCRAFT
33 1
MTF068, Pneumatic Channel Wng, Phase 111, CL vs CM, Run 799, Cp Channel=O.O, Channel Only, Prop ON, CT.3.0. Tail OFF
a) 9
8
7
6
d 5 4
3
2
1
0
1
0 CM
1
2
10 9 8
7
6 5 4
3 2
I_____
+xmom=.375 I-+Xmom=.SO
I_ _.______ _ _I.______ __. __ __________ I...._________________I-__ ____________ __. ...
1 0
2
1
0
1
CM
Fig. 19 Effect of aft center of gravity location on pitching moment curves for the = x,.Jc: a) CT = 3, C, = 0, tail-less PCW at CT = 3 and two blowing values, x, b) C T = 3, C, = 0.3.
IV. Comparison of Measurements and Predictions In Fig. 20 are compared the results of these investigations with previously predicted lift and drag data, which were estimated from existing CCW/USB wind-tunnel data and from A-6/CCW flight-test data. Whereas the prop/electric
R. J. ENGLAR AND B. A. CAMPBELL
332 a)
lC 9
.
.
.
,
,
.
.
Solid Line Data Derived from CCWNSB, a1 Tunnel Data from GTRl MTF055
.
,
a=lW
8 7
cL 6 5 4
3
cl=fl.fl
-
2
I 0
::y
Solid , Line ,Data Derived , ,from CCWIUSB, , , Tunnel Data from GTRI MTF055
1 0.1
0.2
0.3
0.4
0.5
CCChW
0.6
0.7
0.8
0.9
1.0
9
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
a=lW , 0.8
,I
0.9
1.0
%hW
Fig. 20 Comparisons of predicted and experimental PCW lift and drag data at constant Ch a = 10 deg, outboard CCW on: a) Measured (symbols) vs predicted lift (no symbols) and b) measured (symbols) vs predicted (no symbols) drag.
motor currently available did not allow higher CTvalues than about 2.2 (outboard wing on), these lower-thrust, wind-tunnel data considerably surpass the predicted lift data (Fig. 20a). If the ratio of measured-to-predicted holds linearly up to CT = 10, then C, values over 14 are to be expected at a = 10 deg. The experimental drag data (Fig. 20b) are similar to the predicted values at lower
NOTE: -17,000 Ib Increase in TO Gross Wt over Baseline Tilt Rotor at 100 ground roll
Fig. 21 Pneumatic channel wing predicted super-STOL takeoff performance.
PNEUMATIC POWERED-LIFT SUPER-STOL AIRCRAFT
333
C, but show less drag than predicted at higher blowing. These estimated data have been used to predict Super-STOL takeoff distances on a hot day at 3000 ft altitude to be less than 100 ft and, in some instances, 0 ft. (Fig. 21).9 The composition of measured and predicted results in Fig. 20 seems to suggest that even better takeoff performance might be obtained (higher lift, lower drag). However, the lower measured drag values indicate that additional attention will need to be paid to obtaining greater drag values for steeper glide slopes on STOL approaches (when desired and chosen by the pilot). V. Potential Applications Design and mission studies conducted at NASA LaRC based on the preceding tunnel data have led to consideration of several new pneumatic powered-lift PCW-type configurations. The capability of the PCW to significantly augment lift, drag, and stall angle to the levels reported herein demonstrates that this technology has the potential to enable simple/reliable/effective STOL and possibly VTOL operations of personal and business-sized aircraft operating from remote or small sites as well as increasingly dense urban environments. Such capability now opens the way for alternative visions regarding civilian travel scenarios, as well as both civilian and military aerial missions. One such vision is represented by the Personal Air Vehicle Exploration (PAVE) activity at NASA Langley Research Center. Another vision, a military Super-STOL transport, is discussed in the mission study of Ref. 9 and Fig. 21.
VI. Conclusions Results from subsonic wind-tunnel investigations conducted at GTRI on a 0.075-scale powered semispan model of a conceptual PCW transport have confirmed the potential aerodynamic payoffs of this possible Super-STOL configuration, including very high lift and overload capability. These results include the following features. Lift and drag augmentations and/or reductions as desired for Super-STOL operation have been confirmed, with C, = 9 measured at a = 10 deg (C, = 10-11 at higher a), and drag coefficient (including thrust) varying between -2 and +2, depending on blowing and thrust levels. C, values nearing 14 are predicted if higher CT is available, say on takeoff. Blowing C, and thrust CT variations were both found to enhance circulation, thrust deflection significantly, and lift. However, if evaluated as incremental lift per unit of input thrust or momentum (C, or C,), blowing was far more efficient than thrust. By varying only C, and/or C,, all the aircraft’s aerodynamic characteristics (forces and moments) can be augmented or reduced as desired by the Super-STOL aircraft’s pilot or its control system without mechanical moving parts (such as tilting rotors or wings) and without resorting to high a to acquire larger vertical thrust components for lift. The blown channel wing itself, without thrust applied, was able to double the C , capability of the baseline aircraft configuration, and multiply its lift at a = 0 deg by a factor of 10. Addition of blowing on the outboard CCW section can increase this further, and can also add drag as needed for SuperSTOL approaches.
334
R. J. ENGLAR AND B. A. CAMPBELL
Even with the unblown outboard wing stalling at a = 15-17 deg, the blown and thrusting channel continued to increase lift up to a stall angle of 4045 deg as a result of channel flow entrainment. Although this high a may not prove practical as a takeoff/landing operational incidence, it does show significant improvement over the asymmetric LE separation of the conventional channel wing’s stalled channel and the resulting low-speed control problems. Pneumatic Channel Wing conversion of thrust into either drag decrease or drag increase without moving parts is also quite promising for STOL operation. Large nosedown pitching moments are produced by these blown configurations, and thus longitudinal trim capability needs to be addressed in future evaluations. Unlike a tilt rotor, in Super-STOL or V/STOL there is no download on the wing from prop thrust because the PCW props do not tilt. The potential of PCW for an integrated lift, thrust/drag interchange, and control system, all from one set of devices, holds promise in terms of simplicity, weight reduction, and reliability /maintainability. The projected operational benefits based on these early data suggest SuperSTOL and possible V/STOL capability with significantly increased payload, reduced noise signatures, and increased engine-out control, all without variable geometry or mechanical engine/prop tilting. A PCW aircraft thus equipped could provide a simpler, less costly way of achieving Super-STOL or V/STOL capability without the complexity, weight, or reliability issues of rotating the propulsion system, carrying large engines and rotors on the wing tips, or thrusting downwards on fixed wings during hover. Additionally, the integration of pulsed-blowing technology with circulation control (currently being investigated)12 may further increase lift efficiency and reduce already low blowing requirements by up to 50% or more, while further enhancing stability and control. Successful application of these results can lead to positive technology transfer to personal, business, and military-sized aircraft. In addition to the military Super-STOL transport discussed in Fig. 21, these experimental data and pneumatic technology results have been included in preliminary design studies of other possible pneumatic powered-lift configurations, including smaller personal and business-type aircraft. Future testing, evaluation, and development still need to be accomplished to address possible pitch-trim problems, performance at higher CT and lower C,, and associated stability and control. In the future, the existing model or larger three-dimensional models should be modified to include blown tail surfaces and additional improvements to the pneumatic thrust deflection system. The following should be experimentally investigated: 1) Use of pulsed blowing to further reduce required blowing mass flows (both inboard on the channel and outboard on the CCW). 2) Higher propulsor solidity for greater thrust and powered lift, or improved propeller characteristics for greater CT availability. 3) Further evaluation of low-speed controllability and trim, including evaluation of improved tail surfaces, which might even be blown to reduce tail area and drag. 4) Further evaluation of low-speed controllability and trim by novel aerodynamic/pneumatic trim and control devices (blown canards, for example).
PNEUMATIC POWERED-LIFT SUPER-STOL AIRCRAFT
335
The earlier mission analyses should be revised to incorporate the experimentally developed aeropropulsive and stability and control characteristics of the PCW concept. If the projected benefits are confirmed, and further benefits come to light, then larger-scale, higher-Reynolds-number testing on a full three-dimensional PCW model with variable yaw capability should be conducted to facilitate greater strides toward this pneumatic powered-lift technology’s maturation.
Acknowledgments The primary author would like to thank personnel of the NASA Langley Research Center for their ongoing support of this powered-lift research at GTRI. References ‘Pasamanick, J., “Langley Full-Scale-Tunnel Tests of the Custer Channel Wing Airplane,’’ NACA RM L53A09, April 1953. ’Mitchell, K. A., “Mr. Custer and His Channel Wing Airplanes,” Journal of American Aviation Historical Society, Spring 1998. 3Blick, E. F. and Homer, V., “Power-on Channel Wing Aerodynamics,” Journal of Aircraft, Vol. 8, No. 4, 1971, pp. 234-238. 4Englar, R. J., “Circulation Control Pneumatic Aerodynamics: Blown Force and Moment Augmentation and Modification; Past, Present and Future,” AIAA Paper 20002541, AIAA Fluids 2000 Meeting, Denver, CO, June 19-22, 2000. ’Englar, R. J., and Applegate, C. A., “Circulation Control-A Bibliography of DTNSRDC Research and Selected Outside References (Jan. 1969 through Dec. 1983),” DTNSRDC-84/052, Sept. 1984. 6Englar, R. J., “Development of Circulation Control Technology for Powered-Lift STOL Aircraft,” NASA CP-2432, Proceedings of the 1986 Circulation Control Workshop. ’Englar, R. J., Nichols, J. H., Jr., Harris, M. J., Eppel, J. C., and Shovlin, M. D., “Development of Pneumatic Thrust-Deflecting Powered-Lift Systems,” AIAA Paper 86-0476, AIAA 24th Aerospace Sciences Meeting, Jan. 1986. 8Pugliese, A. J. (Grumman Aerospace Corporation), and Englar R. J. (DTNSRDC), “Flight Testing the Circulation Control Wing,” AIAA Paper 79-1791, AIAA Aircraft Systems and Technology Meeting, Aug. 1979. ’Hines, N., Baker, A., Cartagena, M., Largent, M., Tai, J., Qiu, S., Yiakas, N., Zentner, J., and Englar, R. J., “Pneumatic Channel Wing Comparative Mission Analysis and Design Study, Phase I,” GTRI Technical Rept., Project A-5942, March 2000. “Englar, R. J., and Williams, R. M. “Test Techniques for High Lift Airfoils with Boundary Layer and Circulation Control for Application to Rotary Wing Aircraft,” Canadian Aeronautics and Space Journal, Vol. 19, No. 3, 1973, pp. 93-108. “Englar, R. J., Niebur, C. S., and Gregory, S. D., “Pneumatic Lift and Control Surface Technology for High Speed Civil Transport Configurations,” Journal of Aircraft, Vol. 36, NO. 2, 1999, pp. 332-339. ‘’Jones, G. S., and Englar, R. J., “Advances in Pneumatic-Controlled High-Lift Systems Through Pulsed Blowing”, AIAA Paper 2003-341 1, AIAA 21st Applied Aerodynamics Conference, June 2003.
Chapter 12
Use of Circulation Control for Flight Control Steven P. Frith* and Norman J. Woodt University of Manchester, Manchester, England, United Kingdom
Nomenclature b = span, mm
CD = drag coefficient CL = lift coefficient CLc0,= initial lift coefficient Cl = rolling moment coefficient Cl(o)= initial rolling moment coefficient C , = pitching moment coefficient c = chord, mm c = standard mean chord, mm E = mean aerodynamic chord (MAC), mm c, = root chord, mm c, = tip chord, mm D = drag, N h = slot height, mm L = lift, N 1 = rolling moment, Nm M = Mach number m = pitching moment, Nm riz = mass flow rate, kg/s p = rate of roll pW = freestream pressure, Pa q = dynamic pressure r = trailing edge radius, mm *Postgraduate Research Student, Fluid Mechanics Research Group, Aerospace Engineering. Member AIAA. 'Professor, Head of Department, Aerospace Engineering. Senior Member AIAA. Copyright 0 2005 by the authors. Published by the American Institute of Aeronautics and Astronautics, Inc., with permission.
337
S. P. FRlTH AND N. J. WOOD
338
S = reference area, m 2
vj =jet velocity, m/s V , = freestream velocity, m/s a = angle of attack, deg p = orifice diameter to internal pipe diameter ratio aC,/aC, = lift augmentation aC,/at = rolling moment derivative with respect to aileron deflection A = wing sweep angle, deg p = density 5 = aileron deflection, deg Subscripts c/4 = quarter-chord position
D = drag j =jet L = lift L = left jet on full-span model 1 = rolling moment LE = leading edge m = pitching moment max = maximum plenum = associated with plenum parameters R = right jet on full-span model total = combined left and right jets trim = trim condition p = associated with blowing parameters 6 = aileron deflection co = freestream
I. Introduction IRCULATION control (CC) has been recognized as a technique by which very high lift coefficients can be achieved without the use of mechanical control devices. It exploits the Coanda effect by blowing a high-velocity jet over a curved surface, usually a rounded or near-rounded trailing edge (TE), causing the rear stagnation point to move. In turn, the upper surface boundary layer is energized, resulting in a delay in separation. As the circulation for the entire wing is modified, there is an increase in overall lift, often much greater when compared to more conventional mechanical lift devices. Earlier researchla2has been focused mainly on two-dimensional unswept wings, where the flow is predominantly attached to the airfoil. However, in this work the performance benefits of the application of CC to a low aspect ratio (AR) wing have been investigated. A delta-wing planform was chosen, because the regions of separated flow could reveal additional properties of the technique. Although more recent work3 uses pulsed jets in a bid to reduce the total jet mass flow rate required, a steady jet was used in this investigation for
C
USE OF CC FLIGHT CONTROL
339
model simplicity. With a system with few or no moving parts, the circulation control wing (CCW) has generated considerable interest, as it is mechanically simpler, and therefore cheaper to manufacture, and less prone to mechanical failure in comparison with conventional high-lift devices. Also, lift increments can be similar to those with conventional high-lift control surfaces, but pitch increments can be lower, leading to improved aircraft control. The initial aim of the study was to investigate the effect of various TE configurations with a view to eliminating the cruise drag penalty attributed to large TEs, while still obtaining high lift augmentation. The lift augmentation is calculated as the ratio of increase of lift coefficient with blowing. A half-span cropped-delta model was used to perform a parametric study of TE geometries. This was then extended to a sting-mounted CC demonstrator consisting of a generic unmanned air vehicle (UAV) planform with control surfaces with TE sweep to determine whether there would be an interaction between the two jets and also whether CC could be used for roll control, within the limits of pitch trim and maintaining high lift augmentation. A lift augmentation of approximately 20 was achieved over all the blowing ranges tested, with a maximum lift augmentation of 53 recorded. Nosedown pitching moments were experienced, with a roll authority associated with these measurements. Roll of the aircraft was possible with differential blowing of the CC systems.
11. Half-Span Cropped-Delta Model
A. Experimental Procedure For the preliminary studies to investigate a means of optimizing the CC system, a half-span model was used to represent a circulation control wing (CCW). A schematic of the model is shown in Fig. 1. The CCW consisted of a generic delta-wing LE section and a plenum/TE section, forming a cropped delta-wing planform when connected. The LE section comprised a sharp LE profile with a 50-deg LE sweep angle, incorporating strengthening sections along the wing root to reduce flexing when under aerodynamic load. As shown in Fig. 2, the plenum section was manufactured using 2-mm-thick brass sheet for the lower surface and 3-mm-thick aluminum sheet for the upper surface. The TE consisted of a 6-mrg-diam brass rod, giving a TE radius-to-mean-aerodynamic-chord ratio of 0.005C. A narrow convergent slot provides the jet blowing. This was achieved by manufacturing a “knife-edge” section that ensured that there was a contraction within the plenum section to ensure the exiting fluid would attach to the Coanda surface. This was constructed from aluminum and had a spanwise extent of 500mm, dictating the length of the slot. This was incorporated into the top plate of the plenum section. A series of push-pgl screws allowzd the slot height to be adjusted to 0.15 and 0.3 mm (0.00025C 5 h 5 0.0005 C). The root chord was 853 mm and the tip chord was 254 mm, resulting in an average chord measurement of 553.5 mm. The half-span measurement was 500 mm, with the slot extending the full distance, although because of the positioning of the splitter board, an inboard 5-mm section of the slot length was
S. P. FRlTH AND N. J. WOOD
340 a)
,Rectangularto I round transition
I
Mounting Strut
, ' \
Pressure taps
\
\ CCW trailing edge
Fig. 1 Half-span cropped-delta model geometry: a) Upper surface view, b) crosssectional view.
permanently sealed off. The aspect ratio of the wing was calculated using b2/S, giving a value of approximately 1.7. The model was mounted from the overhead balance in the Avro 2.74 m x 2.13 m (9 x 7 ft) wind tunnel at the Goldstein Laboratory, Manchester, UK, as shown in Fig. 3. A splitter board was mounted to ensure that the windtunnel boundary layer did not interfere with measurements and the Coanda jet. Force and moment data were measured using the six-component balance. The
Fig. 2 Close-up cross-sectional schematic of TE section.
USE OF CC FLIGHT CONTROL
341
Fig. 3 Model mounted in wind tunnel.
freestream velocity was set at 25 m/s, corresponding to a freestream Reynolds number of approximately 8.5 x lo5, and maximum jet velocities were approximately 180 m/s. The air supply was sourced from pressurized receiver tanks fed by an AtlasCopco compressor, delivered to the plenum by a flexible hose, such that tare effects out of the plane of measurement were avoided. The pressure within the plenum was monitored with a pressure tapping and using a pressure transducer. The mass flow rate was determined using an orifice plate rig with pressure transducers and an orifice plate with orifice-to-pipe-diameter ratio p of 0.2401. The pressure and flow temperature data were transferred to the computer via an A-to-D card. A computer program was written to accumulate data and calculate the flow rate. From this the blowing momentum coefficient C , could be calculated using Vjriz c, = -
qs
where vj is the velocity of the Coanda jet, riz is the jet mass flow rate, q is the freestream dynamic pressure, and S is the model surface area. The jet velocity was calculated using the isentropic pressure distribution
to avoid errors that can occur using the jet area as a variable. As interest was directed at low blowing rates, data were recorded at increments of C, of
342
S. P. FRlTH AND N. J. WOOD
0.0005 up to 0.01 and then using increments of 0.005 up to 0.03 to obtain general force or moment curves. Particle image velocimetry (PIV) was also performed to obtain more information on the interaction of the jet with the freestream flow.5 A horizontal lightsheet was fired at the TE of the CCW using an Nd:YAG laser. A megapixel CCD camera, positioned under the wind-tunnel floor, captured a sequence of pairs of images of the seeded freestream flow over the wing. The images were then time-averaged and analysed using TSI Insight and Tecplot 9 software to obtain velocity and vorticity data. As part of a joint project, BAE Systems6 calculated computer fluid dynamics (CFD) data to compare with the experimental data. There was a reasonable agreement between the computed and experimental data, although it was felt that additional refinement of the grid would enhance results.
B. Results The aerodynamic data are represented as a series of carpet plots. The lift carpet plots show the variation of the lift coefficient CLwith blowing at angles of attack ranging from 0 to 15 deg in 5-deg increments. The blowing values at a particular angle of attack were offset, with the lowest angle of attack being offset by the greatest amount, effectively removing the need for a horizontal axis. Lines of constant blowing coefficient C, link the lines of data for each angle of attack. The vertical axis represents the lift coefficient C,, with dashed horizontal gridlines of constant CL used for reference. An example of this is demonstrated in Fig. 4. The lift augmentation is calculated from the gradient of the lift curve for each angle of attack over a particular blowing coefficient range. The pitching moment carpet plots are to a large extent very similar in layout to the lift carpet plots. The main difference is that the offset for the blowing coefficients is reversed, such that the highest angle of attack is offset by the greatest amount. The vertical axis represents the pitching moment C, about a particular
.............................................................
Fig. 4 Example of a lift data carpet plot.
USE OF CC FLIGHT CONTROL
343
I
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
........
5
Increasing
I
Blowing
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Fig. 5 Example of a pitching moment data carpet plot.
reference point (in the case of the half-span model, this is the LE), as shown in Fig. 5 . The results given in Fig. 6 show the effect of CC on the lift characteristics with a variation in slot height. There is an increase in lift with an increase in C,, although the greatest lift increments were found at lower blowing rates. The is of the order 18-25 over the complete level of lift augmentation aC,/aC, range of C, tested, although a maximum incremental augmentation of approximately 53 was recorded. Also, it was found that the smaller slot height yields a stronger lift augmentation at smaller values of C,. This may be due to the 1
0.8
" 2
0.6
4 c
.-0,
If"
8 G
J
0.4
0.2
0 -0.2 Fig. 6 CL vs C,: effect of slot height on CC.
S . P. FRlTH AND N. J. WOOD
344
smaller slot height giving rise to an increase in the ratio between the momentum of the jet and that of the freestream flow. It is anticipated, however, that a minimum slot height will be reached where the jet no longer attaches to the Coanda surface. This requires further research. The pitching moment data are characterized by a negative gradient, depicting a nosedown pitching moment, as shown in Fig. 7. This is typical of a CC system, as also seen by Jones and Engla~-.~ The rate of pitching moment is greatest at the lower blowing coefficients, although it levels off with increasing jet momentum. There are similar trends for each angle of attack, although data for the model at 15deg revealed that the nosedown pitching moment of the model was less than that obtained for the model at l0deg. This can be attributed to the more powerful nature of the LE vortex at higher angles of attack, the increased suction on the upper surface giving rise to a slight nose-up pitching moment. The drag coefficient was also found to increase as the blowing rate is increased, although the drag augmentation is significantly less than the equivalent value for lift, suggesting an overall increase in LID. However, drag measurements are not presented in this chapter because of an inconsistency in the data, which may be due to fluctuations in the Coanda jet or the accuracy range of the balance. Fig. 8 shows the calculated time-averaged velocity vectors obtained from PIV in the form of a contour plot using the TSI Insight and Tecplot softwares for different values of C,. It can be seen that the external flow visibly changes at higher blowing rates, indicated by a downward deflection of the velocity vectors. The data also demonstrate that the downstream extent of the wake was reduced. There is also an area of accelerated flow over the upper surface, just
G J
o‘2 O
T----
........................................................................
E
0” -0.2
a 0,
-0.4 UJ
C
5 -0.6
-I
1 3
-0.8
0
a
-1 a E -1.2
c
C
8
p
-1.4
f
-1.6
.r
-1.8
I
..............................................................................
...............................................................
-5degrees A 10 degrees
________________________________________-------------------------------
Fig. 7 Variation of pitching moment with blowing for half-span model with 0.15mm slot height (lower blowing coefficients range: 0 5 C, 5 0.01).
USE OF CC FLIGHT CONTROL
345 b)
Fig. 8 PIV velocity contour plots with streamlines obtained for angle of attack 10 deg at different blowing coefficients: a) C , = 0, b) C , = 0.005, c) C , = 0.01.
before the jet exit. Because of restrictions with the apparatus it was not possible to seed the jet and investigate the full interaction with the freestream flow. 111. Full-Span UAV Configuration
A. Experimental Procedure A full-span model was designed and constructed at the Goldstein Laboratory, Manchester, to investigate any interaction of the Coanda jets and examine the possibility of roll control, as well as lift enhancement. A schematic is shown in Fig. 9. The main body was constructed using modelboard. The model had a LE sweep angle of 55 deg and a TE sweep of -30 deg, resulting in the diamond-shaped planform as shown in Fig. 9. The plenum sections, made from aluminum for the upper surface and brass for the lower surface, incorporated similar TE dimensions as the previous model: TE diameter of 6 mm and slot height adjustment from 0.05 to 0.30 mm (this was set at 0.15 mm to compare with previous results). The spanwise extent of each slot was reduced to 300 mm and did not extend the full length of the TE. The blowing rate was again controlled using an orifice plate rig for each plenum ( p = 0.3), such that the plenum sections
346
S. P. FRlTH AND N. J. WOOD
Balance Mounting Plate
Fig. 9 Plan view schematic of full-span model.
could be controlled independently. The air supply was controlled by the use of two valves for each plenum, allowing finer and more accurate control. The fuselage section of the model was manufactured from aluminum sheet, to create a theoretical aircraft profile and provide protection for the instrumentation (dualaxis inclinometer and strain-gauge balance) and air supply within the model. The model was mounted on a sting in the Avro 9 x 7 wind tunnel as shown in Fig. 10, incorporating an internal six-component strain-gauge sting balance to measure aerodynamic forces and moments. The air supply was again taken from pressurized tanks and passed through a series of flexible hoses. Tare effects because of flexing of the hoses when under pressure were minimized by incorporating highly flexible hose within the model, adjacent to the calibration center of the balance. Any tare effects resulting from flexing of hoses were measured wind-off. Preliminary tests were performed prior to load data acquisition to determine efficiency of both Coanda surfaces, check for any leakages and uniformity of both slots. Test runs were made in the wind tunnel to examine model integrity and performance. Tests were accomplished at 25 m/s (a freestream Reynolds number of approximately 1.3 x lo6) and the angle of attack was varied from 0 to 15deg in 5-deg increments. The blowing coefficient was varied from zero to 0.004 in increments of 0.0005. Data were taken for various test parameters: symmetric blowing, in which the jet momentum from both plenums was identical, and differential blowing, in which one plenum would maintain a constant C, and the other side would operate over the complete range. Table 1 summarizes the test procedure undertaken.
USE OF CC FLIGHT CONTROL
347
Fig. 10 Sting-mounted model in wind tunnel.
B. Results In quiescent conditions, both Coanda jets performed as expected, with the jets fully attaching to the Coanda surfaces, verified using tufts. It was possible to maintain a tunnel velocity up to 40 m/s without the model experiencing significant fluctuations, although all tests were performed at 25 m/s for consistency. The load data for the full-span model are again represented in a series of carpet plots, as described in Sec. 1I.B. In addition to lift and pitching moment plots, rolling moment data are represented in a similar form to the pitching moment data, with the vertical axis giving values of Cl. Figures 11 to 18 show the effectiveness of the full-span model, in the form of carpet plots with contours of constant C, and angles of attack. A lift augmentation aCL/aC, of 17-24 was achieved, as demonstrated in Fig. 11, in which data are shown for both Coanda jets at the same mass flow rate, and therefore the same C, (symmetric blowing). Although the lift augmentation achieved is not as great as that achieved in other studies,’ it is believed that this can be attributed to the small radius of the Coanda surface. The tradeoff of a lower lift augmentation is that the drag for such a surface is reduced when compared to traditionally large CC Coanda surfaces. Table 1 Test procedure for full-span model Configuration Symmetric Differential 0 Differential 0.002 Differential 0.004
Left plenum section blowing coefficient, CI*(L)
0 -0.004 0 -0.004 0 -0.004 0 -0.004
Right plenum section blowing coefficient, CI*(R)
0-0.004 0 (constant) 0.002 (constant) 0.004 (constant)
S. P. FRlTH AND N. J. WOOD
348 1
T
............................................................................
I
-
15 degrees
I
Fig. 11 Variation of lift with blowing for full-span model with both systems active 5 0.008). (blowing coefficients range: 0 5 Cp(total)
Assuming the center of gravity to be at the quarter-chord position, the pitching moment about this point is nosedown (Fig. 12), which is as expected because the center of lift is located aft of the quarter chord. It is encouraging to see that the CC device could be used to trim the aircraft, while maintaining high values of lift augmentation, as the variation in C, required at various angles of attack is approximately linear, as shown in Fig. 13. This suggests that the control of this parameter could be simply transferred to stick control in a real-flight situation. The investigation in using CC for roll control revealed some interesting characteristics. The variation of lift with asymmetric blowing (zero blowing from the right Coanda jet) is shown in Fig. 14. Again, a lift augmentation of approximately 20-25 is achieved and it was demonstrated that the jet momentum is additive; that is, if the left jet was used at the maximum value of C ,, the activation of the right jet would result in a similar lift curve to that obtained with symmetric blowing. The control of rolling moment by CC is demonstrated in Figs. 15 to 18. Figures 15 and 16 demonstrate the effect on the rolling moment of the use of just one system. It can be seen that a particular rolling moment can be achieved with a particular value of C, independent of the angle of attack, although the leading edge vortex, particularly effective at angles of attack from approximately 7.5 deg produces an additional pro-roll moment. This pro-roll moment results from a secondary effect of the blowing that enhances the vortex suction signature ahead of the blowing slot.4 This can be seen in the kink in the rolling moment curves. The graphs shown in Figs. 17 and 18 indicate how the differential blowing affected the rolling moment of the model. The first of these shows the use of the right system at half the maximum blowing possible (C, = 0.002) held
USE OF CC FLIGHT CONTROL
349
0.04
5 -0.02 0 m *
c)
C
r"m
-0.04
C
5
-0.06
h -0.08
Fig. 12 Variation of pitching moment with blowing for full-span model with both 5 0.008). systems active (blowing coefficients range: 0 5 Cp(total)
constant, while increasing the jet momentum on the left system through the entire range possible (0 5 C, 5 0.004). The effect of the vortex is clearly evident, with the most influence with the right system at C, = 0.002 and the left system at either C, = 0 or C, = 0.004. Figure 18 shows how the system can be returned to a state of zero roll from a rolling motion.
0
5
10
15
S. P. FRlTH AND N. J. WOOD
350 1
0.8
0
-0.2
Fig. 14 Variation of lift with blowing for full-spanmodel with left system active only (blowing coefficients range: 0 5 Ca(r)5 0.004).
Adapting ESDU Data Items Aircraft 06.01.01* and 88013,9 it was possible to determine the values for the rolling moment derivatives resulting from aileron input and damping. These are given in Table 2. From these values, a roll rate of approximately 403 deg/s is achieved with a single downward aileron deflection of 10 deg. ............................................................................
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
/
I +15 degrees I ............................................................................
............................................................................
............................................................................
Fig. 15 Variation of rolling moment with blowing for full-span model with left system active only (blowing coefficients range: 0 5 C a ( ~5) 0.004).
USE OF CC FLIGHT CONTROL
q *
351
0.04
.............................................................................
0.03
.............................................................................
0.02
.-m
!-s
0'01
5 : .-
o
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
...................
-5 -A-
mrees 10 degrees
..
-0.01
C
a"
-0.02
-0.03 -0.04
3 C,=0.004 .............................................................................
Fig. 16 Variation of rolling moment with blowing for full-span model with right system active only (blowing coefficients range: 0 IC f i ( ~I)0.004).
If the ailerons were substituted with the CC system, the parameter aC,/aC, would no longer be valid and the parameter aC,/aC, would replace it. This is essentially the gradient of the rolling moment curves with blowing; a mean value of approximately 7 is obtained. The parameter 8 would also be replaced by C,. Assuming the response of the rolling moment with blowing is approximately linear, along with the response due to aileron deflection, it is 0.04
0.03 ............................................................................. 0- 0.02 *-
.-m
! -
5
... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
0'01
0
5
2 -0.01 0
.--
a"
........................................:.
-0.02
-
-
dl
'
Jc!,(ToTAL)
= 0.002
...........................................................................
-0.03 -0.04
Fig. 17 Variation of rolling moment with blowing for right CC system blowing at constant C p ( ~=) 0.002 and increasing blowing on left CC system (0 5 Ca(L)5 0.004).
S. P. FRlTH AND N. J. WOOD
352
Table 2 Calculated values for rolling moment derivatives of full-span model
acIm
acIm -0.084 rad-
'
-0.052 rad-'
possible to equate the equivalent blowing coefficient required to generate the same rate of roll such that
This gives a C, of approximately 0.0021, equivalent to an aileron deflection of approximately 10 deg. The slight negative rolling moment present at an angle of attack of 0 deg and C, = 0 indicates that there is a slight model asymmetry, although this does not have a significant impact on the effectiveness of the system.
IV. Conclusions An experimental investigation of CC has been successfully modeled, initially on a single delta-wing configuration with varying TE geometry and then on a fullspan model to investigate the potential for roll control. The variation of slot height indicated that a smaller slot height yielded a higher lift augmentation aCL/aC,. However, it is anticipated that there is a limiting 0.04 .............................................................................
I 1
0.03 ............................................................................. 0 *-
0.02 .............................................................................
g.-:-
0.01
.............................................................................
+O
+-lodegrees
-8---
= -0.01 0
.........
-0.02 ..................
-0.03 -0.04
degrees
... ... ... 4.......1........ .......
........
.
TOTAL) = 0,004
.............................................................................
Fig. 18 Variation of rolling moment with blowing for right CC system blowing at constant C a ( ~=) 0.004 and increasing blowing on left CC system (0 5 C p ( ~5) 0.004).
USE OF CC FLIGHT CONTROL
353
height, requiring further work. Lift augmentations of approximately 18-25 for low blowing rates were obtained with both models over the complete lower blowing range. This suggests that useful lift increments can be obtained with C, values of the order 0.005, equivalent to those achieved using existing flap systems (ACL x 0.1). As the CC system is considerably less complex mechanically than other high-lift devices, this may be significantly beneficial when contemplating maintenance, production costs, and reliability. The full-span tests demonstrated that the CC system was effective at generating significant rolling moments at low blowing coefficients. Importantly, the production of roll moments can be superimposed on the lift generation, suggesting minimized interaction and simple control development. More detailed work at even smaller increments of C ,, especially in the lower blowing regions, will enable greater understanding of the physics involved in CC and the areas of higher lift augmentation. Power requirements for blowing need to be studied to determine the overall efficiency of the system compared to conventional systems. Further experimental work using the full-span model will continue to investigate the application of CC to roll control and pitch trim. The implementation of pulsed jets will also reduce the required mass flow bleed, yet provide similar lift augmentation^.^
Acknowledgments The authors wish to acknowledge the contributions of staff and students at the Goldstein Laboratory at the University of Manchester, especially those of the technicians for their help with model manufacture. A special mention must also go to Andrew Kennaugh for his continuous help throughout the project. References ‘Wood, N. J., and Nielsen, J. N., “Circulation Control Airfoils-Past, Present, Future,” AIAA, 23rd Aerospace Sciences Meeting, Jan. 1985. 2 Englar, R. J., and Applegate, C. A., “Circulation Control-A Bibliography of DTNSRDC Research and Selected Outside References (Jan. 1969 through Dec. 1983),” DTNSRDC Rept. 84/052, Sept 1984. 3Jones, G. S., and Englar, R. J., “Advances in Pneumatic-Controlled High-Lift Systems Through Pulsed Blowing,” 21st AIAA Applied Aerodynamics Conference, June 2003. 4Frith, S. P., and Wood, N. J., “Effect of Trailing Edge Geometry on a Circulation Control Delta Wing,” 21st AIAA Applied Aerodynamics Conference, June 2003. ’Raffel, M., Willert, C., and Kompenhans, J., “Particle Image Velocimetry-A Practical Guide,” Springer, Berlin, 1998. %ellars, N. D., Wood, N. J., and Kennaugh, A., “Delta Wing Circulation Control Using The Coanda Effect,” AIAA 1st Flow Control Conference, June 2002. 7 Englar, R. J., “Circulation Control Pneumatic Aerodynamics: Blown Force and Moment Augmentation and Modification-Past, Present and Future,” AIAA Fluids 2000 Conference and Exhibit, June 2000. 8“ Stability Derivative, L,, Rolling Moment due to Rolling for Swept and Tapered Wings,” Engineering Sciences Data Unit, Item A 06.01.01, March 1955. ’“Rolling Moment Derivative, Lg, for Plain Ailerons at Subsonic Speeds,” Engineering Sciences Data Unit, Item 88013, August 1988.
1I.C. Experiments and Applications: Nonaerospace
Chapter 13
Pneumatic Aerodynamic Technology to Improve Performance and Control of Automotive Vehicles Robert J. Englar* Georgia Institute of Technology, Atlanta, Georgia
Nomenclature A = vehicle frontal area, ft2 b = vehicle width, ft C, = drag coefficient = drag/(qA) CL = lift coefficient = lift/(qA) CM= pitching moment coefficient = pitching moment/(qAc) C , = yawing moment coefficient = yawing moment/(qAb) Cy = side force coefficient = side force/(qA) C, = jet momentum coefficient = mVj/(qA) c = vehicle total length = tractor + trailer + gap, ft h = blowing jet slot height, in. m = measured jet mass flow, slugs/sec q = freestream dynamic pressure, psf Re = freestream Reynolds number, based on vehicle length c V = freestream velocity, ft/sec vj = isentropic jet velocity, ft/sec p = freestream or jet density, slugs/ft3 = yaw (side wind) angle, deg I. Introduction ONSIDERABLE interest has arisen recently in improving the aerodynamics of heavy vehicles (HVs) as a means of improving their operating costs,
C
*Principal Research Engineer, Aerospace, Transportation & Advanced Systems Lab., Georgia Tech Research Institute. Associate Fellow AIAA. Copyright 0 2005 by Robert J. Englar. Published by the American Institute of Aeronautics and Astronautics, Inc., with permission.
357
358
R. J. ENGLAR
performance, and safety. As discussed in Ref. 1, for a typical U.S. tractor-trailer rig logging 175,000 miles a year at a fuel price of $lSO/gallon, yearly fuel costs could average over $40,000 ($29,000 if only 125,000 miles are logged). Thus even a 5-10% increase in fuel economy could be meaningful. Although devices that can reduce the HV’s drag coefficient can significantly improve fuel economy, it is also desirable that additional capabilities result from improved aerodynamics. These could include increased stability (both lateral and directional), reduction in side-wind sensitivity, reduction in splash and spray, and improved traction plus aerodynamic braking. One could also include an aerodynamic means to reduce tire rolling resistance. Any such devices being considered for these applications should also be simple and robust, contain few or no moving parts, should not be hampered by weather, and not increase vehicle weight or external dimensions. This paper discusses pneumatic aerodynamic devices based on the use of circulation control (CC) aerodynamics, which thus possess many of these desirable characteristics. These are currently under development at Georgia Tech Research Institute (GTRI) for the DOE Office of Heavy Vehicle Technology. First described in the following sections will be the basics of pneumatic aerodynamics and application to heavy vehicles, and then details of wind-tunnel and full-scale programs conducted, their results, and possible future applications.
11. Basics of Pneumatic Circulation Control Aerodynamics GTRI researchers have been involved for a number of years in the development of pneumatic (pressurized air blowing) concepts to yield efficient yet mechanically simple means to control, augment, or reduce the aerodynamic forces and moments acting on aircraft. This was detailed in Refs. 2 to 4, among others, but will be summarized briefly to familiarize the reader with this technology. Figure 1 shows the basic pneumatic concept, which has become known as circulation control (CC) aerodynamics. Here, an airfoil’s conventional mechanical trailing-edge (TE) device has been replaced with a fixed curved surface and a tangential slot ejecting a jet sheet over that surface. That jet remains attached to the curved surface by a balance between subambient static pressure on the surface and centrifugal force (the so-called Coanda e f f e ~ t )This . ~ entrains the external flowfield to follow the curving jet, and thus enhances the circulation around the airfoil and the aerodynamic forces produced by it. The governing parameter is not angle of attack, but rather the blowing momentum coefficient: mV,
c, =qs where m is the jet mass flow, vj the isentropic jet velocity, S is a reference wing area (or frontal area A for a ground vehicle), and q is the freestream dynamic pressure ( q = 0.5pV2, with p being the freestream density). At lower C, values, augmentation of the aerodynamic lift by a factor of ACl/C, = 80 has been recorded? representing an 8000% return on the invested jet momentum (which in a physical sense is also equal to the jet thrust). Familiarity with
IMPROVING PERFORMANCE OF AUTOMOTIVE VEHICLES
359
POSSIBLE LEADING EDGE BLOWING
TANGENTIAL BLOWING OVER ROUNDED TRAILINQ EDGE SURFACE
t
FORCE BALANCE
BOUNDARY LAYER CONTROL
Y MOMENTUM COEFFICIENT, Cp
JET SHEEF
c p = IilvJIqs
Fig. 1 Basics of circulation control pneumatic aerodynamics on a simple twodimensional airfoil.
blown aerodynamic systems will remind the reader that this is quite extraordinary; thrust-deflecting vertical takeoff and landing (VTOL) aircraft are fortunate if they recover anything near 100% of the engine thrust expended for vertical lift (which must exceed weight), with very little, if any, augmentation of aerodynamic lift occurring. It is because of this high return on blowing, or conversely, because of the very low required blowing input and associated power required to achieve a desired lift, that CC airfoils appear very promising for a number of applications. The A-6/CC Wing short takeoff and landing (STOL) flight demonstrator aircraft (Fig. 2)* showed the STOL performance listed, and also suggested capabilities very useful to ground vehicles: during short takeoff, it demonstrated high lift with reduced drag, and in the approach/landing mode, very high lift with high drag was shown. These advantages led to the application of this pneumatic concept to improve the aerodynamics of an already streamlined car model.5 The resulting large jet turning over the curved rear of this vehicle is shown in Fig. 3. Significant but distinctly different trends were observed during testing, depending upon which portion of the tangential slot located along the trunk break line was blown. Blowing the full-width slot produced the large jet turning shown by the striped tuft in Fig. 3, with drag increases of greater than 70%, showing potential for pneumatic aerodynamic braking. Blowing only the outside segments of the slot weakened the comer vortex rollup, attached separated flow, lessened aft suction, and reduced drag by as much as 35%. Blowing this aft slot also yielded a lift increase of 170%. One can envision a similar slot applied to the lower rear surface that could instead yield negative lift (positive down force). This concept has been patented by GTRI and verified by a similar installation on a wind-tunnel model of a European Formula 1 race car.6
R. J. ENGLAR
360
Flight Test Results: 140% Increase in Usable Lift Coefficient, CL
30-35% Reduction in Takeoff & Approach Speeds 60-65% Reduction in Takeoff & Approach Ground Rolls
Fig. 2 A-6/CCW STOL flight test confirming pneumatic devices for aerodynamic force augmentation.
111. DOE Pneumatic Heavy Vehicle Model Test Results Based on the preceding results, a research program was initiated at GTRI for the Department of Energy’s Office of Heavy Vehicle Technologies. The goal was to apply this pneumatic technology to tractor-trailer configurations to develop an experimental proof-of-concept evaluation leading to an on-the-road
Fig. 3 Pneumatic technology on a streamlined car model: aft view, showing blown jet turning.
IMPROVING PERFORMANCE OF AUTOMOTIVE VEHICLES
361
Fig. 4 Schematic of application of GTRI pneumatic technology to heavy vehicle trailer, showing four aft blowing slots and upper LE blowing slot.
demonstration of an operating pneumatic heavy vehicle (PHV). Figure 4 shows a schematic of a generic PHV with tangential blowing slots on each of the trailer’s four curved aft edges, plus blowing on the rounded upper leading edge (LE) of the trailer. Early portions of that effort, including a preliminary feasibility study and design of baseline and pneumatic wind-tunnel configurations, are detailed in Ref. 6 .
A. Wind-Tunnel Evaluations of Baseline Unblown HV Models To develop a representative PHV configuration prior to full-scale testing, initial baseline wind-tunnel testing was conducted, which was then followed by several phases of blown test configurations. For this, an existing generic HV configuration, the ground transportation systems (GTS) vehicle of Ref. 7, was used. The model is shown in Fig. 5 before the blowing modifications were 0.95
0.9 0.85
0.8 0.75
0.7
CD 0.65
0.6 0.55
0.5 0.45 0.4
0.35 0
5
10
15
20
25
30
35
40
45
9.PSf
Fig. 5 Test results for unblown HV models, showing effects of cab height, gap, wheels, and Reynolds number.
R. J. ENGLAR
362
installed. It is actually representative of a faired cab-over-engine HV based on a Penske racing team car carrier, and is relatively independent of the numerous and varying cab roof fairings employed on a number of current HV. Tests of this unblown model configuration did, however, demonstrate the importance of cab/trailer gap and fairing treatments. These configurations were tested in the GTRI Model Test Facility research tunnel6" and showed some significant drag reductions because of changes in the unblown geometry. Figure 5 shows drag reductions of up to 25% below a low-cab full-open-gap vehicle when the gap was eliminated (filled in) and the cab top was even with the trailer top (trailer leading and trailing edges are square here). An additional 15% reduction was confirmed with a round trailer LE facing into the open gap and a round TE on the trailer (this is the unblown PHV). These data were taken at a typical tunnel speed of 70mph. Also very significant is the tremendous increase in CD in Fig. 6 (more than a doubling is seen) due to a side wind acting at a yaw angle on the HV. (In all of the drag data shown herein, CD is based on projected frontal area of the vehicle A, including the wheels.)
B. Wind-Tunnel Evaluations of Blown HV Configurations Based on the preceding unblown configurations with reduced drag, additional wind tunnel tests were conducted to evaluate aerodynamic improvements 1.8
1.7
1.6 1.5 1.4
1.3
1.2
CD 1.1 1
0.9
0.8
0.7 0.6
-15
-10
-5
0
Yaw Angle, y ~ ,deg
5
lo Nose right 15
Fig. 6 Effects of side wind on drag for various unblown HV configurations.
IMPROVING PERFORMANCE OF AUTOMOTIVE VEHICLES
363
resulting from blown configurations. Details of these investigations are presented in Ref. 8. Unless otherwise noted, the blowing variations were run at tunnel (vehicle) wind speeds of approximately 70-71 mph (dynamic pressure q = 11.86 psf and Reynolds number = 2.5 x lo6, based on total tractor trailer length).
+
C. Drag Reductions (for Fuel Economy) or Drag Increases (for Braking & Stability) The blowing slot heights at each aft edge of the trailer could be varied and tested either unblown or blown in any combination of the four, or even with LE slots on the trailer front face also blown. Flow visualization tufts in Fig. 7 show jet turning of 90deg on all four aft corners, even the bottom slot blowing upwards. Figure 7 also shows the results of this jet turning on reducing or increasing aerodynamic drag by blowing various combinations of these aft slots. The combination of all four slots blowing together yielded the greatest drag reduction, more effective than blowing individual slots. Compared to the typical unblown baseline configuration from above (full gap between cab and trailer, square trailer LE and TE, and cab fairing slightly lower than the trailer front), which produced CD = 0.824 at this Reynolds number, the blown configuration reduced the drag coefficient to 0.459 at C, = 0.065. This is a 44% CD reduction, and the internal plenum blowing pressure required was only 0.5 psig. A second blown configuration (labeled 90°/300 TE) used less jet turning on the upper and lower surfaces to generate even greater drag reduction-at 0.5 psig, CD was reduced by 47%, and at 1.0 psig (C, = 0.13), CD was reduced by 50%. These data are all for a smoothed bottom tractortrailer model with low sides and half-cylinder simulated wheels.
Momealum Coellleieat, C,
Fig. 7 Drag reduction or augmentation on blown trailer with 90-deg turning surfaces, plus flow visualization of jet turning.
R. J. ENGLAR
364
Additional evaluation of the effectiveness of the blown configurations was made. The drag coefficient of the preceding unblown baseline configuration, but with the tractor-trailer gap filled in, is CD = 0.627 (not shown in Fig. 7). Addition in Fig. 7 of the unblown pneumatic surfaces onto the trailer TE reduces CD by 9.7%. Adding blowing at C, = 0.065 reduces that CD by another 23.1%. This combination reduces CD to 30.6% less than the square TE baseline having a smooth fairing filling in the gap. When only the top slot, the bottom slot, or both of these slots were blown in the absence of the side jets, drag was initially reduced slightly, but then significantly increased with the addition of blowing. This represents an excellent aerodynamic braking capability to supplement the hydraulic brakes. Blowing efficiency is plotted in Fig. 8, where ACD is an increment from the blowing-off value (negative ACD is reduced drag). Absolute values of ACD/C, greater than 1.0 represent greater than 100% return on the input blowing C,. It is seen that the 90 deg/ 30 deg four-slotted configuration generates values as high as - 5.50, representing 550% of the input blowing momentum recovered as drag reduction. This figure also shows the opposite trend, with up to 200% of the blowing momentum from top/bottom slots recovered as increased drag for braking. Obviously, these percentages will be modified when the power expended to compress the blowing air is included, but that will have to await a full-scale systems study. Should additional air be available from an onboard source such as an existing turbocharger or an electric blower, additional drag reduction is possible, as shown in Fig. 9. Drag coefficients of less than 0.30 are shown for faired blown HV 3.5 3.0 2.5 2.0 1.5 1.0
0.5
0.0 -0.5 -1.0 -1.5
-2.0 -2.5
-3.0
-3.5 -4.0 -4.5 -5.0 -5.5 0.15
0.12
0.09
0.06
0.03
-0.00
-0.03
-0.06
-0.09
-0.12
-0.15
ACD = C D.CDO
Fig. 8 Blowing efficiency and drag increments caused by blowing slot configuration.
IMPROVING PERFORMANCE OF AUTOMOTIVE VEHICLES
0
0.1
0.2
0.3 0.4 0.5 0.6 Momentum Coefficient, Cp
0.7
0.8
0.9
365
1
Fig. 9 Reynolds number effects and increased blowing values, plus LE blowing and gap plates.
configurations. This is in the arena of streamlined sports cars. The drag coefficient of a 1999 Corvette coupe is C, = 0.29. Figure 9, originally intended to show that the drag curves tend to converge onto one slope independent of Reynolds number, also shows a measured drag coefficient of 0.13 for the PHV model at increased C,. This is about half the drag coefficient value of the Corvette or a Honda Insight hybrid (C, = 0.25). Even though not achieved in the most efficient blowing operation range, this is an 84% drag reduction compared to the unblown baseline configuration. Note that the tractor cab in Fig. 9 has “gap plates” (or fairing extensions) instead of the full “no gap” fairing of Fig. 8, and is thus much closer to an actual tractor/trailer configuration. It also has blowing on the trailer top LE. Figure 10 shows this alternative means of improving upon the gap problem. Note that when comparing these data to other experiments that have been conducted by other researchers on similar GTS models, these GTRI data above and below include simulated wheels, which, as Fig. 5 shows, add about AC, = 0.18 to nonwheeled vehicles’ C, values, perhaps more, depending on how well the tunnel ground effects are treated experimentally. GTRI’s measured data are recorded using test-section tangential floor blowing to eliminate floor boundary-layer i n t e r f e r e n ~ e . ~ ’ ~ , ~
D. Stability and Control Strong directional instability can be experienced by HVs at yaw angles (i.e., when experiencing a side wind) because of large side forces on their flat-sided trailers (Fig. 11). This yaw sensitivity is confirmed by the unblown (C, = 0) yawing moment CN shown, where yaw angle as small as - 8 deg produces a
R. J. ENGLAR
366 0 65
0 60
0 s5
0 so
CD 0 45
0 40
0 35
0 30 OW
005
015
010
025
020
030
Momeolum Cafl5eieot.C p
Fig. 10 Side plates and trailer top blowing: a practical solution to the cab gap problem.
large unblown yawing moment coefficient of C, = -2.0. (It should be noted here that this yawing moment is measured about the rigid model's midpoint at the ground, whereas on a real articulated tractor-trailer, it would be experienced at the tires of the individual units. However, comparisons of blowing on and off
Nose 90°/300 1 / 2 TE ,0.375"R GO", Wheels ON, Left Slot Blowing Only
Right 4.0
p-00
N = Half Chord Yawing Moment Coefficient y=O , M s e Straight Ahead
CN
1.5 1
.o
ose Yawed Left
0.5
0.0 -0.5 -1.0 -1.5
Nose Lefl '
0.00
'
8
'
0.02
'
"
'
'
0.04
'
8
.
.
0.06
" " " " " " " " " " " "
0.08
0.10
0.12
0.14
0.16
0.18
0.20
Momentum Coemcient, C
Fig. 11 Directional control capability provided by HV configuration with left rear slot blowing only.
IMPROVING PERFORMANCE OF AUTOMOTIVE VEHICLES
367
are being made for the same single HV unit, and apparent benefits should be valid.) Blowing only one side slot can easily correct this. With the nose straight ahead (Ic, = 0 deg), blowing the left slot at C, = 0.06 yields the equivalent opposite yawing moment ( C N = +2.0). With the nose yawed left (for example, Ic, = - 8 deg), blowing at C, = 0.06 returns the unstable yawing moment to CN = 0.0. Then, increasing the blowing a bit more causes the nose to yaw in the opposite direction, to the right. The opportunity for a no-moving-part, quick-response aerodynamic control system is apparent.
IV. Pneumatic HV Fuel Economy Testing The preceding model tests led to the conclusion that a full-scale proof-ofconcept test series should be conducted on a PHV test rig to determine if the tunnel results would also occur on the road. Based on the above wind-tunnel results, GTRI team member prototype shop Novatek, Inc. designed and fabricated the PHV blown test trailer, including blowers, drive motors, control systems, and instrumentation. This configuration is shown in Fig. 12 in comparison to a stock (reference) Great Dane trailer. Blown tufts confirming jet turning of 90 deg around the right-side TE curved pneumatic surface are shown in Fig. 13. A. TuningTests Test vehicle fabrication and assembly were completed at GTRI and the modified trailer was then picked up by team member Volvo Trucks of North America and moved to its facility in Greensboro, NC. Here two initial tuning tests were conducted (Fig. 14)." Figure 15 shows a rear view of the pneumatic trailer with the tufts confirming on-road flow turning. These tests verified the test equipment and data system operations, and indicated unofficial fuel economy increases from blowing of up to 15.3% over the baseline trailer when on an open highway, rather than on a test track.
Fig. 12 PHV pneumatic trailer (left) and baseline reference unmodified trailer (right).
368
R. J. ENGLAR
Fig. 13 Right rear corner view, looking up, with tufts showing 90-deg jet turning.
B. Full-scale PHV On-Track Fuel Economy Tests On-track testing of the PHV test vehicle (tractor and modified trailer) was conducted at the Transportation Research Center (TRC) test track in East Liberty, Ohio, along with a control vehicle (a stock Volvo/Great Dane rig). Figure 16 shows these two vehicles while in a pit lane fuel station at the 7.5-mile banked
Fig. 14 On-road PHV tuning test near Volvo facility in North Carolina.
IMPROVING PERFORMANCE OF AUTOMOTIVE VEHICLES
369
Fig. 15 On-road PHV test vehicle rear view with jets blowing and tufts turning.
test track at TRC. SAE Type-I1 fuel economy runs were conducted by the TRC/ GTRI/Volvo/Novatek team members in strict accordance with S A E test procedures (as specified in SAE J1321, Oct. 1986). During these tests, a total of 59 runs was made for the six configurations evaluated, each at three different
Fig. 16 Test and control vehicles in pits at TRC test track.
R. J. ENGLAR
370
Fig. 17 PHV test vehicle on track at 75 mph.
speeds (55,65, and 75 mph) and with each run covering six laps (45 miles) of the TRC test track.g911 -13 Figure 17 shows the pneumatic test tractor-trailer at speed on the TRC track. The six sets of fuel economy runs were made at different blowing rates, including zero blowing. This allowed reference comparisons to be made after the pneumatic test trailer was reconfigured into the baseline trailer and then tested to provide reference fuel economy of the standard vehicle (all fuel economy data achieved with the other test configurations were compared to this one to determine percent fuel efficiency increase, %FEI). Figure 18 shows
0
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
0.09
0.1
Blowing Momentum Coefficient, CF
Fig. 18 Measured PHV fuel economy improvement, with four trailer slots blowing.
IMPROVING PERFORMANCE OF AUTOMOTIVE VEHICLES
371
%FEI as a function of blowing coefficient, C ., The %FEI improvements shown range from 4 to 5% (5 to 6% if the 1% error band is included) above the fuel economy of the baseline standard tractor-trailer, but these are seen to reduce somewhat as blowing increases to values beyond C, = 0.02-0.03. Whereas responses heard from typical HV users indicate this 5-6%FEI to be quite respectable, it is less than we had anticipated based on our smaller-scale wind-tunnel tests.@ Figure 19 compares this on-track data to the predicted fuel efficiency increase that we had expected from the drag reductions of the blown configurations. Whereas the lower blowing values were nearing 70-80% of the expected values, at greater blowing the payoff was reduced. The test team of GTRI, Novatek, and American Trucking Associations identified suspected reasons for this, and we then conducted an experimental test program to confirm these, as discussed in Sec. V.
V. Updated Wind Tunnel Evaluations A new series of wind-tunnel runs was made on the 0.065-scale PHV model. We began with the best blown configuration from Fig. 7 (now seen as Run 205 in Fig. 20), and then we sequentially downgraded the model by making changes suspected of being detrimental when installed on the road-test truck. It was the intent of this new wind-tunnel program to determine if the geometric differences between the full-scale test vehicle and the wind-tunnel model produced the aerodynamic and fuel consumption differences discussed above. Figure 20 shows that as the configurations approached the representation of the
24
%FEI
WT = GTRl Small-scale Wind Tunnel Tests, (from Figure 1) TRC= Full-scale Track Test at TRC
13 12
-
LO 9
8
,Run 157, 2 Side Slots Only
7
6 5
4
3 2 I 0
0
0.01
0.02
0.06 0.07 0.08 0.09 Blowing Momentum Coefficient, C @
0.03 0.04 0.05
0.1
0.11
Fig. 19 Comparison of wind-tunnel results to TRC track results."
R. J. ENGLAR
372
Comparative CD vs Ck, MTF 065
0.85
0.80
0.75
0.70
CD 0.65
0.60
0.55
0.50
0.45
0.40 0.00
0.02
0.04
0.06
0.08
0.10
0.12
0.14
0.16
Momentum Coefficient, C,
Fig. 20 Updated wind-tunnel test results: Drag change with configuration variation and with variations in blowing.
full-scale test vehicle (Run 239), both blown and unblown drag increased. These tests are further detailed in Refs. 14 and 15. Figure 21 compares the percentage drag reduction resulting from each configuration change, whereas Fig. 22 shows the predicted change in %FEI due to each. Major problems on the full-scale rig were the lower surface fairing with aft facing step in front of the bottom blowing slot and the partial gap between tractor and trailer. A comparison in Fig. 22 of Run 239 (model most like the blown full-scale test vehicle) with Run 36 (most like the standard tractor-trailer vehicle) shows that only a 2% FEI occurs for the unblown vehicle and only 7% for the blown one. This confirms the trends of Figs. 18 and 19, and explains the causes of the less-than-expected track-test results. We have since conducted further testing to improve the final PHV configurations in anticipation of a second on-road fuel economy test at TRC. Note from Fig. 22 that if we convert the full-scale PHV test vehicle to a blown configuration much more like the one in Run 205, we can expect FEIs of 16% unblown and 23% blown, which will be very significant results. That plan to reconfigure the test truck and retest is now under way.
IMPROVING PERFORMANCE OF AUTOMOTIVE VEHICLES
373
55
50 45
40
%CD 35 30 25 20 15 10 5 0
36
239
212
210
209
207
205
Test Run No.
Fig. 21 GTRI model drag reductions relative to Run 36 (configuration closest to baseline HV) (positive AC, here is drag reduction).
30
25
20
%FEI 15
10
5
0
36
239
212
210
209
207
205
Test Run No.
Fig. 22 Equivalent fuel efficiency increase (%FEI) relative to Run 36.
R. J. ENGLAR
374
VI. Pneumatic Sport Utility Vehicles (PSUVs) A. Background An analysis of vehicle fuel usage rates in the United States (Fig. 23)16,17shows that, as of about 2001, S W s and light trucks consumed more total fuel in the United States than either automobiles or HVs. It thus seemed quite relevant to determine if this pneumatic technology would be as beneficial to S W s as to HVs, perhaps even more so in the total picture. To prepare for a full-scale evaluation of the pneumatic concept applied to a sports utility vehicle, we acquired the use of the Georgia Tech FutureTruck vehicle, a 2000 Chevrolet Suburban SUV. Preliminary wind-tunnel testing of the conventional SUV was first conducted to determine baseline aerodynamic characteristics and flow separation point locations (Figs. 24 and 25). The unmodified baseline GM Suburban S W was installed on the six-component balance of the Lockheed 16 x 23 ft subsonic wind tunnel in Marietta, GA. Figure 26 shows aerodynamic force and moment variations for the unblown vehicle as functions of yaw angle, and confirms that side winds can have a significant effect on the performance and stability of these large SUVs (much like the HVs). The conventional Suburban was then modified into the pneumatic SUV configuration for the blowing tests. We had received an additional aft door assembly for the Suburban, donated by the GM Technical Center. The modification was conducted at the prototype shop of our team member Novatek, Inc. in Smyrna, ,
150,~
Source: Refs. 16 and 17 BGY=Billions of Gallons per
140.0 130.0
-
120.0
-
iinn
A n n
'w'w 20.0
1
E
10.o
0.0
~
L
, , , ,
1970
'
Repon Date "
1975
' '
"
1980
'
1985
' ' '
' ' '
1990
'
~
(Class 2b-8) & Buses (No Military Vehicles)
, ' ' .'I , ' ' " ,
1995
2000
2005
"
' I , '
2010
"
'
"
2015
"
d
2020
Year
Fig. 23 Highway energy usage comparisons (billions of gallons per year) by vehicle type.
IMPROVING PERFORMANCE OF AUTOMOTIVE VEHICLES
375
Fig. 24 GM Suburban test vehicle undergoing smoke flow visualization in the Lockheed 16 x 23 ft wind tunnel test section.
Georgia. Because it was impractical to cut away the rear sheet metal and door structure of the Suburban, we simply added blowing plenums, slots and turning surfaces onto the outside of the donated door. This modification installed on the vehicle is shown in the Lockheed Low Speed Wind Tunnel in Fig. 27. The blowing slots were adjustable and the TE jet turning angles could be changed. Blowing coefficient was variable, and mass flows, pressures, and jet velocities were measured to enable online calculation and setting of C .,
Fig. 25 GM Suburban test vehicle undergoing tuft flow visualization in the Lockheed wind tunnel.
R. J. ENGLAR
376 0.8 0.7 0.6
0.25
I"""
Pneumatic SUV, Lockheed LSWT T1835,10/18/02 Yaw Sweep, q=12.54 psf, V=71.7 mph, Run 3 Conventional GM Suburban
0.5
0.20
0.15
0.4
0.10
0.3 0.2
0.05
CD, CLfO.1 Cside
-0.00
0.0
-0.1 -0.05
-0.2 -0.3
-0.10
-0.4
-0.15
-0.5
-0.6
-0.20
-0.7 -0.8 -16
-12 Nose Left
-8
-4
0
Yaw Angle,
4
v, deg
8
12
-0.25 16 Nose Right
Fig. 26 Resulting aero forces and moments as functions of yaw angle for baseline Suburban.
B. PSUV Test Results Flow visualizations taken with blowing activated on the pneumatic vehicle showed significant attachment of flow over the new curved aft surfaces and a contracting of the jet wake behind the vehicle. The wind-on, blowing-on data showed different behaviors for the different TE configurations. Greater TE turningsurface angle produced greater jet turning, but also greater suction on these TEs, the latter causing an incremental drag force. The resulting total drag is shown in Fig. 28 for four different blowing configurations. Notice that for some configurations, initial drag reduction reached a minimum point, followed ., This drag reduction at lower blowing is of by drag increase with higher C the order of 3 to 4.15 times the input blowing coefficient, representing as much as a 415% return on the jet momentum invested. Note that when increased blowing yields a rise in drag for some of the configurations, this represents an opportunity for an aerodynamic braking system. What is needed, of course, is an onboard control system to switch from drag reduction to braking if requested by the driver. Note also that the configuration with a 45 deg turning surface on all exposed TEs continued to reduce drag with increased blowing, although at a lesser rate of reduction. Also, the blowing-off drag coefficient for these
IMPROVING PERFORMANCE OF AUTOMOTIVE VEHICLES
377
Fig. 27 Modified PSUV blown test vehicle in the Lockheed Low Speed Wind Tunnel.
V=50mph PSUV Drag Variation with Blowing, V=SOmph 0.56
Run 12, 45° Top, Bottom & Bottom Sides; 90° Top Sides
0.54
Run 11, 45° Top & Bottom; 90° Sides
0.52
CD 0.50
Run 13, 180° All 4 Sides 0.48
Run 14, 45° All Sides 0.46
0.44 0
0.01
0.02
0.03
0.04
0.05
0.06
0.07
Momentum Momentum Coefficient, Coefficient,C Cb µ
Fig. 28 PSUV drag coefficient changes with varying C,.
0.08
R. J. ENGLAR
378
nonoptimized pneumatic configurations was the same as that of the stock reference Suburban tested earlier, indicating no blowing-off drag penalty for installing this system on a typical SUV. An additional benefit of the blowing system is its ability to provide increased safety of operation. Aerodynamic braking was already mentioned, but Fig. 29 shows an additional strong potential. To counteract the adverse effects of side winds on both yawing and rolling moments shown in Fig. 26, we tested blowing of only one side slot, the left side. In Fig. 26, the baseline SUV is directionally unstable (for instance, nose-left yaw produces nose-left negative yawing moment, which tends to yaw the vehicle more), but blowing on the left side produces an aft aerodynamic side force to the left and a restoring yawing moment that returns the SUV's stability. Figure 29 shows the amount of left-side blowing required to eliminate the destabilizing yawing moment at each of three side-wind angles 4. In each case, blowing at a slightly higher rate produced yaw in the opposite direction, so that the vehicle's stability in either direction could be controlled by varying blowing alone. It is to be noted from the above that we have not yet achieved the optimum configuration to maximize drag reduction and yawing moment generation while requiring minimum blowing input, but we have otherwise verified that blowing on SUVs can be a powerful means to reduce or increase drag as needed, and to increase vehicle stability, all with no external moving parts.
0.09
Nose Right
PSUV Aero Data, LSWT Test 1853, Run 17, Left Side Slot ONLY, Cmu Sweep, V=50 mph, All Turning Corners are 45"
1
. . .
,
. .
.
,
. .
.
,
. . .
,
. . .
,
. . .
0.08
0.07 0.06
required to offset
v=-lOounblown
0.05
Cyaw 0.04
C+,,
0.03
required to offset
~ = - 5 'unblown
0.02 0.01
O
1 0
0.01
0.02
0.03
0.04
0.05
0.06
Left Slot Momentum Coefficient, Cmu
Fig. 29 Yawing moment control by blowing the left-side slot only.
IMPROVING PERFORMANCE OF AUTOMOTIVE VEHICLES
379
In a related application, GTRI and Novatek are also currently developing a patented aerodynamic heat exchanger that is based on these pneumatic princ i p l e ~ . ’ ~This ” ~ can be combined with the above devices to further reduce vehicle drag by reducing the drag associated with the conventional vertical radiator and related cooling system, while also adding favorable aerodynamic and control characteristics to the vehicle.
VII. Conclusions Under DOE-sponsored research programs, GTRI and its teammates on the DOE Pneumatic Heavy Vehicle project have completed experimental investigations to confirm the use of pneumatic aerodynamics on these vehicles. We have verified these capabilities by designing, fabricating, and testing smallscale PHV models in three separate wind-tunnel tests, and by designing, fabricating, and conducting three road or track tests of a full-scale PHV demonstrator. We have also conducted full-scale, wind-tunnel tests of this technology applied to a typical S W . It has thus been verified that these blowing concepts can reduce aerodynamic drag, favorably modify other aerodynamic characteristics, and thus improve the performance, stability and control, handling qualities, safety of operation, and economics of both HV and SUV. The very favorable capability of controlling all aerodynamic forces was shown for the PHV and pneumatic SUV configurations, as was the ability of a no-external-moving-part pneumatic control system to restore directional stability by eliminating unstable yawing moment and providing counter-yaw in the opposite direction. The preceding test programs and analyses have confirmed the following capabilities for pneumatic aerodynamics applied to HVs or SUVs. 1) Pneumatic devices, using one to four blowing slots and nonmoving downstream jet turning surfaces on HVs and SUVs, have reduced drag by up to 84% in tunnel tests. This is a result of the prevention of flow separation and increase in base pressure on the rear of the vehicle. Recent tunnel tests of a new PHV configuration soon to be tested full-scale indicate FEI of up to approximately 23%, corresponding to about 46% CD reduction at highway speeds. 2) 2)Specific blowing on only certain of the slots can cause a deliberate increase in drag, which can be used for rapid-response aerodynamic braking. 3) Specific blowing on only one side slot can cause a deliberate increase in side force and yawing moment to overcome the directional instability of these flat-sided vehicles caused by side winds and/or wind gusts. 4) Blowing on only the top slot can cause a deliberate increase in lift to reduce tire rolling resistance and thus increase fuel economy; or blowing on only the bottom slot can deliberately increase down force and thus provide an increase in load on the wheels to increase both traction and braking. 5 ) Because blowing can be quickly directed to whichever slot it is needed in, these devices provide a very-rapidly-responding pneumatic control system without external moving parts. Integrated with an onboard sensor and controls, a pneumatic system could thus control all aerodynamic forces and moments acting on HVs and S U V s and increase operational safety.
380
R. J. ENGLAR
6 ) Pneumatic reduction of the vehicle’s wake should lead to reduced spray and reduced effect on following vehicles, and thus also increase safety of operation. 7) Low-pressure blowing required could be supplied from onboard sources such as a turbocharger or supercharger, or from an existing auxiliary engine, such as is currently used for HV refrigeration units. 8) These pneumatic aerodynamic systems can be integrated with the patented GTRI/Novatek aerodynamic heat exchanger to further increase fuel economy from additional drag reduction and reduced radiator size requirements for cooling. 9) Nonmoving external components can yield an all-pneumatic system and components with very small (if any) component drag. These small-size aft trailer extensions should incur no vehicle length limitations. 10) For safety, stability, and/or economy, positive use can be made of aerodynamic forces/moments (lift, down force, side force, yaw, roll), which are currently not employed to influence HV or SUV operation. The relevance of the preceding results is shown by the DOE fuel usage data of Fig. 23. It appears that it is time to approach this problem for HVs, light trucks, and S U V s in order to reduce the deficit (“gap”) shown between U.S. oil usage and domestic supply. The concepts demonstrated by these pneumatic vehicle results suggest that certain options are now available to do so. In addition to HV and S U V application, the above results appear quite promising for other forms of automotive vehicles. Clearly, buses, motor homes, and trains are also prone to large drag values and directional stability issues due to aft flow separation and large vertical panels exposed to side winds. Of course, the application of improved blown aerodynamics to increase the performance, traction, braking, and handling of race cars is a very related and promising subject.
VIII. Recommendations The preceding aerodynamic data confirm the PHV as a viable concept for improving the aerodynamic performance, economy, stability, handling, and safety of operation of large tractor-trailers. The following recommendations are made to suggest a meaningful continuation of this program. 1) Conduct additional wind-tunnel evaluations to further reduce the required blowing momentum needed from an air source on board the tractor-trailer rig or S U V . Include slot height variation, improved blowing surface geometry, alternative jet turning surface geometry, pulsed blowing, or other innovative means. 2 ) Feasibility studies are needed, where the above results are transferred to the HV and SUV industries and interactions occur with tractor, trailer, and S U V manufacturers, as well as with engine manufacturers, turbocharger builders, or other possible air-supply specialists. Blowing sources and required compressor power must yet be considered. 3 ) Conduct additional full-scale, on-the-road demonstration and development of this technology with improved aerodynamic configurations.
IMPROVING PERFORMANCE OF AUTOMOTIVE VEHICLES
381
Acknowledgments The author wishes to thank Sidney Diamond, Jules Routbort, and Rogelio Sullivan of DOE for their continued support and encouragement of this work, as well as Victor Suski for the continued very valuable involvement of the American Trucking Associations. The technical assistance of Ken Burdges of Novatek, Inc., Skip Yeakel of Volvo, and Charlie Fetz of Great Dane is also greatly appreciated, as are the experimental efforts of GTRI Co-op students Graham Blaylock, Warren Lee, Chris Raabe, Erik Kabo, and Brian Comer from the Georgia Tech School of Aerospace Engineering, and researchers Rob Funk and Paul Habersham of GTRI.
References ‘Hammache, M., Michaelian, M., and Browand, F., “Aerodynamic Forces on Truck Models, Including Two Trucks in Tandem,” Society of Automotive Engineers Paper 2002-01-0530, Feb. 2002. ’Englar, R. J., Hemmerly, R. A., Taylor, D. W., Moore, U. H., Seredinsky, V., Valckenaere, W. G., and Jackson, J. A., “Design of the Circulation Control Wing STOL Demonstrator Aircraft,” AIAA Paper 79-1842, Aug. 1979. 3Englar, R. J., and Applegate, C. A., “Circulation Control-A Bibliography of DTNSRDC Research and Selected Outside References (Jan. 1969 to Dec. 1983),” DTNSRDC Rept. 84/052, Sept. 1984. 4Englar, R. J., “Circulation Control Aerodynamics: Blown Force and Moment Augmentation and Modification; Past, Present and Future,” AIAA Paper 2000-2541, June 2000. 5 Englar, R. J., Smith, M. J., Niebur, C. S., and Gregory, S. D., “Development of Pneumatic Aerodynamic Concepts for Control of Lift, Drag, and Moments plus Lateral/ Directional Stability of Automotive Vehicles,” Society of Automotive Engineers Paper 960673,26-29 Feb. 1996. 6 Englar, R. J., “Development of Pneumatic Aerodynamic Devices to Improve the Performance, Economy and Safety of Heavy Vehicles,” Society of Automotive Engineers Paper 2000-01-2208, 19-21 June 2000. ’Gutierrez, W. T., Hassan, B., and Rutledge, W. H., “Aerodynamics Overview of the Ground Transportation Systems (GTS) Project for Heavy Vehicle Drag Reduction,” Society of Automotive Engineers Paper 960906, June 1996. ‘Englar, R. J., “Advanced Aerodynamic Devices to Improve the Performance, Economics, Handling and Safety of Heavy Vehicles,” Society of Automotive Engineers Paper 2001-01-2072, May 2001. 9 Englar, R. J., “Development and Evaluation of Pneumatic Aerodynamic Devices to Improve the Performance, Economics, Stability and Safety of Heavy Vehicles,” DOE Quarterly Progress Rept. No. 14, April-June 2002. “Englar, R. J., “Preliminary Results of GTRI/DOE Pneumatic Heavy Vehicle Tuning Tests,” GTRI Rept. A-5871, March 2002. “Englar, R. J., “Preliminary Results of the Pneumatic Heavy Vehicle SAE Type-I1 Fuel Economy Test,” GTRI Draft Rept. Sept. 2002. ”Dotson, R., “SAEJ1321 Class-Eight Truck Aerodynamic and Tire Comparison Fuel Economy Tests,” Transportation Research Center Report, Project 20020465, Sept. 2002.
382
R. J. ENGLAR
13Englar, R. J., “Development and Evaluation of Pneumatic Aerodynamic Devices to Improve the Performance, Economics, Stability and Safety of Heavy Vehicles,” DOE Quarterly Progress Rept. No. 15, July-Sept. 2002. 14Englar,R. J., “GTRI Updated Wind-Tunnel Investigation of Pneumatic Heavy Vehicle Road-Test Configurations,” GTRI Rept. Projects A-587 1 and A-6395, Jan. 2003. ‘’Englar, R. J., “Drag Reduction, Safety Enhancement and Performance Improvement for Heavy Vehicles and SUVs Using Advanced Pneumatic Aerodynamic Technology,” 2003 SAE International Truck and Bus Meeting and Exhibition, Society of Automotive Engineers Paper 2003-01-3378, Nov. 2003. ‘6“Transportation Energy Data Book: Edition 19,” DOE/ORNL-6958, Sept. 1999. ”“EIA Annual Energy Outlook 2000,” DOE/EIA-0383 (2000), Dec. 1999; also AEO 2001. “Gaeta, R. G.,Englar, R. J., and Blaylock, G., “Wind Tunnel Evaluations of an Aerodynamic Heat Exchanger,” Proceedings of the UEF Conference The Aerodynamics of Heavy Vehicles: Trucks, Buses and Trains, Dec. 2002. ‘’Gaeta, R. J., and Englar, R. J., “Pneumatically Augmented Aerodynamic Heat Exchanger,” Paper #18 presented at the NASA/ONR Circulation Control Workshop, March 2004; also to be published in Workshop Proceedings.
Chapter 14
Aerodynamic Heat Exchanger: A Novel Approach to Radiator Design Using Circulation Control Richard J. Gaeta,* Robert J. Englar,+ and Graham Blaylock* Georgia Institute of Technology, Atlanta, Georgia
Nomenclature C, = pressure coefficient or specific heat at constant pressure CL = section lift coefficient CD = section drag coefficient C , = momentum coefficient m, = coolant mass flow rate, gal/min q = freestream dynamic pressure, psf Q = heat energy rejected by coolant, kW s = wing reference area, ft2 Tci, = inlet coolant temperature, O F T,,,, = outlet coolant temperature, O F V , = freestream velocity, mph a = angle of attack I. Introduction A. Motivation ROPER aerodynamic design of automotive vehicles lead to improved fuel efficiency. This usually means that aerodynamic drag, both profile drag and friction drag, are minimized. Design strategies for low profile drag include
P
*Senior Research Engineer, Aerospace and Acoustics Technologies Branch, ATAS, Georgia Tech Research Institute. Associate Member AIAA. 'Principal Research Engineer, Aerospace and Acoustics Technologies Branch, ATAS, Georgia Tech Research Institute. Associate Member AIAA. *Undergraduate Student, School of Aerospace Engineering. Copyright 0 2005 by the authors. Published by the American Institute of Aeronautics and Astronautics, Inc., with permission.
383
384
R. J. GAETA, R. J. ENGLAR, AND G. BLAYLOCK
establishing small cross-sectional areas and gentle transitions from the front to the rear of the vehicle. There is a long history of streamlining automotive vehicles, which began in earnest in the 1920s and continues to the present.' The majority of the work in this area has been aimed at reducing the profile drag.' In general, vehicle size, and thus its frontal area, is largely determined by other requirements, such as engine size and passenger room. Heat exchangers used in these vehicles are, typically, finned radiators that are positioned in the engine compartment away from the freestream flow, which is rammed into the vehicle body through a duct or an open area at the front. The radiator must pass a sufficient amount of air through its core to remove engine heat. However, its location and how the vehicle is shaped around are largely governed by the engine placement and the vehicle styling. A conventional, state-ofthe-art radiator is installed perpendicular to a freestream flow and employs part of the freestream total pressure to provide a pressure drop that aids the heat transfer across the radiator. Unfortunately, this also produces large aerodynamic drag coefficients on the vehicle. The frontal area needed for engine cooling air flow varies in the extreme from heavy vehicles like tractor-trailer rigs to high-performance race cars (Fig. 1). Profile drag could potentially be reduced by allowing the radiator to assume a smaller frontal cross-sectional area relative to the oncoming flow. A novel approach to this problem is the aerodynamic heat exchanger (AHE) concept, which starts with the notion of housing the heat exchanger in a low-drag package: a wing.
B. Aerodynamic Heat Exchanger (AHE) Concept All current liquid-cooled internal combustion engines used in automotive vehicles use heat exchangers that rely on stagnation pressure drop across a porous flat plate. Air flow from the freestream is directed either through a duct or through louvers to reach the face of the heat exchanger. This pressure difference is large, but it occurs at the price of a large drag force. A wing is a device that naturally produces a pressure difference but in a way that produces a low drag force (Fig. 2). The pressure difference produced by a wing is not as great as that produced by the stagnation flow across a conventional radiator, but by employing established pneumatic flow control techniques, the wing lift (or
Fig. 1 Radiators for large heavy vehicles and passenger cars present large frontal area to oncoming flow, thus contributing to profile drag. High-performance race cars use pods or ram scoops that are necessary tradeoffs for low profile drag designs.
AERODYNAMIC HEAT EXCHANGER USING CC
Flat Plate
Wing
385
AP at expense of high drag
P with low drag
Fig. 2 A wing has an order of magnitude lower drag coefficient than a flat plate and has a mechanism to produce a pressure differential needed for heat transfer.
pressure difference) can be radically augmented. Figure 3 shows the AHE concept in conjunction with a blown elliptical airfoil, which can generate suction pressure coefficients of AC, = - 5 to - 6 across the center compared to AC, = +0.4 to +0.5 across a standard radiator core. This concept is embodied in a patent that was granted to GTRI and Novatek, Inc. in 2000 and it involves the use of a very effective high-lift airfoil section to generate the pressure differences needed across a conventional automotive radiator.’ As the natural pressure difference is formed by flow over the wing, the difference in static pressure forces air through the porous thickness of the wing. The greater the lift or pressure difference, the greater the flow through the wing. The AHE device embeds the radiator core within an airfoil shape aligned parallel to the wind flow. Blown airfoils can generate suction rises on the order of 10 to 15 times the conventional radiator pressure drops (typically 40-50% of freestream total pressure). Thus the potential exists to enhance engine cooling and reduce aerodynamic drag. This pneumatic-based lift augmentation concept, also known as circulation control (CC), is based on the physics of the Coanda e f f e ~ t This . ~ effect postulates that a fluid stays attached to a curved
h i t o Reswe A?
Porous Heat Conducting Material
Fig. 3 AHE concept with pneumatic lift augmentation control. Pressure difference arising from lift provides air to the heat exchanger in a low drag envelope.
386
R. J. GAETA, R. J. ENGLAR, AND G. BLAYLOCK
surface as it flows over it by virtue of a balance between the pressure gradient normal to the surface and the centrifugal force caused by the streamline curvature. A significant amount of study of this effect and how it can augment the lift of a wing can be found in the literature for both fixed and rotating Figure 4 shows how the lift on a symmetric airfoil can be controlled by changing the momentum of air blown through a thin slot at the trailing edge (TE) of the wing. The design challenge for the AHE is to provide enough flow through the wing (via porosity), but still retain a high enough pressure difference to create aerodynamic force, if needed. This novel heat exchanger concept has the potential to become a dual-use system for automotive vehicles by providing both aerodynamic advantages and cooling simultaneously. A proof-of-concept model of the AHE was built and tested in a low-speed wind tunnel at GTRI to evaluate its feasibility. Several radiator core configurations were tested for their aerodynamic and heat-exchanger performance. The technical approach and results of this testing are presented in Secs. I1 and 111.
11. Technical Approach A. Facilities and Experimental Setup The testing of the AHE model was performed in GTRI’s Model Test Facility. This facility is a closed-return wind tunnel with an operating dynamic pressure range of 5 to 50 psf. The flow is conditioned upstream of the test section with
High Velocity Jet Sheet
Increasing Momentum Ratio, C, Increases Velocity on upper surface, thus increases lift [see Coanda Effect, Circulation]
Circulation Control produces up to 8000% increase in CL relative to momentum force input
Phenomenal Increase in P
Fig. 4 Controllable lift, thus controllable heat transfer, with pneumatic flow control.
AERODYNAMIC HEAT EXCHANGER USING CC
387
honeycomb screens and the nominal freestream turbulence is approximately 1.0%. The test section is 30 x 30 in. and aerodynamic forces are measured with a six-component balance attached to a turntable for easy changes of the model’s angle of attack. The airfoil shape chosen for the AHE concept was elliptical, with a round TE, similar to that shown in Fig. 4. The airfoil has a plenum near the TE that can be pressurized with air to produce a variable amount of flow through a TE slot for increased circulation or lift. Internal pressure transducers and flow meters measured blowing air mass flows, velocities, and blowing coefficients. Static pressure taps were located on the pressure and suction side of the airfoil at approximately midchord. Data from these taps were used to compute an average ACp. The baseline configuration used a nonporous center section to generate a reference for aerodynamic performance. Three AHE radiator configurations were tested by fabricating a two-dimensional airfoil with a reconfigurable center section along with three porous center sections. The elliptical wing with the radiator core was installed vertically in the wind tunnel and was attached to the force balance. The airfoil was connected via flexible hoses to a three-phase electric 3600 W water heater. Water was heated and pumped into one side of the wing and into an inlet reservoir attached to the radiator. After the water made its way through the radiator core, it exited into an outlet reservoir and into the water heater, closing the coolant loop. Figure 5 shows a schematic of the coolant flow path. Coolant mass flow was measured with a water flow meter and thermocouples were placed in both inlet and exit reservoirs to monitor the temperature drop across the core. The coolant mass flow and temp-
Inlet Reservoir
Outlet Reservoir
Fig. 5 Schematicof AHE experimentshowing the coolant path through the test airfoil.
3aa
R. J. GAETA, R. J. ENGLAR, AND G. BLAYLOCK
eratures were acquired using a LabView program, where a simple heat balance was used to quantify heat rejection of the coolant to the air passing through the airfoil into the tunnel. The heat transferred from the coolant can be expressed as
A typical run for a given radiator configuration would include a “sweep” of slot blowing pressure at constant angle of attack and tunnel speed, to record and evaluate aerodynamic characteristics. Then, for each radiator core installed, the coolant lines were added (these would have caused balance tares during the aero runs) and temperature data were taken at constant coolant flow rates for variable blowing pressures. Variation in tunnel speeds was also conducted for the radiator airfoils at constant flow rates while varying blowing pressures. For reference, the conventional Visteon radiator was evaluated without blowing or airfoil frame but perpendicular to the freestream flow so as to simulate a standard radiator’s cooling characteristics. All aerodynamic characteristics are based on a wing planform area of 2.871 ft2 and the blowing momentum coefficient is defined as riZV, c, = 9s
It should be noted that the model airfoil is mounted inverted in the tunnel, with negative lift (positive down force) towards the ground as the lifting side of the airfoil is towards the road, and negative angle of attack a is LE downward.
B. Test Articles All radiator test articles were made to fit in the center section of the twodimensional elliptical wing. The nominal dimensions of the radiators are 8 x 13 x 1.42 in. In addition to a solid wing configuration, the radiator types of Secs. 1I.B.1-1I.B.3 were tested. 1.
Conventional Aluminum Fin Core This core was a conventional aluminum finned core used in a Formula SAE race car operated by the Georgia Tech Motorsports Club. It had relatively low pressure drop or a high porosity and was produced by the Visteon Company. Each radiator core had cooling tubing passing through internal channels or through the foam core. Sealing of coolant leaks was a significant problem for the Visteon core. Note in Fig. 6, the coolant channels marked in red stripes were not able to be sealed within the core, and thus were taped over with metal tape to prevent leakage into the airfoil. Thus the Visteon radiator shown was tested with only 10.5 of its 18 passages open, or only 58% flow capacity. 2.
ORNL Very Dense Graphite Foam Core ORNL supplied a radiator that had the same planform dimensions and was in the same envelope as the Visteon radiator, but was made from solid pieces of carbon-graphite foam material. This material has phenomenal heat conduc-
AERODYNAMIC HEAT EXCHANGER USING CC
389
Fig. 6 Conventional aluminum finned radiator core made by Visteon, showing coolant passage inlets.
tivity properties." Although it is porous, the bulk density is such that it has a significant pressure drop. Brass tubes were press fit into the foam to carry the coolant through the material for heat exchange, as shown in Fig. 7. 3. ORNL PorouslSerpentine Graphite Foam Core A second ORNL supplied radiator core consisted of smaller carbon-graphite foam fins arranged in such a way that flow could follow the serpentine, as shown in Fig. 8. These were brazed to narrow water channels in a manner similar to the aluminum radiator. The manufacturing of this core was such that some of the coolant passages were blocked off, so its full heat rejection potential was not realized. Furthermore, it was made about half an inch thinner than the thickness of the wing, so a perforated sheet had to cover the wing to maintain smooth flow. Figure 9 shows two different AHE radiator core configurations installed in the wind-tunnel test section.
111. Results A. Note on Measurement Uncertainties The aerodynamic data presented in this chapter were acquired from a sixcomponent force balance that supported the test article in the wind tunnel. The accuracy of the load cells are approximately 1% of the reading. All pressure measurements were acquired with piezoresistive gauges with f0.5% reading accuracy. Omega thermocouples were used that were accurate to within a
390
R. J. GAETA, R. J. ENGLAR, AND G. BLAYLOCK
Fig. 7 Dense ORNL carbon-graphite foam radiator core.
degree Fahrenheit. Perhaps the largest uncertainty was in the water mass flow rate measurement. An Omega flow meter using a turbine wheel was used to obtain the liquid flow rate. The manufacturer specified the meter to have a 0.5% of reading accuracy. In practice, the repeatability of our data acquisition system was approximately 10% of reading accuracy. This accounts for most of the observed scatter in the heat transfer results presented. Weighted curve fits are used to signify trends in the data.
Section A-A Fig. 8 A more porous carbon-graphite radiator core.
AERODYNAMIC HEAT EXCHANGER USING CC
391
Blowing Slot
Tufts
Pressure Taps
ConventionalVisteon Radiator Dense Graphite Foam Core
Fig. 9 AHE installed in GTRI low-speed wind tunnel showing two radiator core configurations.
B. Aerodynamic Results The aerodynamic portion of the tests (i.e., the radiator installed and blowing applied, but no coolant hoses connected and no coolant flowing) was conducted first to evaluate the effects of porous sections in the center of the lifting airfoil. Figure 10 shows pressure coefficients plotted as functions of blowing, all at a = 0 deg. Increasing blowing C, dramatically increases the static pressure differential (- AC, avg), implying suction on the lifting (down force) side. The pressure coefficients (as defined in Fig. 10) are averaged between the pressure and suction side of the radiator, and the suction rise is equal to the pressure drop. As the radiator gets more porous, the AC, is reduced. For reference, the pressure drop of the conventional radiator at 90deg is shown (AC, = +0.4 to +0.5) along with a blown wing. The latter’s AC, is a factor of 7 times greater. Note that the ORNL dense foam radiator performs almost exactly as the baseline blown airfoil with no radiator installed. The implication is that this foam is so dense as to allow little if any air to pass through the radiator core. Also note that the conventional Visteon radiator reaches a pressure drop commensurate with the typical 90-deg radiator when the blowing momentum coefficient approaches 0.4, whereas the two other configurations reach higher pressure drops after a momentum coefficient of 0.05. This implies that the porous wing can produce the magnitude of pressure drop across the radiator needed to remove the required heat from the coolant. Figure 11 shows how blowing and porosity affect the lift and drag. These results conform directly to the pressure differences in Fig. 10. As porosity increases, lift decreases and drag increases, but still increased blowing is very effective. As down force (- CL)increases, due to blowing, the high circulation
R. J. GAETA, R. J. ENGLAR, AND G. BLAYLOCK
392
MTF059 Pressure Increment Across Airfoil Radiator, q=5 psf, a=O0
Conventional Radiator at c(=90"
0.5
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
0.5
Cmu
Fig. 10 Average pressure coefficient as a function of blowing momentum for various AHE configurations.
around the airfoil causes the LE to separate and thus the discontinuity in the lift curves to occur. This can be corrected by improving the LE shape. There is still improvement to be realized: the 20%elliptic airfoil of Ref. 5 is a thicker airfoil (i.e., has a greater LE radius) version of the current baseline blown ellipse airfoil, and it shows no sign of separation, reaching a C, of - 8 or more. Thus great down force potential is confirmed with blowing (no increase in airfoil angle of attack is necessary) and this will impact the heat transfer potential. The aerodynamic test also confirms that flow through the radiator core can be varied by controlling the circulation with trailing edge (TE) blowing.
C. Heat Transfer Results Results for the conventional radiator core indicated that a maximum coolant temperature drop of about 5" F was possible for a flow rate of 5 gal/min with
AERODYNAMIC HEAT EXCHANGER USING CC
393
0.5 0.0
-0.5 -1.0 -1.5
-2.0 CL
-2.5
-3.0
-3.5 -4.0
-4.5
-5.0 -5.5
0.00
0.10
0.30
0.20
0.40
0.50
CD
Fig. 11 Drag polar as a function of blowing momentum for various AHE configurations.
a 64 mph freestream velocity. Figure 12 shows coolant temperature drop for the Visteon core as a function of blowing coefficient C, and coolant mass flow. Note that for the smaller coolant mass flows, larger temperature drops are observed. This is quite likely caused by the longer exposure of the coolant to the heat exchanger (longer residence times). It should also be noted that because of fabrication anomalies, some (42%) of the coolant flow tubes were blocked off so the Visteon radiator was not flowing in a evenly distributed manner and it is likely that its performance was inhibited to some degree (see Fig. 6). Figure 13 shows the AHE heat removal as a function of C, and radiator configuration at a nominal coolant flow rate of 5 gal/min. As expected, the effect of the pneumatic lift augmentation (the increasing C ), is to increase the heat
R. J. GAETA, R. J. ENGLAR, AND G. BLAYLOCK
394 0.0 -1
.o
-2.0 -3.0
-5
-4.0
2 +
-6.0
=- -5.0 6-
-7.0
-8.0 -9.0 -1 0
0
0.05
0.1
0.15
0.2
0.25
0.3
Fig. 12 Influence of coolant flow rate on coolant temperature difference; Visteon core, V, = 64 mph.
removal rate via increasing air flow through the wing. A notable exception is the high-density carbon-graphite configuration. Figure 14 shows the heat removal for the various AHE radiator configurations at a coolant flow rate and freestream velocity that are close to nominal automotive vehicle values. It is interesting that the high-density graphite core performs as well as the Visteon core, which is somewhat surprising because it has little or no airflow through the core. This performance is likely because of the superior conductive performance of the foam; that is, almost all of the heat transfer takes place in the form of forced convection along the surface of the airfoil (both upper and lower; see Fig. 7). This result was intriguing and suggests that the heat removal can be varied and/or augmented by simply varying the turbulence level of the flow over the wing surface. There are many flow control methods (active and passive) that can aid this forced convection process. The high-density graphite radiator core was also the best aerodynamic performer. This makes this configuration all that much more attractive, because the AHE can function as an effective aerodynamic and heat transfer device. For comparison, a typical passenger automobile radiator removes about 1520 kW in normal operation for a full-sized engine. The model AHE tested here produced roughly half of this heat rejection, but with a radiator core of less
AERODYNAMIC HEAT EXCHANGER USING CC
395
10
9.0 8.0
-z E
7.0
I
-m0
6.0
8 E t
5.0
8 $
4.0
3.0
U
8
I 2.0 1 .o
0.0 0
0.2
0.4
0.6
0.8
1
CP Fig. 13 Rejected heat from three different AHE configurations; V, = 32 mph, coolant mass flow = 5 gal/min.
than half the area. When one accounts for the heat removed per square foot of radiator, it can be shown that the AHE does the heat transfer job the conventional radiator does with approximately three times less aerodynamic drag.
IV. Conclusions Initial wind-tunnel evaluations of the aerodynamic heat exchanger concept employing both conventional and ORNL graphite foam radiator cores have been performed. This new concept has been shown to adequately transfer heat at the same or similar rates as convectional radiators at 90-deg to the flow, but at much lower drag coefficients when enclosed in a lifting surface parallel to the flowfield. The dense graphite foam core of ORNL has been shown to be both an effective heat transfer medium, employing forced convection and an excellent aerodynamic surface, and allowing almost no air to pass through the wing. The following conclusions can be drawn from this proof-of-concept test of the AHE: 1) An aerodynamic heat exchanger (AHE) with pneumatic lift control was successfully tested in a wind tunnel and the basic concept was validated.
396
R. J. GAETA, R. J. ENGLAR, AND G. BLAYLOCK
Fig. 14 Rejected heat from three different AHE configurations; V , = 64 mph, coolant mass flow = 15 gal/min.
Fig. 15 AHE installation into GT Formula SAE race car.
AERODYNAMIC HEAT EXCHANGER USING CC
397
2) Lift and drag are dramatically affected by the porosity of the radiator core section, but pneumatic augmentation is still a powerful control. 3) The AHE demonstrated nonoptimized heat rejection performance, but optimized sizing should further improve results. 4) The AHE has great potential for exhibiting both controllable aerodynamic force and low drag penalty for engine cooling. 5 ) Carbon-graphite foam enables optimal performance of the radiator core within the AHE concept. It is important to note that system integration issues will pose a (surmountable) challenge to designers of cooling systems. Two important issues that need to be addressed are the production of steady high-pressure air for the pneumatic system and coolant pump size and ducting for the AHE. It is recognized that any fuel savings obtained from a lower drag configuration will be offset somewhat by the energy needed to produce the circulation control blowing air. It is the plan of GTRI to demonstrate this technology on the GT Motorsports Formula SAE race car as a technology demonstrator. Initial work has highlighted the need for good system integration design. Figure 15 shows one of the Formula SAE student cars with the AHE model being prepared for installation.
Acknowledgments The authors would like to thank James Klett and April McMillan of ORNL for being receptive to the concept of the AHE and providing funds and material for a part of this work. References ‘Hucho, W. (ed.), “Aerodynamics of Road Vehicles,” Butterworth-Heinemann, London, 1990, Chaps. 1, 3-9. ’Burdges, K. P., and Englar, R. J., “Vehicle Heat Exchangers to Augment & Modify Aerodynamic Forces,” U.S. Patent No. 6,179,077, Aug. 2000. 3Metral, A. R., “On the Phenomenon of Fluid Veins and Their Application, the Coanda Effect”, AF Translation, F-TS-786-RE, 1939. 4Cheeseman, I. C., and Seed, A. R., “The Application of Circulation Control by Blowing to Helicopter Rotors,” Journal ofthe Royal Aeronautical Society, Vol. 71, July 1966. ’Williams, R. M., and Howe, H. J., “Two-Dimensional Subsonic Wind Tunnel Tests on a 20%thick, 5% Cambered Circulation Control Airfoil,” NSRDC TN AL-176, Aug. 1970. Wilkerson, J. B., Reader, K. R., and Linck, D. W., “The Application of Circulation Control Aerodynamics to a Helicopter Rotor Model,” American Helicopter Society Paper AHS-704, May 1973. ’Englar, R. J., “Experimental Investigation of the High Velocity Coanda Wall Jet Applied to Bluff Trailing Edge Circulation Control Airfoils,” M.S. Thesis, Dept. of Aerospace Engineering, Univ. of Maryland, College Park, MD, June 1973. 8Wilkerson, J. B., Barnes, D. R., and Bill, R. A., “The Circulation Control Rotor Flight Demonstrator Test Program,” American Helicopter Society Paper AHS-795 1, May 1979.
398
R. J. GAETA, R. J. ENGLAR, AND G. BLAYLOCK
’Pugliese, A. J., and Englar, R. J., “Flight Testing the Circulation Control Wing,” AIAA Paper 79-1791, Aug. 1979. “Englar, R. J., “Circulation Control Pneumatic Aerodynamics: Blow Force and Moment Augmentation and Modification; Past, Present, and Future,” AIAA Fluids 2000 Conference, AIAA Paper 2000-2541, June 2000. “Klett, J., Ott, R., and McMillian, A. “Heat Exchanger for Heavy Vehicles Utilizing High Thermal Conductivity Graphite Foams,” Society of Automotive Engineers Paper 2000-01-2207, Washington, DC, June 2000.
1II.A. Tools for Predicting Circulation Control Performance: NCCR 1510 Airfoil Test Case
Chapter 15
Investigation of Turbulent Coanda Wall Jets Using DNS and RANS Hermann F. Fasel,* Andreas Grosst, and Stefan Wernz’ University of Arizona, Tucson, Arizona
Nomenclature A = area per unit span, m B = blowing ratio b = nozzle height, m c = chord length, m cp = wall pressure coefficient c p =jet momentum coefficient d = cylinder diameter, m f = frequency, Hz k = number of spanwise Fourier mode L = domain size, m M = Mach number R = gas constant, J/(kg K) Re = Reynolds number T = temperature, K p = pressure, kPa r = radius of curvature, m v = velocity, m/s riz = mass flux per unit span, kg/(m s) x = streamwise location (from leading edge), m y = wall-normal location (from chord), m y2 = wall-jet half-thickness, m z = spanwise location, m a = angle of attack, deg *Professor, Department of Aerospace and Mechanical Engineering. Member A I M . ‘Research Associate, Department of Aerospace and Mechanical Engineering. Member AIAA. Copyright 02005 by the authors. Published by the American Institute of Aeronautics and Astronautics, Inc., with permission.
401
402
H. F. FASEL, A. GROSS, AND
S. WERNZ
r = circulation, m2/s y = ratio of specific heats 6* = displacement thickness, m 8 = momentum thickness, m h = wavelength, m p = molecular viscosity, k3/(m s) v = kinematic viscosit m s v, = eddy viscosity, m Y,/s / p = density, kg/m3 6 = streamwise (azimuthal) angle, deg w = vorticity, l/s
Subscripts
in = inflow inside plenum jet = nozzle exit max = wall-normal maximum wall = wall, surface z = spanwise direction 6 = streamwise direction 03 = free stream Superscript
+ = wall coordinates I. Introduction ALL jets over curved surfaces have great potential for technical applications. Coanda wall jets over convex surfaces can effectively provide aerodynamic side forces or change the circulation of an airfoil. An existing application is the “No-Tail-Rotor’’ (NOTAR) helicopter. Possible future applications are the enhancement of low-speed maneuverability of underwater vehicles or high-lift wings for short take off and landing (STOL) aircraft. However, without profound understanding of the mechanisms that keep the wall jet attached to the surface for large downstream distances, any implementation of Coanda flow technology must rely on empiricism and hence requires excessive safety margins to account for unknowns. In this paper, results from numerical investigations of two separate Coanda flow experiments are presented that may help to shed some light on the relevant physical mechanisms. One of the most intriguing phenomena of the Coanda wall jet is the competition/interaction of naturally occurring streamwise and spanwise vortical structures, which are a consequence of a centrifugal, Gortler-type instability (leading to streamwise coherent structures) and a Kelvin-Helmholtz-type instability (leading to spanwise coherent structures), respectively. It can be conjectured that the intensity of these structures, both absolute and relative to each other, will significantly influence the separation location and, as a consequence, will have a key effect on the side forces that can be generated and thus on the
TURBULENT COANDA WALL JETS AND DNS AND RANS
403
effectiveness and reliability of this technique. The amplitudes and wavelengths of the coherent structures will also determine the intensity and the frequency spectrum of the associated aerodynamic/hydrodynamic noise. In addition, because both the streamwise and the spanwise structures are a consequence of hydrodynamic instabilities, instability mechanisms may be exploited advantageously for active flow control (AFC) strategies. Tani was among the first to report on streamwise vortices in a turbulent boundary layer along a concave wall.’ In his experiments he observed regularly spaced spanwise modulations of the velocity profiles, which he attributed to a Gortler instability mechanism. To compare with stability theory results for a laminar boundary layer, he assumed a constant eddy viscosity v r = 0 . 0 1 8 ~S*,~and a displacement thickness, S*= 1.38( 8 is the momentum thickness). Moser and Moin2 performed direct numerical simulations (DNS) of a curved turbulent channel flow to determine the effects of curvature in wall-bounded turbulent flows. They found stationary Gortler vortices, which had a significant impact on the mean Reynolds shear stresses and which enhanced the asymmetry of the channel flow. Sufficiently close to the wall, the mean velocity profiles followed the law of the wall. For a curved wall with curvature S*/r = 0.1 the turbulence intensities and shear stresses were, in some cases, twice as large as for a plane wall. In the Reynolds-averaged Navier-Stokes (RANS) calculations considered in this chapter, the prediction of the spreading rate depends on the turbulence model employed. Pajayakrit and Kind3 used the Baldwin-Lomax, the Dash et al. K - E , the Wilcox K--w, and the Wilcox multiscale turbulence models for the calculation of plane and curved turbulent wall jets. They tuned the model constants to obtain better agreement with experimental data for the streamwise development of the skin friction and the half-thickness of the jet. They also pointed out that the Boussinesq approximation mandates zero shear stress at the velocity peak, although it is well known that the zero shear stress location in wall jets occurs substantially closer to the wall. For the curved wall jet the nondimensional velocity profile predicted by the K - E model matched the experimental profile whereas the profile predicted by the K--W model had the velocity maximum slightly closer to the wall.
11. Investigated Configurations At first, in collaboration with an experimental effort by Wygnanski and coworkers4 a turbulent wall jet on a circular cylinder was investigated. For this configuration extensive numerical simulations, including DNS, large eddy simulation (LES), and unstead Reynolds-averaged Navier-Stokes (URANS) calculations were conducted?” The flow parameters in the simulations were chosen to match the ex eriment, with cylinder diameter d = 0.2032 m, nozzle height b = 2.34 x 10- Bm, and jet-exit velocity vjet = 48 m/s. The Reynolds number based on jet-exit velocity and cylinder diameter was Re = 6.15 x lo5 (Reb= 7,080 based on jet-exit velocity and nozzle height). The experiment was conducted in a quiescent environment. Secondly, the flow around a NCCR 1510-7067 N circulation control airfoil was computed using steady RANS. This flow configuration was posed as a benchmark problem to the CFD community for the 2004 NASA/ONR Circulation Control Workshop.’ Experiments by Abramson* on this particular airfoil
H. F. FASEL, A. GROSS, AND
404
S. WERNZ
Table 1 Elliptic airfoil CFD challenge cases
a,deg hjet, CLL
kg/ms
Case 283
Case 321
0 0.196 0.209
-8 0.182 0.184
served as a reference for the numerical simulations. The airfoil has 15% relative thickness (maximum thickness to chord length c) and a Coanda trailing edge (TE). A blowing slot is located at x/c = 0.967 with slot height b = 0.003~. The tests were conducted at a freestream Mach number M, = 0.12 for various angles of attack a. This flow configuration was computed by Slomski et al.' using the commercial flow software Fluent on computational grids with approximately 1.6 x lo5 points. Computations with the standard and the realizable K--E turbulence model only yielded realistic results for the jet momentum coefficient cp = 0.026. The jet momentum coefficient was defined as
with jet-exit velocity vjet, jet-mass flux hjet= pjetvjetb,and freestream dynamic pressure 1/2p,vk. For a higher momentum coefficient, cp = 0.093, the same two turbulence models predicted the wall-jet separation slightly farther downstream than observed in the experiment. At the even higher momentum coefficient, cp = 0.209, the jet wrapped around the entire elliptic airfoil 1.5 times when the realizable K--E model was used. Only the full Reynolds stress model predicted the correct separation locations and hence the correct overall circulation for all momentum coefficients studied. In general, turbulence models based on the Boussinesq approximation predicted separation too far downstream. Another simulation for the same flow geometry was carried out by Paterson and Baker." They studied the two workshop CFD challenge cases (Table 1) using the incompressible CFDSHOP-IOWA code. For two-dimensional RANS, the two-equation shear stress transport (SST) turbulence model by Menter" was employed. The predictions of wall-jet separation location and wall-pressure distribution (and therefore circulation) were in good agreement with the experiment.* 111. Numerical Approach
For the computational results presented in this paper two different numerical approaches were taken. Each of these approaches is tailored and optimized for certain subtasks, so computational resources can be focused effectively. When combined, they will help in understanding the different physical mechanisms involved.
A. Direct Numerical Simulations (DNS) An existing incompressible Navier-Stokes was adopted to allow for highly accurate DNS of turbulent Coanda wall jets for Reynolds numbers in the
TURBULENT COANDA WALL JETS AND DNS AND RANS
405
range of the laboratory experiments by Wygnanski and coworkers. In this code the incompressible Navier-Stokes equations in vorticity-velocity formulation are solved. The governing equations are discretized using fourth-order accurate compact differences in the streamwise and wall-normal directions in combination with a sixth-order compact filter for filtering out disturbances at grid level. The spanwise direction is assumed to be periodic and is discretized using a pseudospectral decomposition into Fourier cosine or sine series.13 This expansion reduces the number of spanwise modes by a factor of two when compared with the full Fourier transform. However, spanwise symmetry is imposed in addition to periodicity. Metric terms were included to allow for computations on orthogonal curvilinear grids. The velocity Poisson equations are solved using an iterative solver with multigrid acceleration.
B. Reynolds-averaged Navier-Stokes (RANS) calculations A multidomain, compressible, finite-volume Navier-Stokes code with highorder accurate upwind schemes was developed to allow for robust computations of complex geometries. The convective terms are discretized with fifth-order upwind schemes based on a weighted essentially nonoscillatory (WENO) extrapolation and the Roe scheme,14 and the viscous terms are fourth-order accurate. A second-order accurate Adams-Moulton method is used for time integration. Various turbulence models were implemented. The standard 1988 and 1998 K-6.1 models and the K--E model15 can be combined with both a Reynolds stress based on the Boussinesq approximation, and an explicit algebraic stress model (EASM).16 The Menter SST" and Spalart-Allmaras" models were included as well. IV. Turbulent Wall Jet on a Circular Cylinder A. Direct Numerical Simulations (DNS) The objective of our DNS on a segment of the Coanda cylinder from the experiments4 was to investigate the development of coherent structures in the turbulent flow upstream of separation and the impact of forcing on these structures and on the mean flow development. An illustration of the computational domain is provided in Fig 1. At the inflow boundary (6= -5.6 deg), a laminar Glauert wall jet with maximum velocity v ~= 50,m/s ~ and~ momentum thickness 8 = 3 mm is prescribed (Re0 = 10,000). The laminar flow is transitioned to turbulence at 6 = 0 deg using a volume forcing technique by which a time-dependent local force field is applied inside the flow through forcing terms added to the right-hand side of the Navier-Stokes equations5 For actively forcing the wall jet to enhance spanwise or streamwise coherent structures inside the flow, additional time-harmonic or steady volume forcing is applied at 6 = 0 deg. Inside a buffer domain near the outflow boundary the turbulent flow is relaminarized to prevent reflections of turbulent fluctuations from the 0utfl0w.l~Also shown in Fig. 1 is the computational grid for the present simulations. In the azimuthal direction, 573 points with constant step-size A 6 = 0.28 deg are used (Ax+ RZ 50 in wall-coordinates). An additional 100 points on a stretched grid are placed inside the buffer domain. In
406
H. F. FASEL, A. GROSS, AND
x
S. WERNZ
[mml
Fig. 1 Computational grid used for DNS.
the radial direction, 193 points are clustered toward the surface such that, throughout the domain of interest, the wall-next points are located within y+< 1 from the surface. The spanwise direction is discretized with 21 modes over a domain of width L, = 20 mm x 0. Id, resulting in Az+ = 20 between collocation points. Evidence from experiments4 and from earlier numerical investigations5 suggests that both spanwise and streamwise coherent structures are present in the turbulent Coanda wall jet. The streamwise structures develop as a result of a centrifugal, Gortler-type instability while the spanwise structures originate from an inviscid, Kelvin-Helmholtz-type instability (inflection point of velocity profile). It may be conjectured that in the natural (unforced) turbulent Coanda wall jet (under “clean” experimental conditions) the two instability mechanisms balance each other. For example, the Gortler-type, centrifugal instability and the resulting Gortler vortices may inhibit the spatial growth of the spanwise coherent structures that result from the inflectional instability. To probe this conjecture, DNS were performed, where deliberate forcing was introduced to enhance certain structures, or where the simulations were set up such that certain instability mechanisms were weakened. Three DNS cases will now be discussed. In the “unforced” case, which serves as a reference, the flow is transitioned without applying additional forcing. In the second case, the spanwise rollers are enhanced using time-harmonic volume forcing (frequency f = 340 Hz) that is two-dimensional, that is, without modulation in the spanwise direction. In the third case, streamwise vortices with a fixed spanwise wavelength are generated by steady volume forcing with a periodic modulation in the spanwise direction (A, = 20 mm). The downstream development of the streamwise structures for the three simulation cases is visualized with the iso-surface plots in Fig. 2 of the time-averaged
TURBULENT COANDA WALL JETS AND DNS AND RANS
b)
407
C)
Fig. 2 DNS of turbulent Coanda wall jet, showing iso-surface plots of time-averaged streamwise vorticity (light-shaded surfaces, clockwise rotation; dark-shaded surfaces, counter-clockwise rotation): a) “Unforced” reference; b) harmonic twodimensional forcing; c) steady three-dimensional forcing.
streamwise vorticity, we. The counter-rotating streamwise vortices are represented as light- and dark-shaded iso-surfaces ( w q = 300/s and w q = - 300/s, respectively). An impression of the spanwise vortical structures is provided with the snapshots in Fig. 3 showing gray scales of instantaneous spanwise vorticity w, averaged in the spanwise direction. When time-harmonic two-dimensional forcing is applied, the intensity of the spanwise coherent structures is strongly enhanced, as seen from a comparison of Figs. 3a and 3b. However, forcing of the spanwise structures leads to an increase in intensity of the streamwise vortices, not a decrease as may have been expected (compare a)
Fig. 3 DNS of turbulent Coanda wall jet showing instantaneous spanwise vorticity, spanwise average: a) “Unforced” reference; b) harmonic two-dimensional forcing; c) steady three-dimensional forcing.
408
H. F. FASEL, A. GROSS, AND
S. WERNZ
Figs. 2a and 2b). It is possible that the strongly enhanced spanwise structures in Fig. 3b promote the generation of streamwise vortices through a secondary instability (braid vortices), a conjecture that requires further exploration. On the other hand, by forcing the streamwise structures, the intensity of the naturally occurring Gortler vortices is significantly increased (compare Figs. 2a and 2c), whereas the intensity of the spanwise coherent structures is strongly decreased (compare Figs. 3a and 3c). The time-development of the spanwise coherent structures can be visualized nicely with their footprint on the wall, namely, fluctuations in the spanwise wall vorticity qwall. Shown in Fig. 4 for the three cases are time-space diagrams of the spanwise-averaged w ~ plotted , versus ~ ~ streamwise ~ ~ angle and time. Dark lines in the diagrams (amplitude peaks in the wall vorticity) correspond to propagating spanwise vortices inside the flowfield. A merging of these lines reflects the pairing of subsequent vortices. These pairings occur repeatedly in a subharmonic cascade. Regions of local flow separation are indicated by the black areas (negative wall vorticity) in the downstream part of the flow domain (Figs. 4a and 4b). Although two-dimesional harmonic forcing enhances the wall-vorticity fluctuations in the upstream part of the flow and leads to frequent flow separation in the downstream part of the flow (compare Figs. 4a and 4b), three-dimensional steady forcing strongly reduces both wall-vorticity fluctuations and local flow separation (compare Figs. 4a and 4c). This suggests that the presence of streamwise vortices indeed inhibits the development of spanwise coherent structures.
Fig. 4 DNS of turbulent Coanda wall jet showing time-space diagrams for spanwise-averaged wall-vorticity: a) “Unforced” reference: b) harmonic twodimensional forcing; c) steady three-dimensionalforcing.
TURBULENT COANDA WALL JETS AND DNS AND RANS
409
b)
Fig. 5 DNS of turbulent Coanda wall jet showing effect of forcing on the mean-flow development: a) Inverse-squareof streamwise mean-velocity maximum; and b) walljet half-thickness vs streamwise angle. Experimental data4 are also plotted for reference. To compensate for different initial development near the nozzle, experimental data in (a) are matched at 6 = 35 deg with the unforced case.
The results from the DNS also showed that a strengthening or weakening of the streamwise or spanwise structures changes the downstream development of the Coanda wall jet. For example, both the decay of the streamwise mean velocity and the radial spreading of the jet in the downstream direction are significantly increased in response to the forcing (Fig. 5). Individually, both streamwise and spanwise structures facilitate entrainment of low-momentum fluid from the ambient into the near-wall region of the jet, causing the observed increase in spreading and velocity decay. Although the separated flow region is not computed in our DNS, it may be conjectured that the wall jet will separate from the cylinder surface farther upstream as a direct result of the increased spreading and decay of the turbulent mean flow. This, in turn, has an effect on the side force that is being generated. However, most of the interacting mechanisms between spanwise and streamwise vortical structures have to be investigated in considerably more detail as numerous physical aspects are not yet fully understood. This understanding is essential for the implementation of the Coanda technology for practical applications.
B. Reynolds-Averaged Navier-Stokes (RANS) Calculations The applicability of the different available turbulence models for Coanda flow calculations was scrutinized in two-dimensional RANS calculations of the Coanda wall jet experiment by Wygnanski and coworker^.^ The computational grid used for these investigations is shown in Fig. 6 and consists of three blocks. The grid sizes for the blocks consist of 200 x 75, 50 x 50, and 150 x 20 cells, respectively. For the turbulence models used in these calculations
41 0 a)
H. F. FASEL, A. GROSS, AND
b)
S. WERNZ C)
Fig. 6 Computational grid used for two-dimensional RANS calculations: a) Entire grid; b) close-up of cylinder; c) close-up of nozzle region.
the laminar sublayer needed to be resolved. The y+ values of the wall-next grid points were between 0.2 and 1, and the Ax+ values were between 50 and 300. The grid resolution in the jet was between 40 and 180 times the local Kolmogorov length scale. A top-hat velocity profile was prescribed at the nozzle inflow. The ambient was quiescent. The flow was assumed to be laminar at the nozzle inflow and in the ambient. Generally, most turbulence models gave disappointing results, some to a larger degree than others. Typical results in the form of iso-contours of eddyviscosity from such RANS calculations are given in Fig. 7. The 1988 K-6.1 model facilitates the strongest turbulent mixing across the wall jet and hence leads to the fastest jet velocity decay and largest jet spreading and the earliest separation. In contrast, when the K--E or the Spalart-Allmaras model was used, the jet wrapped around the cylinder more than once. For some of these turbulence models the jet-velocity decay and jet-halfthickness are plotted in Fig. 8 against streamwise angle. When the 1988 K-6.1 model was used in combination with the EASM model, a close match of the jet-velocity decay with the measured data was achieved. However, even with this model, the downstream development of the jet-half-thickness was poorly predicted. The second-best model was the 1988 K--W model.
Fig. 7 Two-dimensionalRANS computationsof Coanda flow showing eddy viscosity normalized by laminar viscosity (Note that the K--E and S-A results are transient): a) 1988 K--O model; b) K--E model; c) Spalart-Allmaras model.
TURBULENT COANDA WALL JETS AND DNS AND RANS
a)
411
b)
Fig. 8 Two-dimensional RANS computationsof Coanda flow: a) Jet-velocity decay; b) jet half-thickness vs streamwise angle.
The shape of the normalized velocity profiles is predicted best by the K--E model (Fig. 9). The second-best results were obtained from the 1988 K--W model with EASM. However, because the predicted half-thickness was too small for all models (Fig. 8), the non-normalized velocity profiles still do not match the experimental velocity profiles. With the 1988 K--W model (with Boussinesq or EASM Reynolds stress), very good predictions of the wall-pressure distribution were possible (Fig. 9). For the EASM model the separation location was slightly closer to the experiment. When the K--E and Spalart-Allmaras models were used, the jet remained attached to the cylinder for more than 360 deg. To allow for a comparison with the K--W model results, the data shown for these two models are not from steady-state solutions but from transient solutions at a time instant before the drifting separation location had reached 6 = 360 deg. For all but the 1988 K--W model with EASM, jet spreading and velocity decay were underpredicted. Based on the DNS results one may assume that the turbulence models failed to account for (or underpredicted) the additional mixing facilitated by the strong coherent turbulence structures that are present in the flow. Because the separation location was predicted within 10% of the experimental result when the standard 1988 K--W turbulence model was used, this model was then chosen for subsequent three dimensional RANS stability investigatiom6 For these three-dimensional computations, 48 grid cells were used in the spanwise direction over a domain of width L, = 0.3d, resulting in Az+ values between 50 and 200. A periodicity boundary condition was applied in the spanwise direction. In these three-dimensional RANS simulations, several steady perturbations with a periodic modulation in the spanwise direction were introduced simultaneously at the nozzle exit, each with a different amplitude and spanwise wavelength h,(k) = L,/k, where k = 1,2,. . . represents the spanwise Fouriermode number of a perturbation. One such case is illustrated in Fig. 10. The streamwise development of these perturbations and their interaction was then studied by plotting the amplitudes of the spanwise Fourier modes representing the perturbations.
412
H. F. FASEL, A. GROSS, AND
S. WERNZ
Fig. 9 Two-dimensional RANS computations of Coanda flow: a) Velocity profiles at three downstream stations; b) wall-pressure coefficient cp = 2 ( p pce)/(pjetv?et)*
Forcing at small amplitudes allows for a comparison with linear stability theory. From the experiment by Wygnanski and coworkers4 it was found that the spanwise wavelength of the locally predominant structures scales roughly with the local half-thickness of the jet (Fig. 11). This can be confirmed by computation. When the streamwise structures were forced with larger (nonlinear) disturbance amplitudes, nonlinear subharmonic resonances could be observed (Figs. 11 and 12). The results obtained for nonlinear amplitudes depend on the relative phase between the modes. This becomes evident from the total
TURBULENT COANDA WALL JETS AND DNS AND RANS
413
Fig. 10 RANS computation of Coanda wall jet. Spanwise Fourier modes k = 1, 2 forced at nonlinear amplitudes of 0.01 and O.lvjet (mode k = 2 phase-shifted by m/2 relative to mode k = 1). Iso-surfaces of azimuthal velocity component. As the jet passes along the cylinder in the downstream direction the higher wavenumber structures disappear, while the lower wavenumber structures emerge.
b) 1o4
6 lo4
3 1o4
10-10
6
0
50
100
150
200
6
Fig. 11 RANS computation of Coanda wall jet showing amplitude of spanwise Fourier modes k: a) Linear case, all modes forced at small disturbance amplitudes; b) Fourier modes k = 1, 2 forced at large, nonlinear amplitudes of 0.01 and 0.lvjet (solid lines). Comparison with linear case (dashed lines). In particular, close to the nozzle (8= 0 deg) the growth rates for the nonlinear forcing deviate substantially from the growth rates for the linear forcing.
H. F. FASEL, A. GROSS, AND
414
a)
10-l
10-1
a
-P--
b
S. WERNZ
no phase shift d2 phase shift
\
10-l
L
10-2
0
100
200
300
6
Fig. 12 Two-dimensional RANS computation of Coanda flow for Fourier modes k = 1, 2 forced at amplitudes of 0.01 and 0.lvjetand modes 1 and 2 forced in phase and at a relative phase shift of n/2: a) Total circulation r(8)= 1 ~ 4 1dA; and b) mode amplitudes.
circulation for 6 > 150 deg These preliminary investigations suggest that both linear instability as well as nonlinear subharmonic resonance are possible viable mechanisms for the merging of the longitudinal vortices that was observed in the experiments. Based on our calculations, the linear process appears to be more likely for the present experimental conditions. However, for possible control of the Coanda wall jet, the nonlinear resonance mechanisms might also be exploited. Because RANS underpredicted the wall-normal mixing (and hence the jet-velocity decay and jet spreading), and because our DNS results clearly indicate that strong turbulent coherent structures play a dominant role in
Fig. 13 Two-dimensional FSM computation of a Coanda wall jet: a) Vorticity; b) contribution function. Because three-dimensional streamwise vortices are deliberately excluded, the two-dimensional structures have a high intensity. The spatial distribution of the contribution function clearly correlates with dominant flow structures.
TURBULENT COANDA WALL JETS AND DNS AND RANS
415
the turbulence mixing, application of our flow simulation methodology (FSM)'8919appeared to be a logical choice. With FSM, depending on the local turbulence characteristics and grid resolution, small-scale turbulent motion is modeled, while large-scale coherent structures are computed in a time-accurate fashion. Results from a preliminary two-dimensional FSM are shown in Fig. 13. Large spanwise coherent structures arise as a consequence of the inflectional wall-jet profile (Fig. 9a). The turbulence-model contribution is clearly linked to the flow structures, as shown in the right plot of Fig. 13.
V. Circulation Control Airfoil A. Case Description The airfoil-chord length was c = 8 in (or 0.2032 m). The freestream velocity was v, = 39.18 m/s, the freestream density p, = 1.226 kg/m3, and the freekg/ms. Assuming a gas constant stream molecular viscosity p, = 1.790 x of R = 287.1 J/(kg. K) and a ratio of specific heats y = 1.4, the freestream temperature can be computed as T, = (v,/w2/(yR) = 265.21 K. The Reynolds number based on freestream velocity and chord length was Re=-- pwv'ooc - 5.455
105
PCu
If the assumption p, = pjet is made, the jet-blowing ratio B = vjet/vW = c,p,c/(2pj,,b) is 5.90 for case 283 and 5.54 for case 321. However, this x 0.7 and requires the use of a comisults in a nozzle-exit Mach number pressible code. The nozzle-inflow area is Ai,/c = 0.03188. The nozzle-area ratio is 10.2.
B. Computational Grid The computational grid used for the investigations discussed here is shown in Fig. 14. The number of cells around the airfoil was 500, and the nozzle interior
Fig. 14 Computational grid for circulation control airfoil: a) Entire grid; b) closeup of airfoil and block boundaries; and c) close-up of Coanda flow region.
416
H. F. FASEL, A. GROSS, AND
S. WERNZ
was resolved by 100 x 80 cells. The resolutions of the individual blocks were 700 x 80 (block l), 40 x 40 (block 2), and 400 x 50 (block 3). This results in a total number of cells of 77,600. The total extent of the grid was 1Oc in both x and y measured from the center of the airfoil. The y+ value of the wall-next grid points was smaller than one.
C. Boundary Conditions Following common practice, velocities and temperature were set at the freestream inflow boundary, while the static pressure was extrapolated. At the outflow boundary all flow quantities were extrapolated, except for the static pressure, which was prescribed. A stable and realistic nozzle-inflow condition was found by extrapolating the static pressure and prescribing = pinvinAin and the total temperature (the total tempthe mass flux hi, = hjet erature at the nozzle inlet was chosen to match the total temperature of the freestream). Inflow velocity vin and temperature, Ti, were then obtained by solving
and
y-1
Tm+-v;1
2
YR =Tin
y-1
1 2 + -vin 2
(4)
The wall was considered to be adiabatic and hydraulically smooth.
D. Results With the 1988 K--W model and the Menter SST model the wall jet stayed attached to the wall for too long (Fig. 15). Shown therefore are transient solutions for these models. On the other hand, very good results could be obtained when the EASM model was used. Case 321 was computed with the 1988 K--W model and EASM only (Fig. 16). For both cases the jet-exit velocity vjet x 6.7v,, resulting in a jet-exit Mach number Mjetx 0.85. The nozzle-pressure ratio (nozzle inflow to nozzle exit) was approximately 1.6 and the nozzle-density ratio was about 1.4. Wall-pressure distributions are shown in Fig. 17. For both cases the prediction is in very good agreement with the experiment. When the 1998 K--W model with EASM was used, the wall jet separated somewhat earlier, leading to a slightly smaller circulation augmentation and a slightly smaller area enclosed by the pressure coefficient curves. The LE stagnation point moved backward as a result of the increase in total circulation (Figs. 18 and 19).
TURBULENT COANDA WALL JETS AND DNS AND RANS
417
1988K-0
Menter SST
1988~-0 EASM
1988~-0 EASM
Fig. 15 RANS calculation of CC airfoil, Case 283 (a= 0 deg). Eddy viscosity normalized by laminar viscosity is (left) and turbulence kinetic energy (right) (result for 1988 K--0 and Menter SST model are transient).
1988~-0 EASM
Fig. 16 RANS calculation of CC airfoil, Case 321 (a= -8 deg). Eddy viscosity is normalized by laminar viscosity (left) and turbulence kinetic energy (right).
418
H. F. FASEL, A. GROSS, AND
S. WERNZ
b)
a)
Fig. 17 RANS calculation of CC airfoil showing the wall-pressure coefficient cp = 201 - pm)/(pjetvfet): a) Case 283 (a= 0 deg); b) Case 321 (a= -8 deg). a)
b)
c)
Fig. 18 RANS calculation of CC airfoil showing total velocity and streamlines: a) Case 283 (a= 0 deg) 1988 K-W, EASM; b) Case 283 (a= 0 deg) 1998 K-W, EASM; c) Case 321 (a= 8 deg) 1988 K-w EASM.
Fig. 19 RANS calculation of CC airfoil showing total velocity and streamlines (1988 K-w model with EASM): a) Case 283 (a= 0 deg); b) Case 321 (a= -8 deg).
VI. Conclusions Coanda wall jets for two different configurations were investigated numerically: 1) The circular cylinder from the experiments by Wygnanski and coworkers; and 2 ) the NCCR 1510-7067 N CC airfoil from the experiments by Abramson (the workshop CFD challenge).
TURBULENT COANDA WALL JETS AND DNS AND RANS
419
Configuration 1 was investigated using DNS and RANS computations. In the DNS, both spanwise and streamwise coherent structures were present in the flow. It was conjectured that in the natural, unforced case both types of structures keep each other at bay and that if either one was favored or forced by active flow control, the other one would be weakened. This conjecture was probed by separately forcing the spanwise and streamwise coherent structures at the nozzle inflow. Forcing of the spanwise structures indeed strengthened their downstream coherence, but did not noticeably weaken the streamwise structures. The reason for this is unclear and necessitates further research. Forcing of the streamwise structures weakened the spanwise structures and strengthened the streamwise structures, as expected. The downstream development and interaction of both types of structures and their influence on the turbulent flow are ultimately responsible for the downstream development of the wall jet. The goal here is to actively control the jet spreading and velocity decay by application of AFC at the nozzle exit. Configuration 1 was also used to evaluate turbulence models for steady RANS of Coanda wall jets. None of the models tested correctly predicted all relevant aspects of the flow. Evidently, important physical mechanisms are not modeled correctly. For example, none of the employed turbulence models had a curvature correction. Also, the strong turbulent coherent structures that are not captured in steady and two-dimensional RANS may significantly contribute to the mean flow and turbulence characteristics. Relatively speakening, the models based on an EASM Reynolds stress model performed best. Configuration 1 was also used for steady RANS stability investigations. Steady streamwise structures were introduced at the nozzle, and their development in the downstream direction was investigated. At low disturbance amplitudes (linear case), the local size of the dominant streamwise structures roughly scales with the local wall jet halfthickness, an observation that was also made in the experiment. Overall, the amplification of the streamwise coherent structures by the centrifugal Gortler instability was rather small. If the streamwise coherent structures observed in the experiment were of similar strength as in the linear three-dimensional RANS computation, the vortex mergings observed in the experiment may be explainable by linear stability mechanisms. Based on the experience gained from studying configuration 1, the elliptic CC airfoil (configuration 2) was then computed using the RANS and by employing the 1988 and 1998 K-6.1 models and the Menter SST model. In our calculations, only use of the EASM Reynolds stress model resulted in good predictions of the wall jet separation from the airfoil. For both angles of attack, excellent agreement with the experimental data could be obtained with this model.
Acknowledgments The authors gratefully acknowledge the Office of Naval Research for funding of this research under grant number N00014-01-1-09, with Ronald J o s h serving as program manager. References ‘Tani, I., “Production of Longitudinal Vortices in the Boundary Layer Along a Concave Wall,” Journal of Geophysical Research, Vol. 67, No. 8, 1962, pp. 3075-3080.
420
H. F. FASEL, A. GROSS, AND
S. WERNZ
’Moser, R. D., and Moin, P., “The Effects of Curvature in Wall-Bounded Turbulent Flows,” Journal of Fluid Mechanics, Vol. 175, 1987, pp. 479-510. 3Pajayakrit, P., and Kind, R. J., “Assessment and Modification of Two-Equation Turbulence Models,” AIAA Journal, Vol. 38, No. 6, 2000, pp. 955-963. 4Neuendorf, R., and Wygnanski, I., “On a Turbulent Wall Jet Flowing Over a Circular Cylinder,” Journal of Fluid Mechanics, Vol. 381, 1999, pp. 1-25. ’Wernz, S., Valsecchi, P., Gross, A., and Fasel, H. F., “Numerical Investigation of Turbulent Wall Jets Over a Convex Surface,” AIAA Paper 2003-3727, June 2003. 6Gross, A., Wernz, S., and Fasel, H. F., “Numerical Investigation of Coherent Structures in a Turbulent Coanda Wall Jet,” AIAA Paper 2003-4020, June 2003. 7Jones, G., and Joslin, R. D. (eds.), Proceedings of the 2004 NASAIONR Circulation Control Workshop, NASA/CP 2005-213509, June 2005. ‘Abramson, J., “Two-Dimensional Subsonic Wind Tunnel Evaluation of Two Related Cambered 15-Percent-Thick Circulation Control Airfoils,” DTNSRDC Tech. Rept. ASED-373, Sept. 1977. ’Slomski, J. F., Gorski, J. J., Miller, R. W., and Marino, T. A., “Numerical Simulation of Circulation Control Airfoils as Affected by Different Turbulence Models,” AIAA Paper 2002-0851, Jan. 2002. “Paterson, E. G., and Baker, W., “Simulation of Steady Circulation Control for Marinevehicle Control Surfaces,” AIAA Paper 2004-0748, Jan. 2004. “Menter, F. R., “2-Equation Eddy-Viscosity Turbulence Models for Engineering Applications,” AIAA Journal, Vol. 32, No. 8, 1994, pp. 1598-1605. ”Meitz, H. L., “Numerical Investigation of Suction in a Transitional Flat-Plate Boundary Layer,” Ph.D. Dissertation, Dept. of Aerospace and Mechanical Engineering, Univ. of Arizona, Tucson, AZ, 1996. 13Meitz,H. L., and Fasel, H. F., “A Compact-Difference Scheme for the Navier-Stokes Equations in Vorticity -Velocity Formulation,” Journal of Computational Physics, V O ~157, . NO. 1, 2000, pp. 371 -403. ‘‘Gross, A., and Fasel, H., “High-Order WEN0 Schemes Based on the Roe Approximate Riemann Solver,” AIAA Paper 2002-2735, June 2002. ‘’Wilcox, D. C., Turbulence Modeling for CFD, 2nd ed., DCW Industries, La Canada, CA, 2000. 16Rumsey, C. L., and Gatski, T. B. “Recent Turbulence Model Advances Applied to Multielement Airfoil Computations” Journal of Aircraft, Vol. 38, No. 5, 2001, pp. 904-910. 17Spalart,P. R., and Allmaras, S. R., “A One-Equation Turbulence Model for Aerodynamic Flows,” AIAA Paper 92-0439, Jan. 1992. “Fasel, H. F., Seidel, J., and Wernz, S., “A Methodology for Simulations of Complex Turbulent Flows,” Trans. ASME, Journal of Fluid Engineering, Vol. 124, 2002, pp. 933-942. ‘’Fasel, H. F., von Terzi, D. A., and Sandberg, R. D., “A Methodology for Simulating Compressible Turbulent Flows,” FEDSM 2003-45334, 4th ASMEIJSME Joint Fluids Engineering Conference, July 2003; also ASME Journal of Applied Mechanics (to be published).
Chapter 16
RANS and Detached-Eddy Simulation of the NCCR Airfoil Eric G. Paterson* and Warren J. Bakert Pennsylvania State University, University Park, Pennsylvania
Nomenclature a = speed of sound, ft/s CD = section drag coefficient, F ~ / ( 1 / 2 ) p U i S CL = section lift coefficient, F ~ / ( 1 / 2 ) p U i S C , = section moment coefficient, M z / ( 1/2)pUiSc C, = pressure coefficient, ( p - p o o ) / ( 1 / 2 ) p ~ L C , = j e t momentum coefficient, r i z ~ j / ( l / 2 ) p ~ L ~ c = foil chord length, in. FD = drag force, lbf FL = lift force, lbf f* = nondimensional frequency, f c / U , g = gravitational acceleration, ft/s2 h = slot height, in. k = turbulent kinetic energy, ft/s2 C, = k-w, or subgrid, length scale, in. C = DES length scale, in. M = Mach number, U / a M , = moment about the z-axis, ftelbf m = mass flow rate, pUjhw, lbm/s p = pressure, lbf/ft2
*Senior Research Associate, Applied Research Laboratory and Associate Professor of Mechanical and Nuclear Engineering. AIAA member. 'Graduate Research Assistant, Department of Aerospace Engineering. Member AIAA. Copyright 02005 by Eric G. Paterson and Warren J. Baker. Published by the American Institute of Aeronautics and Astronautics, Inc., with permission.
421
422
E. C. PATERSON AND W. J. BAKER
Re = Reynolds number, p U w c / p s = planform area, cw, ft2 U , V , W = velocity components, ft/s U , = friction velocity, ft/s w = foil span, in. x, y , z = Cartesian coordinate y+ = wall coordinate, U , j / v p = distance from wall, in. a = angle of attack, deg A = maximum dimension of local grid cell At* = nondimensional time step, AtU,/c S,, ,a 6, = dimensions of local grid cell in each curvilinear coordinate direction p = dynamic viscosity, lbm/ft.s 8, 7,t = curvilinear coordinates p = density, lbm/ft3 u = DES blending function or cavitation number T~ = wall-shear stress, lbf/ft2 w = turbulent dissipation rate, ft2/s3
m,
Subscripts = freestream j = at jet orifice
00
min = minimum Superscripts r = resolved turbulence s = subgrid turbulence tot = total
I. Introduction IRCULATION control (CC) for lift augmentation via the Coanda effect has been studied for many years.192In comparison to mechanical means of CC (e.g., shape change and leading- and trailing-edge flaps), the use of a wall jet on a convex curved trailing-edge (TE) surface is attractive for many reasons. Based upon aerospace flow-control applications3 and previous hydrodynamic assessment^,^'^ anticipated benefits for naval vehicles include simplification of actuation, reduction in weight and number of parts, dual-mode operation (i.e., cruise and high-lift scenarios), contribution to novel design options such as placing control surfaces at nontraditional locations and arrangement of sensors and payloads on control surfaces, and improved shock resistance. As with all flow control scheme^,^-^ there are technical as well as economic and operational issues that must be overcome for systems to be transitioned into practical application. For example, for CC schemes to be incorporated in the marine environment, they must address the inherent drag penalty of a blunt TE at cruise condition, overcome operator reluctance to fixed control surfaces, not
C
DETACHED-EDDY SIMULATION OF NCCR AIRFOIL
423
suffer from orifice fouling or shock damage, and, for applications where stealth is important, have limited impact on the hydroacoustic ~ignature.~ The work presented herein is ultimately motivated by these issues. Continued development of new actuation methods' potentially leads to novel solution of issues. Actuators such as high-performance solenoid valves, smart materials, zero-net-mass actuators, synthetic jet actuators, and plasma control actuators find application to CC as well as other forms of flow control. Of particular interest to CC are high-performance solenoid valve^,^ which can achieve efficient pulsed blowing, a mode of CC that has been known to reduce mass-flow requirements for a given performance increment."-12 However, detailed understanding of both the unsteady flow physics and their application in water-based scenarios is lacking. Even for steady blowing CC, there are important flow physics that computational fluid dynamics (CFD) models must be able to simulate if such tools are to be used in design. Most notable are streamwise curvature effects on the turbulent boundary layer and spanwise coherence of the wall jet. Nearly the entire range of Reynolds-averaged Navier- Stokes (RANS) turbulence models from algebraic to full Reynolds-stress transport models (RSTMs) have been modified for curvature effects.13-15 Unfortunately, the state of affairs is poor in that modifications to algebraic and one- and two-equation models are limited in range due to empiricism, whereas RSTMs have yet to convincingly demonstrate capability to resolve subtleties in the way curvature impacts mean flow and turbulence struct~re.'~ Nonetheless, numerical experiments for a CC configuration16 have demonstrated that baseline RSTMs can improve simulation results in comparison with baseline two-equation models, especially at large jet momentum coefficients. Moreover, this study showed that simulations using two-equation models demonstrated nonphysical behavior with a dramatic reduction in lift and a wall jet that remained attached to the surface for 1.5 revolutions around the foil.16 Unfortunately, the source (e.g., model limitations or numerical accuracy) of this discrepancy, and whether or not it is flow-code-specific, was not identified. Detailed understanding of the high-Reynolds-number turbulent wall jet on the Coanda surface would best be facilitated by direct numerical simulation (DNS), or possibly large eddy simulation (LES). For the usual reasons, that is, lack of computer power, this is not yet realizable. Therefore, the approach pursued here is one based upon the detached-eddy simulation (DES),17 which is a hybrid RANS/LES method. In this approach, the foil fore body and the nearwall region is treated as RANS and the outer regions of the after body boundary layer and near wake are treated as LES. Detached-eddy simulation has been shown to improve accuracy for massively separated and has been applied to an active flow control application with zero-net-mass actuation,20 albeit with inconclusive results. The ability of DES to resolve curvature effects, or the need for curvature modifications in the RANS portion of the DES model, is unknown. Although the objective of our research is to develop validated simulation tools using recently acquired incompressible water-tunnel data for a low-aspect-ratio tapered control surface21and wind-tunnel data for a pulsed CC c ~ n f i g u r a t i o n , ~ ~ ' ~ the work presented herein represents our initial efforts to apply RANS and DES to a simpler steady-blowing CC configuration.22 It has been selected as a preliminary validation exercise because of the fact that it can be treated as a
424
E. C. PATERSON AND W. J. BAKER
two-dimensional geometry and has previously been studied using RANS CFD.16 Our progress is reported in Secs. 11-VII. 11. Geometry, Conditions, and Data
The NCCR-1510-7067N CC foil was tested in a wind tunnel at the David Taylor Naval Ship Research and Development Center in 1977.22The geometry was a 15% thick elliptical cambered foil with a single jet orifice on the upper surface at x/c = 0.967. The model chord length was c = 8 in., the slot heightto-chord ratio h / c = 0.003, and the Coanda surface a nominal circular arc. A cross-section of the model is shown in Fig. 1. Although a wide range of C, and a were studied in the original experiment, two cases are studied here. For the first, designated as Case 283, C, = 0.209 and a = Odeg. For the second, designated as Case 321, C, = 0.184 and a = - 8 deg. Both are assumed to have the following common parameters: freestream velocity U , = 128.54 fps, freestream density p, = 0.07654 lbm/ft3, and kinematic viscosity p = 3.73 x lo-’ slug/ft-s. This yields a Reynolds number of Re = 5.45 x lo5 and a Mach number of M , = 0.12. Assuming that the jet is incompressible (i.e., pj/pm = l), the nondimensional jet-orifice velocities can be computed as
&;:;
vj/Um = - l - C ,
= 5.90 and 5.54
for Cases 283 and 321, respectively. Although this assumption introduces an unknown modeling error, a posteriori evidence suggests that it is small. Available experimental data are somewhat limited in comparison to modem experiments, consisting of surface pressure measured via pressure taps placed at midspan. Experimental lift and moment were computed by integrating the surface pressure, and drag was evaluated using a wake survey and a momentumdeficit method. In addition, estimates of experimental uncertainty are not available; however, several possible sources have been identified such as slot-height
TRAILING EDGE
Fig. 1 Cross-sectional geometry of NCCR 1510-7067N.
DETACHED-EDDY SIMULATION OF NCCR AIRFOIL
425
growth because of plenum pressure, and Coanda jet interaction with tunnel walls, especially at large C,, such that the effective a is different from the geometric a. 111. Computational Methods
A. Unsteady RANS CFDSHIP-IOWA23 is a general-purpose parallel unsteady incompressible RANS CFD code. The computational approach is based upon the pressureimplicit split-operator (PISO) approach, which iteratively solves the momentum and pressure-Poisson equations. Discretization is achieved using structured overset grids and the finite-difference method, where convective terms are discretized using a general five-point stencil that permits a user-specified orderof-accuracy ranging from first-order upwind to fourth-order central. Viscous and temporal terms are discretized using second-order central and secondorder backward methods, respectively. Turbulence is modeled using a linear closure and the blended K- W / K - - E SST two-equation Efficient parallel computing is achieved using coarse-grain parallelism via MPI distributed computing. For time-accurate unsteady simulations, global solution of the pressurePoisson equation is achieved using preconditioned GMRES and the PETSc libraries.25926
B. Detached-Eddy Simulation Detached-eddy simulation is a three-dimensional unsteady numerical method using a single turbulence model, which functions as a subgrid-scale model in regions where the grid density is fine enough for LES, and as a RANS model in all other regions. Implementation of DES in CFDSHIP-IOWA was accomplished by modifying the turbulence model and convective-term discretization. The turbulence model is modified by introducing a DES length scale
t = min ( e k w , C D E S h )
(1)
which compares the subgrid length scale to the local grid size, where the former can be written as
CDEsis a model constant with a value between 0.78 and 0.61 weighted by the Menter k-w1k-E blending function,24 and A is based on the largest dimension of the local grid cell:
A = max (&, a, 8,) The new length scale equation
(3)
t replaces t k w in the destruction term of the k-transport Dk,,,
pk3I2
= pp*kw = ekw
(4)
426
E. C. PATERSON AND W. J. BAKER
which results in a new destruction term:
The effect of this modification on the turbulence budget is to shift energy from subgrid, or modeled, scales to resolved scales as defined by the filter width CDESA. The second modification aims to reduce numerical dissipation inherent in the upwind convective-term discretization scheme. The implemented approach is based upon a hybrid central/upwind approximation of the convective terms (or fluxes):
where u is defined as
(7) The result is that u smoothly transitions between 1.0 in the RANS regions, resulting in an “almost upwind” scheme, and 0.0 in the LES regions, resulting in an “almost centered” scheme. In addition, a Courant-number constraint of 1.0 has been imposed, which requires that the time step be sufficiently small to support turbulent eddies. The coefficients n and m permit the interface between RANS and LES regions to be arbitrarily “sharpened”; however, currently we use n = m = 1 because of the fact that higher-order coefficients have resulted in unstable simulations. In CFDSHIP-IOWA the convective terms are discretized with the following higher-order upwind formula
where
DES implementation is accomplished by redefining these equations
DETACHED-EDDY SIMULATION OF NCCR AIRFOIL
427
where W,, W,, W,, W,, and W,, are hybrid coefficients defined as the blending between second-order upwind and fourth-order central schemes:
w,,= (1 - a ) w E + ow: w,= (1 - a ) w Z + mv? w,= (1 - a ) w F + mv? w, = (1 - a)wtlf + w,, = (1 - a)wtlf, + ow:,
(13)
Finally, as discussed in the following section, it is noted that CFDSHIP-IOWA is an overset-grid capable CFD code with an interface to PEGASUS 5.1.27 This capability will be exploited to perform local grid refinement and flow adaptation in the wall-jet, wake, and LES regions.
IV. Grid Generation Overset grids are generated primarily using hyperbolic extrusion, although transfinite interpolation and elliptic smoothing is used for blocks that do not lend themselves to that approach, that is, the background mesh and plenum mesh. Overset interpolation coefficients and holes are computed using Pegasus 5.1.27 CFDSHIP-IOWA employs double-fringe outer and hole boundaries so that the five-point discretization stencil (i.e., in each curvilinear coordinate direction) and order-of-accuracy does not have to be reduced near overset boundaries. Level-2 interpolation capability of PEGASUS 5.1 is also used so as to achieve optimal match between donor and interpolant meshes. Grid design is based upon a domain size of - 2 5 x/c 5 4, - 2 5 y / c 5 2, and 0I z/c I 0.2, and a near-wall spacing of 1.0 x the latter of which aims to resolve the sublayer of the turbulent boundary layer with a wall spacing of y+= 1. The grid system used for RANS simulations is shown in Fig. 2. Nested orthogonal uniform box grids are used for the far-field and a simple 0-grid is used for the foil. Preliminary solutions were used to locate streamlines, and wake-refinement blocks were built off these streamlines for subsequent higher-fidelity simulations. RANS simulations were computed in a pseudo-two-dimensional fashion that requires five points in the spanwise direction. The entire grid system consists of 323,000 points and comprises eight blocks ranging in size from 30,000 to 5 1,000 points. For DES, the approach is the same as described above, except that the spanwise resolution must be increased in regions where turbulent eddies are to be resolved. Overset grids are effectively used to locally refine the simulation. As shown in Fig. 3, the fore body and far-field, which is in the RANS region, is resolved with five points in the spanwise direction. In contrast, the TE and near-wake blocks are resolved with 41 points in the spanwise direction. The wake refinement mesh shown in Fig. 3 is designed for unblown C= , 0
E. C. PATERSON AND W. J. BAKER
428 a)
Fig. 2 Overset grid system for RANS simulation: a) Overall view; b) foil view; c) plenum and TE view.
simulations and has an isotropic spacing of A = 0.005. The entire grid system consists of 855,000 points and comprises 15 blocks ranging in size from 31,000 to 67,000 points. It is noted that translational periodicity is imposed in the spanwise direction and that the extent of the domain in this direction is
DETACHED-EDDY SIMULATION OF NCCR AIRFOIL
429
a)
Fig. 3 Overset grid system for DES: a) Side view; b) TE detail, showing spanwise resolution.
20% of chord length. It is acknowledged that this can potentially affect flow structures in the wake, particularly if this dimension is smaller than the spanwise turbulent correlation len th scales, which for this problem are unknown. However, other TE flows55 have shown correlation lengths of four foil thicknesses, which in this case would be 0.60~.Because periodic boundaries are used, domain size in the spanwise direction would need to be 1 . 2 ~ As . such, our domain may be one-sixth the required size. V. Initial and Boundary Conditions Initial conditions for the steady RANS simulations are straightforward: U = U,, V = 0, k = k, = 1 x lop7, w = w, = 9.0, and p = 0. For unsteady
430
E. C. PATERSON AND W. J. BAKER
RANS and DES, a cubic polynomial is used to accelerate the foil from rest over a nondimensional time of 2.0. No-slip boundary conditions were applied on all surfaces of the foil and the top and bottom walls of the plenum. On the inlet face of the plenum, a top-hat velocity profile was prescribed with the magnitude computed using conservation of mass, known Uj/U, at the jet orifice, and a plenum contraction of 10.63. For Cases 283 and 321, this velocity magnitude corresponds to 0.555 and 0.521, respectively. In addition, it was assumed that the inflow at this location was laminar. Inlet, far-field, and exit conditions were applied on the outer boundaries of the largest box grid and translational periodicity was applied on all spanwise faces. Neumann conditions were used for pressure on all boundaries. As already mentioned, outer and hole boundary trilinear interpolation coefficients were computed using Pegasus 5.1 Mathematical formulation of all boundary conditions are described in the CFDSHIP-IOWA users’ manual.21 Finally, it is noted that boundary conditions are set and input file created using the CFDSHIP-IOWA filter in the GRIDGEN software from Pointwise, Inc.
.*’
VI. Results Research has been undertaken along two paths, both of which are presented. First, RANS simulations for Cases 283 and 321 will be presented. Second, DES results for the unblown C, = 0 case will be shown and discussed.
Steady RANS Simulation A comparison of experimental and simulated surface pressure is shown in Fig. 4. Relatively good agreement is demonstrated for both cases. The largest discrepancy is the underprediction of the suction peak aft of the jet orifice. Case 283 shows a strong LE low pressure, relatively uniform loading over the majority of the chord, and a Cp,min of - 17 and - 18 at the TE for the simulation and experiment, respectively. Because of the negative angle of attack, Case 321 lacks the LE low pressure. It also shows larger error in comparison to the data across the chord, but especially on the Coanda surface. The predicted and experimental Cp,minare - 13 and - 18, respectively. Lift, drag, and moment about the z-axis centered at midchord were computed by integrating C, and T~ on all external surfaces. All plenum surfaces were neglected in the CFD integration process so that comparison could be made to experimental values. Experimental values were computedz2by directly integrating the discrete (and fairly coarse) surface-pressure data. Results are tabulated in Table 1. The lift coefficient for Case 283 is within 5% of the data, whereas Case 321 shows a discrepancy of 30% because of the larger underprediction of the suction peak on the Coanda surface. Drag for both cases shows a very large difference from the data. The data, which were measured using a wake profile corrected by the jet momentum, show a negative drag, whereas the CFD values (which includes both viscous and pressure components) are positive and substantially larger in magnitude. Moment coefficient CM is positive (LE down, TE up) for both cases because of the large suction peak on the Coanda surface. Data for CMare not available. A.
DETACHED-EDDY SIMULATION OF NCCR AIRFOIL
43 1
a)
Fig. 4 Comparison of experimental (symbols) and computational (lines) surface pressure: a) Case 283, C, = 0.209, a = 0 deg; b) Case 321 C, = 0.184, a = -8 deg.
Figure 5 illustrates the impact of the Coanda effect upon the overall circulation. For both cases, velocity-magnitude contours show a high velocity on the top surface that is consistent with the surface pressure shown in Fig. 4. The streamlines show the effect of the change in angle of attack on the overall flowfield and on the locations of stagnation points. Table 1 Lift, drag, and moment coefficients
CL
Case 283 Case 321
CM
CD
Data
CFD
Data
CFD
Data
CFD
4.2 3.1
4.0 2.4
- 0.05
0.18 0.12
-
2.07 1.21
CFD,computational fluid dynamics.
- 0.08
-
E. C. PATERSON AND W. J. BAKER
432 a)
Fig. 5 Overall view of velocity-magnitude contours and streamlines: a) Case 283, C , = 0.209, (Y = 0 deg; b) Case 321, C , = 0.184, (Y = -8 deg.
A close-up view of the LE flowfield is shown in Fig. 6 . For Case 283, the stagnation point is located at x / c = 0.07 on the foil bottom surface. For Case 321, the stagnation point is located at x / c = 0.01 on the foil bottom surface, despite the negative angle of attack. Comparing the two cases, the relative magnitudes of velocity are shown to be consistent with the differences in the LE suction peak shown in Fig. 4. A close-up of the TE flowfield is shown in Fig. 7. Both cases are similar in that they demonstrate a high-velocity jet emanating from the plenum, increased velocity around the initial curvature of the Coanda surface, clean jet detachment, and flow separation on the bottom surface upstream of the jet. For Case 283, the wall jet stays attached longer and the thickness of the separated region is smaller in comparison to Case 321. Contours of turbulent kinetic energy near the TE are shown in Fig. 8. Again, both cases demonstrate similar behavior. The contours show two primary sources
DETACHED-EDDY SIMULATION OF NCCR AIRFOIL
433
a)
Fig. 6 Leading-edge view of velocity magnitude contours and streamlines: a) Case 283, C, = 0.209, a = 0 deg; b) Case 321, C, = 0.184, a = -8 deg.
of kinetic energy, both of which correspond to regions of high mean shear. The first is downstream of the jet-slot knife edge and grows along the wall-jet shear layer. The second, which is larger in magnitude, starts at the point of wall jet separation and grows into the wake. It is noted that the maximum k is approximately 0.7, which is two orders-of-magnitude larger than k in the turbulent boundary layer. To better understand the evolution of the wall jet, profiles of velocity magnitude and turbulent kinetic energy are extracted at two locations for Case 283, as shown in Fig. 9. Location A is slightly aft of the jet orifice, and location B is along a y = 0 line. At location A, the wall jet and its correspondingly strong shear layer are clearly shown. The turbulent kinetic energy shows spikes downstream of the plenum walls, the outer of which merges with k from the suction-side boundary layer. At location B, the peak velocity magnitude is close to that at location A; however, the wall-jet shape has greatly thickened as a result of viscous and turbulent stresses near the wall and along the shear layer. The turbulent kinetic
E. C. PATERSON AND W. J. BAKER
434 a)
Fig. 7 Trailing-edge view of velocity magnitude contours and streamlines: a) Case 283, C, = 0.209, a = 0 deg; b) Case 321, C, = 0.184, a = - 8 deg.
energy has significantly grown in both magnitude and thickness, both of which are consistent with velocity profiles and k contours shown in Fig. 8. In preparation for future DES of the blown cases, the length scale in Eq. (2) was computed for Case 283 and is shown in Fig. 10. This shows that the largest eddies in the boundary layer and near wake are of the order of 0.02~. However, the length scale is much smaller (i.e., &, 5 0.002) in the near orifice region. Therefore, target grid spacing in this area should be approximately A = 0.001, which is five times finer than the grid used in the unblown simulations discussed in the next section.
B. Detached-Eddy Simulation Detatched-eddy simulation (DES) was performed for 10,000 time steps with At = 0.001 (or 10 flow-through periods). Animations of the instantaneous isosurface of vorticity shaded by spanwise velocity were made and snapshots are
DETACHED-EDDY SIMULATION OF NCCR AIRFOIL
435
Fig. 8 Contours of turbulent kinetic energy: a) Case 283, C , = 0.209, a = 0 deg; b) Case 321, C , = 0.184, a = - 8 deg.
shown in Fig. 11. The side view clearly shows the dominant vortex shedding of spanwise eddies. The overset grid is also shown in the background to illustrate the effect of switching from high to low, that is, LES-to-RANS, grid resolution in the near wake (i.e, at about 0 . 4 ~downstream of the TE). All spanwise structure is filtered and only the “two-dimensional” vortex passes through this interface. The top view clearly displays the longitudinal vortices, which are intertwined with the spanwise vortices. Again, the impact of switching from high to low grid resolution is shown. The lack of spurious numerical reflections at this overset boundary is noted. Mean and root-mean-square (RMS) statistics for all dependent variables were computed over 6000 time steps. Figure 12 shows the contours of the mean axial velocity streamlines through the mean field, and RMS axial velocity The mean flowfield shows a typical wake with two eddies. The RMS velocity field also shows a typical wake pattern28’29with two peaks across the wake corresponding to vortices shed off the top and bottom sides of the foil. It is noted
u,
fi.
E. C. PATERSON AND W. J. BAKER
436 a) 0.014 0.012
.
h
0
2
0.01
I
1 0
0.006 0.006
0.004 0.002
00
1
3
5
6
Vel&ity magnitude (U2+4/2)1’2
6
Fig. 9 Extracted profiles: a) Velocity magnitude; b) turbulent kinetic energy.
that computed statistics were not yet fully two-dimensional, thus indicating that a larger integration time is needed to reduce uncertainty in the computed statistics. Analysis of the turbulent kinetic energy is shown in Fig. 13. Subgrid turbulence kS is computed from the modified k - w turbulence model, whereas the resolved turbulence is computed from the velocity correlations k‘ = $(El W WW). Total kinetic energy is the sum of these two parts. These figures show that kS is significant only in the boundary layer upstream of the separation. Downstream, total k is comprised of resolvable scales only. A region of particular interest is the potential “gray region” where the solution switches from RANS to LES,and where the model’s response to the underlying grid does not yield either a fully LES or a fully RANS solution. Contours of total k show a
+ +
DETACHED-EDDY SIMULATION OF NCCR AIRFOIL
437
Fig. 10 k-w length scale for Case 283.
Fig. 11 Instantaneous iso-surface of vorticity shaded by spanwise velocity component: a) Side view; b) top view.
E. C. PATERSON AND W. J. BAKER
438 a)
Fig. 12 Statistical analysis of axial velocity: a) Mean velocity; b) root-mean-square velocity.
slight decrease in magnitude as the TE is approached, and highlights a deficiency in the overall approach, which is consistent with other recent high Re TE DES
sir nu la ti on^.^^
Finally, Fig. 14 shows spectral analysis of velocity at a single point ( x / c , y / c ) = (1.117, 0.016), the location of which was shown in Fig. 12. The time history and Fourier transform show a shedding frequency at f$ =f c / U , = 3.8. If a new length scale is defined as the vertical distance between points of mean separation at the TE, which is d / c = 0.052, a more appropriate shedding frequency is computed to be f =f d / U , = 0.198, which is consistent with a typical Strouhal number of 0.2. The Fourier transform shows higher harmonics at ff = 7.5 and f; = 12, which are 2f$ and 3f$, respectively, and a decay of the higher frequencies at - 5/3 slope up to a frequency of about 30, the latter of which is consistent with a grid spacing of 0.005 and the assumption of 10 grid points per wavelength.
DETACHED-EDDY SIMULATION OF NCCR AIRFOIL
439
a)
Fig. 13 Analysis of turbulent kinetic energy: a) Subgrid, kS; b) resolved, k'; c) total, kS k'.
+
E. C. PATERSON AND W. J. BAKER
440
0.75
3 0.5
0.25
time (UVC)
b)
Powerspectral density of axial velocity
Fig. 14 Frequency analysis of axial velocity at (x/c, y / c ) = (1.117, 0.016): a) Time history; b) Fourier transform.
C. Cavitation-Free Operating Depth and Speed Given the low pressure on the Coanda surface, cavitation is a concern for ship hydrodynamics. As a rough estimate, cavitation occurs when the magnitude of minimum pressure coefficient exceeds the cavitation number:
-cp 2
(T
DETACHED-EDDY SIMULATION OF NCCR AIRFOIL
44 1
Given that p m = pgz, u increases linearly with depth. Substituting p m into Eq. (13, an expression for cavitation-free operation relating Cp,,,in, depth z, and vehicle speed U , can be derived:
Using properties of water at 15°C ( p = 1000 kg/m 3 , pv = 1.7 Wa), a family of curves can be computed that relates the three variables. Such a figure is shown in Fig. 15. It illustrates, for example, that for a Cp,fin= -20, cavitation can be avoided at all depths greater than 50 ft as long as speed remains lower than 1Okn. Because CC is envisioned for low-speed littoral operation where traditional control surfaces lose control authority, this is a favorable observation. On the other hand, a speed of 30 kn would require a depth of 750 ft to achieve cavitation-free operation, at least for the C, studied herein. Fortunately, because dynamic pressure increases with Urn,lower C, and CL,and therefore decreased Cp,min, would be required at high speed, thus permitting CC to be used throughout the operation envelope.
VII. Conclusions A CC foil was studied using incompressible RANS and DES CFD methods. RANS simulations of large jet-momentum coefficient cases demonstrated that a linear closure with blended k- W / k - - E turbulence model was able successfully to predict the pressure distribution trends in comparison to benchmark data. This 30
25
20
10
5
OO
10
20
40
50
Fig. 15 Cavitation-free operation curves.
60
442
E. C. PATERSON AND W. J. BAKER
contrasts with other published results,16 which indicate the need for higher-order curvature-corrected models such as a full Reynolds-stress model. The reason for this discrepancy is unknown, but recent work by Baker and Paterson3’ indicates that near-wall grid resolution on the Coanda surface plays an important role when using two-equation turbulence models. Details of the simulated flow were presented through analysis of the integral forces and moment, velocity field, and turbulent kinetic energy. Detached-eddy simulation was undertaken for the unblown case, and demonstrated that the method is capable of resolving turbulent vortex shedding. Statistical and spectral analysis was undertaken to explain the simulation results; however, as with the RANS simulations, lack of data precludes validation for this problem. Nonetheless, results are encouraging and suggest further application of DES to both CC studies as well as other TE applications (e.g., propulsor blades and nozzles). Future work will focus on validation using modern water-tunnel data for a low-aspect-ratio ta ered control surface*l and wind-tunnel data for a pulsed CC config~ration.~’ P In addition to providing high-fidelity flowfield data, these cases will permit study of three-dimensional effects and pulsed blowing, both of which are important issues for practical application and improved understanding of basic CC flow physics.
Acknowledgments The authors gratefully acknowledge support from both the Office of Naval Research through Grant Number N00014-03-1-0122 (Program Officer: Ron Joslin) and NAVSEA SUB-RT (Program Manager: Meg Stout), the latter of which was in the form of a graduate student fellowship for the second author. The DoD High Performance Computing Modernization Office (HPCMO) and Army Research Laboratory-Major Shared Resource Center are acknowledged for providing computing resources through DoD HPCMO Challenge Project Number C1E.
References ‘Englar, R., “Circulation Control Pneumatic Aerodynamics: Blown Force and Moment Augmentation and Modifications; Past, Present, and Future,” AIAA Paper 2000-2541, June 2000. *Wood, N., and Nielson, J., “Circulation Control Airfoils Past, Present, and Future,” AIAA Paper 1985-0204, Jan. 1985. 3McLean, J. D., Crouch, J. D., Stoner, R. C., Sakurai, S., Seidel, G. E., Feifel, W. M., and Rush, H. M., “Study of the Application of Separation Control by Unsteady Excitation to Civil Transport Aircraft,” NASA/CR-1999-209338, June 1999. 4Joslin, R., Kunz, R., and Stinebring, D., “Flow Control Technology Readiness: Aerodynamic versus Hydrodynamic,” AIAA Paper 2000-44 12, June 2000. 5Hess, D., and Fu, T., “Impact of Flow Control Technologies on Naval Platforms,” AIAA Paper 2003-3568, June 2003.
DETACHED-EDDY SIMULATION OF NCCR AIRFOIL
443
6Bushnell, D., “Application Frontiers of ‘Designer Fluid Mechanics’ Visions versus Reality or An Attempt to Answer the Perennial Question ‘Why Isn’t It Used?’,’’ AIAA Paper 1997-2110, June 1997. ’Howe, M., “Noise Generated by a Coanda Wall Jet Circulation Control Device,” Journal of Sound and Vibration, Vol. 249, No. 4, 2002, pp. 679-700. ‘Schaffler, N., Hepner, T., Jones, G.,and Kegerise, M., “Overview of Active Flow Control Actuator Development at NASA Langley Research Center,” AIAA Paper 20023159, June 2002. ’Jones, G.,Viken, S., Washburn, A., Jenmins, L., and Cagle, C., “An Active Flow Circulation Controlled Flap Concept for General Aviation Aircraft Applications,” AIAA Paper 2002-3157, June 2002. “Oyler, T., and Palmer, W., “Exploratory Investigation of Pulse Blowing for Boundary Layer Control,” Tech. Rept. NR72H-12, North American Rockwell, Jan. 1972. "Waiters, R., Myer, D., and Holt, D., “Circulation Control by Steady and Pulsed Blowing for a Cambered Elliptical Airfoil,” Aerospace Engineering TR-32, West Virginia Univ., Morgantown, WV, July 1972. 12Jones, G.,and Englar, R., “Advances in Pneumatic-Controlled High-Lift Systems Through Pulsed Blowing,” AIAA Paper 2003-341 1, June 2003. ‘3Wallin, S., and Johansson, A., “Modeling Streamline Curvature Effects in Explicit Algebraic Reynolds Stress Turbulence Models,” International Journal of Heat and Fluid Flow, Vol. 23, 2002, pp. 721-730. 14Patel,V., and Sotiropoulos, F., “Longitudinal Curvature Effects in Turbulent Boundary Layers,” Progress in Aerospace Science, Vol. 33, 1997, pp. 1-70. ”Gatski, T., and Speziale, C., “On Explicit Algebraic Stress Models for Complex Turbulent Flows,” Journal of Fluid Mechanics, Vol. 254, 1993, pp. 59-78. ‘6Slomski, J., Gorski, J., Miller, R., and Marino, T., “Numerical Simulation of Circulation Control Airfoils as Affected by Different Turbulence Models,” AIAA Paper 20020851, Jan. 2002. ”Strelets, M., “Detached-Eddy Simulation of Massively Separated Flows,” AIAA Paper 2001-0879, Jan. 2001. “Squires, K., Forsythe, J., Morton, S., Strang, W., Wurtzler, K., Tomaro, R., Grismer, M., and Spalart, P., “Progress on Detached-Eddy Simulation of Massively Separated Flows,” AIAA Paper 2002-1021, Jan. 2002. ”Forsythe, J., Squires, K., Wurtzler, K., and Spalart, P., “Detached-Eddy Simulation of Fighter Aircraft at High Alpha,” AIAA Paper 2002-0591, Jan. 2002. 2oSpalart, P., Hedges, L., Shur, M., and Travin, A., “Simulation of Active Flow Control on a Stalled Airfoil,” Proceedings of IUTAM Symposium on Unsteady Separated Flows, Apr. 2002. ”Rogers, E., and Donnelly, M., “Characteristics of a Dual-Slotted Circulation Control Wing of Low Aspect Ratio Intended for Naval Hydrodynamic Applications,” AIAA Paper 2004- 1244, Jan. 2004. ”Abramson, J., “Two-Dimensional Subsonic Wind Tunnel Evaluation of Two Related Cambered 15-Percent Circulation Control Airfoils,” DTNSRDC ASED-373, Sept. 1977. 23Paterson, E., Wilson, R., and Stem, F., “General-Purpose Parallel Unsteady RANS Ship Hydrodynamics Code: CFDSHIP-IOWA,” Tech. Rept. 432, IIHR Hydroscience and Engineering, Univ. of Iowa, Ames, IA, Nov. 2003. 24Ames, I. A., and Menter, F., “Two-Equation Eddy Viscosity Turbulence Models for Engineering Applications,” AIAA Journal, Vol. 32, No. 8, 1994, pp. 1598- 1605.
444
E. C. PATERSON AND W. J. BAKER
25Balay, S., Buschelman, K., Gropp, W. D., Kaushik, D., Knepley, M., McInnes, L. C., Smith, B. F., and Zhang, H., “PETSc Users Manual,” Tech. Rept. ANL-95/11-Revision 2.1 S , Argonne National Lab., Jan. 2003. 26Balay, S., Gropp, W. D., McInnes, L. C., and Smith, B. F., “Efficient Management of Parallelism in Object Oriented Numerical Software Libraries,” Modern Software Tools in Scient$c Computing, edited by E. Arge, A. M. Bruaset, and H. P. Langtangen, Birkhauser Press, Cambridge, MA, 1997, pp. 163-202. 27S~hs,N. E., Rogers, S . E., Dietz, W. E., and Kwak, D., “PEGASUS 5: An Automated Pre-Processor for Overset-Grid CFD,” AIAA Paper 2002-0101, June 2002. ”Blake, W., “A Statistical Description of Pressure and Velocity Fields at the TrailingEdges of a Flat Strut,” DTNSRDC Rept. 4241, Dec. 1975. 29Paterson, E. G.,and Peltier, L. J., “Detached-Eddy Simulation of High-Reynolds Number Beveled-Trailing-Edge Boundary Layers and Wakes,” ASME Journal of Fluids Engineering, Vol. 127, 2005, pp. 897-906. 30Baker, W. J., and Paterson, E. G.,“Simulation of Steady Circulation Control for the GACC Wing,” Applications of Circulation Control Technologies, AIAA, Reston, VA, 2005.
Chapter 17
Full Reynolds-Stress Modeling of Circulation Control Airfoils Peter A. Chang III,* Joseph Slomski,* Thomas Marho,+Michael P. Ebert,+ and Jane Abramson* Naval Sur$ace War$are Center-Carderock Division, West Bethesda, Maryland
Nomenclature A = airfoil planform area, m2 c = chord length, m CL = lift coefficient; see Eq. (2) C, = pressure coefficient, see Eq. (3) C, = blowing rate; see Eq. (1) h = slot height, m k = turbulence kinetic energy, m2/s2 riZ = mass flow rate, kg/s Re = Reynolds number based on U,, c, and v, S = span, m U , = freestream velocity, m/s u, v = fluctuating horizontal and vertical velocity, respectively, m/s u, = friction velocity, m/s vj = mean jet velocity at slot opening, m/s x, y = in-plane coordinates, m y+ = wall normal distance in viscous units; yu,/vm a = angle of attack, rad E = turbulence dissipation rate, m2/s3 r ) = distance from wall, m w = specific dissipation rate, 1/s *Propulsion and Fluid Systems Department. Member AIAA. 'Propulsion and Fluid Systems Department. *Marine and Aviation Department (retired). Member AIAA. This material is declared a work of the U.S.Government and is not subject to copyright protection in the United States.
445
446
P. A. CHANG ET AL.
p.. = freestream fluid density, kg/m3 T~ = wall shear stress, kg/(m. s2) .v = free stream kinematic viscosity, m2/s V T = turbulence viscosity, m2/s
- -- (overbar) time average
I. Introduction ECENTLY, low-speed maneuverability has become an important design equirement for aircraft, ships, and submarines. At low speed, the control authority (that is, the normal, or lifting force) associated with conventional hinged control surfaces is often insufficient to perform certain maneuvers. As a result, designers have begun to investigate the use of circulation control (CC) airfoils to achieve the required control authority at low speeds. Circulation control technology has been investigated both experimentally”2 and a n a l y t i ~ a l l yover ~ , ~ the past 25 years. True CC airfoils typically have bluff trailing edges. These airfoils employ the Coanda effect to obtain lift augmentation by tangentially ejecting (blowing) a sheet of fluid near the trailing edge (TE) on the upper surface. Because of the Coanda effect, the jet sheet remains attached to the bluff TE and provides a mechanism for boundary layer control (BLC). The blowing can be thought of as a movement of the stagnation point, producing an increase in circulation around the airfoil. Experimental results for Coanda-type TE blowing5 have shown lift coefficient increases of as much as a factor of four when compared to the case of no blowing. Because of the difficulty and expense involved in experimentally investigating different CC configurations for parametric design studies, researchers and designers have begun to focus on the use of computational fluid dynamics (CFD) to analyze CC devices. Although most of the computational problem of the CC airfoil is straightforward, complications arise in the area of the Coanda jet itself. This jet is bounded by a curved wall on one side and a free shear layer on the other, and contains very-high-momentum fluid. This high momentum enables the jet to remain attached to the curved TE. The extent to which the jet remains attached controls the circulation and, hence, the lift generated by the airfoil. Thus, any computational technique, in order to be successfully applied to the CC problem, must be able to accurately predict the spreading rate of the jet and the location at which the Coanda jet finally separates from the curved TE of the airfoil. To accomplish this, the computational flow solver must be able to correctly predict the exchange of momentum between the Coanda jet and the surrounding fluid, the entrained upstream boundary layer, from the airfoil. Consequently, the computational mesh in the vicinity of the jet must be fine enough to adequately resolve the boundary layer between the wall and the jet, and the shear layer between the jet and the surrounding fluid. In addition, the type of turbulence model chosen for the problem will be crucial to successful modeling the Coanda jet and its interaction with the surrounding fluid, and subsequent prediction of the lift force generated by the CC airfoil. A recent paper6 reports good results from numerical solutions for CC airfoils using algebraic7 and one equation’ eddy-viscosity turbulence models. However,
R
FULL-REYNOLDS STRESS MODELING
447
the CC for these airfoils is essentially a blown flap method, where the jet separates from a sharp, rather than bluff, TE, which fixes the separation point. The general CC airfoil problem requires the jet to separate at some point along a curved wall (the bluff TE). Figure 1 depicts the streamlines around such an airfoil at zero degrees angle of attack and some finite free stream velocity. In the figure, the flow is from left to right, and the jet emerges from a slot above the curved trailing edge on the right hand side of the airfoil. The jet remains attached to the TE for some distance before finally separating. Also, the circulation increase caused by the jet has moved the leading edge (LE) stagnation point to a position below the LE. In general, curved wall jets like those on the CC airfoil have been problematical for simple eddy viscosity based turbulence models to predict. Although eddy-viscosity models can often be modified to improve their predictive accuracy for curved wall jets, these modifications are largely ad hoc, and cannot be easily eneralized for arbitrary flows and configurat ion^.^ For example, Slomski et al.% demonstrate that standard isotropic, twoequation turbulence models yield nonphysical solutions for a CC airfoil as blowing rate increases, whereas a full Reynolds-stress turbulence model reproduces the correct lift/blowin rate behavior for the same airfoil. Recently, however, Paterson and Baker 1% reported a successful simulation of the highest blowing rate case reported in Slomski et al.,9 using a blended k-w/k-E SST (shear stress transport) two-equation turbulence model. This chapter explores the performance of the Full Reynolds Stress Model (FRSM) for two-dimensional CC airfoils beyond the cases investigated in Slomski et al.9 and Paterson and Baker." Specifically, a full range of blowing slot heights, airfoil angles of attack, and two airfoil TE shapes are simulated.
Fig. 1 Typical CC airfoil showing Coanda jet and surrounding streamlines. Flow is from left to right. The jet is depicted by the thick group of streamlines at the trailing edge of the airfoil.
448
P. A. CHANG ET AL.
Based on the encouraging results reported in Paterson and Baker,” the performance of the k - w / k - e SST model in some of these new conditions is investigated.
11. Mathematical Development The steady, two-dimensional Navier-Stokes equations are solved using the finite volume code, Fluent. The segregated solver, with SIMPLE pressurevelocity coupling, is utilized. Second-order upwinding is used to discretize the convective terms in the momentum equations with second-order central differencing used on the viscous terms. First-order upwinding is used on density, energy, k, E and Reynolds stress equations. The effect of turbulent flow on the steady state solution is obtained using the FRSM of Launder, Reece and Rodi (LRR),” as well as the blended k - w / k - e SST model. In two dimensions, the FRSM introduces an additional five equations -three equations for each of the correlations UU, UV, and VV, and equations for k and E are solved in order to evaluate at the walls. A wall reflection term is invoked, which damps the normal stresses at the wall while enhancing the stresses parallel to the wall. Enhanced wall treatment is utilized, which solves to the wall where y+ 5 3 and uses wall functions valid in the buffer region including the effect of pressure gradients. The wall function is important because of the wide range of velocities over the foils, where upstream of the slot the grid has y+ x 1 but in the Coanda jet, y+ RZ 3-10. Numerical simulations of airfoils with 15% thickness-to-chord ratio, 1% camber, with a slot located at 97% chord, with a 6.7% thickness at the slot location, and two Coanda TE shapes5 are undertaken. Both the “nominal” circular TE foil, NCCR 1510-7067N, shown in Fig. 2, and the logarithmic spiral TE foil, NCCR 1510-70678, are used. The slot-height-to-chord ( h / c )ratios include 0.0015, 0.0022, and 0.0030. Incidence angles are 0, -4, and - 8 deg. The logarithmic spiral curve has a constantly increasing radius of curvature with the smallest radius at the slot. A comparison between the circular and logarithmic-spiral TE geometries is shown in Fig. 3. The rationale for a
LEADING EDGE
TRAILING EDGE
Fig. 2 Geometry of the NCCR 1510-7067N airfoil.
449
FULL-REYNOLDS STRESS MODELING
I 0.92
I 0.94
I 0.96
I 0.98
I 1.00
X/C
Fig. 3 Comparison of circular and logarithmic-spiralTE geometry; -, TE; - - -, logarithmic-spiral TE.
circular
logarithmic spiral TE is that for a given blowing rate the Coanda jet may stay attached a longer distance around the TE because of the decreasing curvature where the jet would tend to detach for a circular TE. This would reduce the power requirement necessary to obtain a given lift augmentation ratio (Rogers, E., personal communication, March 2004). When computing the solutions for the logarithmic spiral it was thought that the geometry had h / c = 0.0015 and, thus, the C, values were set to match the h = 0.0015 cases. After the fact, however, it was found that the geometry actually had h / c = 0.0020. This is between the experimental h/c values of 0.0015 and 0.0022. For comparison to results, the C, values were re-computed and the h = 0.0022 cases closest to the actual C, values are used for comparison. The computational grids have between 100,000 and 150,000 cells, depending on slot height. An 0-grid topology is used near the body with an H-grid in the wake extending approximately 13 chord lengths downstream. The LE and TE regions are shown in Figs. 3 and 4, respectively. The hybrid mesh consists of quadrilaterals with triangular elements in the slot exit as shown Fig. 6. On the body, boundary conditions are specified as no-slip except at the upstream end of the plenum where rit (mass flow rate) and pressure are specified. For the incompressible startup conditions, the upstream, outer boundary is set to a velocity inlet condition where the freestream speed is set to 41.65 m/s, vT/v, = 5, and k / U k = 0.05. Also, for the incompressible startup conditions, the downstream boundary is set to a pressure outlet with zero pressure. When the flow is assumed to be compressible, the air is assumed to be governed by the ideal gas law with the Sutherland law applied to the evaluation of molecular
450
P. A. CHANG ET AL.
Fig. 4 Grid for h / c = 0.0030 airfoil showing detail of LE.
viscosity; the outer boundaries are set to far-field pressure with M , = 0.12 and zero pressure. It is assumed that the freestream temperature is 288 K, with a freestream kinematic viscosity voo= 1.462 x lOP5m2/s and density, p, = 1.224 kg/m3. The chord length of the airfoil, c is 0.203 m, giving a freestream Reynolds number Re = 5.8 x lo5. In order to change the angle of attack a, the freestream velocity is rotated appropriately. A negative value of a denotes that the nose is pitched downward. The mass flow rate is nondimensionalized as the jet momentum coefficient
my c -- 1/2p,u2,c
Fig. 5 Grid for h / c = 0.0030 airfoil showing detail of Coanda jet region.
FULL-REYNOLDS STRESS MODELING
451
Fig. 6 Grid for h / c = 0.0030 airfoil showing detail of slot region.
where p, and V , are the freestream density and velocity, respectively, and c is the airfoil chord length. The experimental riz values were measured using a calibrated venturi meter that was inserted in the air supply line and the jet velocity vj was calculated as an isentropic expansion from duct pressure to freestream static pre~sure.~ Table 1 lists the cases for the circular arc TE with case numbers corresponding to those given in A b r a m ~ o nTable . ~ 2 lists the cases for the logarithmic spiral TE,
Table 1 Circular arc TE runs
293 289 283 311 307 302 330 326 321 60 57 53 229 227 223
0.050 0.092 0.209 0.048 0.093 0.189 0.047 0.090 0.184 0.052 0.104 0.201 0.053 0.103 0.198
0.0030 0.0030 0.0030 0.0030 0.0030 0.0030 0.0030 0.0030 0.0030 0.0015 0.0015 0.0015 0.0022 0.0022 0.0022
0 0 0 -4 -4 -4 -8 -8 -8 0 0 0 0 0 0
P. A. CHANG ET AL.
452
Table 2 Logarithmic spiral TE runs case’ 56
CFD 361 53
CFD 358 51 CFD 357
0.054 0.041 0.039 0.107 0.080 0.077 0.140 0.105 0.090
0.0015 0.0020 0.0022 0.0015 0.0020 0.0022 0.0015 0.0020 0.0022
showing the experimental C, values for h / c = 0.0015 and h / c = 0.022, as well as the C, values actually run with h/c = 0.0020. The flow is assumed to be compressible in order to validate the wind-tunnel experiments. Obtaining a well-converged solution is difficult because of the large range of length and velocity scales (e.g., the ratio of the jet to freestream velocities is as high as 6). Typically, the compressible RSM solutions are obtained using a multistep procedure: 1) Initial Coanda jet development: Incompressible flow, k--E turbulence model, underrelaxation factors (URFs) less than 0.2, run for several thousand iterations. 2) Coanda jet development and prediction of approximate separation point: Incompressible flow, FRSM turned on, URFs lowered to less than 0.1, about 5000 iterations. 3) Incorporation of compressibility effects: Compressible flow, k- E turbulence model, URFs less than 0.1, run for about 10,000 iterations. 4) Final jet development: Compressible flow, FRSM, URFs less than 0.1 for Reynolds stress equations, 0.2 for other equations, about 10,000 iterations. 5 ) Ensuring convergence to stable solution: Compressible flow, RSM, larger URFs 0.3-0.5, run for 20,000-30,000 iterations. For the solution with the k-w/k--E SST model this procedure is modified: 1) Initial Coanda jet development: Incompressible flow, k--E turbulence model, URFs less than 0.2, run for 10,000 iterations. 2) Incorporation of compressibility effects: Compressible flow, k-w turbulence model, URFs less than 0.1, run for about 5000 iterations. 3) Turn on k-w/k--E SST model: Compressible flow, URFs less than 0.1 for all equations, run for about 2000 iterations. 4) Ensuring convergence to stable solution: Compressible flow, k-w SST, URFs increased to 0.4, run for 10,000-20,000 iterations. Solutions are considered converged when there is no change in the integrated lift force over 10,000-20,000 iterations. The lift forces converge to steady-state values with no transient oscillations.
FULL-REYNOLDS STRESS MODELING
453
111. Results In this section qualitative aspects of the computed flows are shown, then the integrated lift vs angle of attack curves and surface pressure distributions about the foil are compared with experimental data. The lift coefficient is computed by
where Fy is the force in the y direction, pmand Urnare the freestream values of density and velocity magnitude, respectively, and A is the reference surface area cS, where c is the chord length and S the 1 m span (for the two-dimensional calculations, the forces are given in terms of force/unit span). The presure coefficient C, is computed by P c -- 1/2pwU&
(3)
A. C, Variation with h / c = 0.0030, a = 0 deg The Coanda jet changes the location of the detachment point on the trailing edge (TE) and with it, the circulation around the airfoil. As can be seen in Fig. 7, the TE detachment point and the LE stagnation point migrate around to the bottom of the foil as C, increases. The FRSM does not predict streamlines that wrap around to the bottom (referred to henceforth as “trailing edge pressure drawdown”) as was shown by Slomski et al.9 for isotropic turbulence models. The lift vs. C, curve (Fig. S), shows that the lift is underpredicted throughout the C, range, and where the experimental curve has a small amount of curvature, the predicted curve is almost linear. However, this is a major improvement over the large drop in lift due to trailing edge pressure drawdown as shown in Slomski et a1.9 Surface pressure distributions (Fig. 9) show that the reason why the predicted lift is low is because of an underprediction of the midchord pressure differential. The C, = 0.092 case, which has the largest discrepancy in midchord pressure differential, has the largest discrepancy in the predicted C,. The highest C, case (Fig. 9c), matches up with the experimental pressure data very well across the foil, but has just enough of a discrepancy in the midchord pressure differential to cause the under prediction of C, shown in Fig. 8.
B. Angle-of-AttackVariation with h / c = 0.0030 Figures 10 and 11 show the streamlines for a = -4 and - 8 deg, respectively. For both cases it can be seen that at the lower blowing rate the stagnation point is at the LE or on the upper surface. As C, is increased, the Coanda jet induces the stagnation point to migrate around to the bottom, in essence modifying the angle of attack. The integrated lift coefficients for the circular TE with h / c = 0.0030 vs. a for a = 0, -4, and - 8 deg are shown in Fig. 12. The experimental data show that as a becomes more negative, the amount of positive lift decreases.
P. A. CHANG ET AL.
454
Fig. 7 Streamlines for h / c = 0.0030 and a = 0 deg: Upper, C, = 0.050; middle, C, = 0.092; bottom, C, = 0.209.
5 4
3
0' 2
1
0 0.00 0.05
0.10
0.15
0.20
0.25
c, Fig. 8 Lift coefficient vs C, at h / c = 0.0030, a = 0 deg, comparing FRSM results to experimental results: -Experiment;' 0, FRSM.
FULL-REYNOLDS STRESS MODELING
455
b)
Fig. 9 Surface pressure distributions for hlc = 0.0030 at a = 0 deg: -, FRSM; 0,experiment-top; +, experiment-bottom. a) C, = 0.050, b) C, = 0.093, c) C, = 0.209.
However, because of the additional circulation caused by the Coanda jet, negative angles of attack can still have positive lift. There is a constant difference between the curves of constant a and a slight decrease in slope with increase in C,. The computational FRSM results show similar behavior, although they are low by about ACL % 0.5. Figures 13 and 14 show the pressure distributions for h / c = 0.0030 at -4 deg and - 8 deg, respectively. The results are consistent with the streamline plots (Figs. 10 and 11), which show that for the low C, cases the stagnation point is on the upper surface, migrating around to the lower surface as the C, increases. In all cases, the TE pressure peak is underpredicted with the discrepancy decreasing with increasing C, and smaller angle of attack. This seems to indicate that there is a discrepancy in the jet detachment point-that as angle of attack becomes more negative, the predicted jet detaches relatively earlier with respect to the experiment, but that as C, increases, the jet detachment points get closer to their experimental values.
456
P. A. CHANG ET AL.
Fig. 10 Streamlinesfor h / c = 0.0030 and cu = -4 deg: Upper; C, = 0.048; middle, C, = 0.093; bottom, C, = 0.189.
Fig. 11 Streamlinesfor h / c = 0.0030 and cu = -8 deg: Upper: C, = 0.047; middle, C, = 0.090; bottom, C, = 0.184.
FULL-REYNOLDS STRESS MODELING
457
5 4
3
0' 2 1
0
0.00
0.05
0.10
0.15
0.20
0.25
c, Fig. 12 Lift coefficient vs C, at h / c = 0.0030 for three angles of attack, comparing cu = 0 deg; ---,cu = -4 deg; FRSM results to experimental results. FRSM: -, ---, cu = -8 deg. Experiment5 symbols: 0, cu = 0 deg; 0, cu = -4 deg; H, cu = -8deg.
C. Slot Height Variation with (Y = 0 deg Lift coefficient vs C, for three slot heights, h / c = 0.0015,0.0022, and 0.0030 are shown in Fig. 15. For a given C,, the product hvj is constant so that as h / c decreases, vj must increase in inverse proportion to a decrease in h. Thus, jet velocities for the cases with smaller slot heights and higher C, values are very close to being supersonic. For h / c = 0.0015 the two higher C, cases did not converge. The experimental results show that at the lower values of C, there is very little change in CL with variation in slot height. As C, increases, the CL for h / c = 0.0030 falls away from the two smaller slot heights. The FRSM results show that for the lower two values of C, as h/c decreases, the discrepancies between experiment and predicted values decreases, with the smallest slot height, h / c = 0.0015, being right on the experimental data. For h / c = 0.0022 the FRSM values shows the correct trend at higher C,, a decreasing slope as C, increases. The pressure distributions for h/c = 0.0022 are shown in Fig. 16. They show that as C, increases, the peak TE pressure as well as the overall pressure compares increasingly well with experimental data. Fig. 17, the pressure distribution for h/c = 0.0015, C, = 0.052 shows very good comparison to experimental data, with only a very small underprediction of the peak TE pressure. These trends in the pressure distributions indicate that for the circular TE, for the jet detaches early compared with experiments, resulting in low values of I$ low midchord pressure differentials and lift. However, as vj increases, the jet detachment point extends further around, eventually matching up with the experimental location. In these cases, the midchord pressure differential and lift are well predicted.
458
P. A. CHANG ET AL.
a)
b)
Fig. 13 Surface pressure distributions for h / c = 0.0030 at a = -4 deg; -, FRSM; 0, experiment-top; +, experiment-bottom. a) C , = 0.048, b) C , = 0.093, c) C , = 0.189.
D. Logarithmic-Spiral TE Figure 18 shows the streamlines for the three C, cases run on the logarithmic spiral TE. Figure 18a shows that for the lower C, case the detachment point is well predicted. However, as C, increases, TE pressure drawdown is predicted as shown in Figs. 18b and 18c. Figure 19 shows that for the lower C, case, C, = 0.041, the pressure at the lower TE is correctly predicted. However, the predicted suction side pressures are low. For the C, = 0.080 and C, = 0.105 cases, the pressures on the lower side of the TE do not increase to their constant suction-side values because of the TE pressure drawdown. The pressure amplitudes at the LE are overpredicted, indicating excessive circulation. These results indicate that with the logarithmic spiral’s increasing radius of curvature, the FRSM is not sensitive enough to predict the correct detachment point.
E. Blended k-wlk-e SST Model Figure 20 compares Case 283 results with the k-w/k-• SST model with FRSM and experimental results. The k - W / k - - E SST results are similar to
FULL-REYNOLDS STRESS MODELING
459
b)
Fig. 14 Surface pressure distributions for h / c = 0.0030 at a = -8 deg; -, FRSM; experiment-top; +, experiment-bottom. a) C , = 0.047, b) C , = 0.090, c) C , = 0.184.
previous computations" in that they predict the Coanda jet detachment at the TE, rather than the TE pressure drawdown effect typical for other isotropic models.' In this case, however, the k-w/k--E SST results predict lower airfoil circulation than the experiment, as evidenced by a smaller difference in surface pressure magnitudes between the upper and lower surfaces of the airfoil. As shown in Fig. 21, this results in a lower CL = 2.82 as compared with 4.25 from experiments and 3.81 from FRSM. Paterson and Baker" obtained a value of CL = 4.0, using the blended k-w/k--E SST model. The difference between the results reported herein and Paterson and Baker's results may be due their use of overset gridding which allows a finer grid in the Coanda jet region. Pressure distributions and streamlines for the logarithmic spiral TE using the k-w/k--E SST model are shown in Figs. 22 and 23, respectively. Using the k-w/ k--E SST model, the TE pressure drawdown is not as severe as for the FRSM, with a pressure distribution on the lower side of the TE much closer to the experimental results. This generates a midchord pressure distribution much closer to the experimental values, although the peak pressure at the TE is slightly
P. A. CHANG ET AL.
460 5 1 4
3
0' 2 1
0 0.00
0.05
0.10
0.15
0.20
0.25
c, Fig. 15 Lift coefficient vs C, at (Y = Odeg for three values of slot height, hlc, comparing FRSM results with experimental results. FRSM: -, h / c = 0.0030, . . . . . ., h / c = 0.0022; ,h / c = 0.0015. Experiment5 symbols: 0,h / c = 0.0030; 0, h / c = 0.0022; A,h / c = 0.0015.
____
underpredicted. Table 3 shows that the experimental C, values for the h / c = 0.0015 and h / c = 0.0022 are 3.86 and 3.62, respectively. The FRSM result, CL = 3.97 is high because of the circulation induced by the trailing edge pressure drawdown. The k-w/k-• SST value, CL = 3.15, is low, consistent with the predictions for the circular arc TE.
F. Discussion It is difficult to say conclusively which turbulence models are best for the CC foil problem. The results presented in this paper have not been shown to be grid independent, for example. However, the following trends are evident: 1) Isotropic turbulence models. The Menter k - W / s - - E SST model appear to offer the best performance of the isotropic turbulence models. The results herein and from Paterson and Baker" bear this out. Notwithstanding Paterson Table 3 Lift coefficients for logarithmicspiral case with C, = 0.105 case'
CL
Expt. Case 56 Expt. Case 356 FRSM k-rn1k-E SST
3.86 3.62 3.97 3.15
FULL-REYNOLDS STRESS MODELING
461
b)
XIC
Fig. 16 Surface pressure distributionsfor h / c = 0.0022 at a = 0 deg, -, 0, experiment: a) C, = 0.053, b) C, = 0.103, c) C, = 0.198.
FRSM,
Fig. 17 Surface pressure distributions for h / c = 0.0015 at a = 0 deg for C, = 0.052; -, FRSM, 0, experiment.
P. A. CHANG ET AL.
462
b)
Fig. 18 Streamlines for logarithmic-spiral TE: a) C, = 0.041, b) C, = 0.080 and c) C, = 0.105.
a)
Fig. 19 Surface pressure distributions for logarithmic spiral cases: -, FRSMh / c = 0.0020, 0, experiment-h/c = 0.0015; +, experiment-h/c = 0.0020, a) C, = 0.041, b) C, = 0.080, c) C, = 0.105.
FULL-REYNOLDS STRESS MODELING
463
-20
-1 5
-5
0
0.0
0.2
0.6
0.4
0.8
1.0
xlc
Fig. 20 Surface pressure distributions comparing turbulence models for circular TE Case 283 ( h / c = 0.0030, a = 0 deg, C, = 0.209); -, FRSM: A,k - m SST; 0, experiment.
Fig. 21 Lift coefficient vs C, at a = 0 deg, h / c = 0.0030, comparing FRSM, k-m, and experimental results: -, experiment5;0, FRSM; A,k-w SST.
P. A. CHANG ET AL.
464
-20
-1 5
-1 0 0"
-5
0
0.0
0.2
0.6
0.4
0.8
1.0
X/C
Fig. 22 Surface pressure distributions comparing turbulence models for logarithmic-spiral TE, C, = 0.105: FRSM; A, k - o SST, 0,experiment-h/ c = 0.0015 (Case 51); 0, experiment-h/c = 0.0022 (Case 356).
Fig. 23 Streamlines for logarithmic-spiral TE using k - o turbulence model (Case 51, C, = 0.105).
and Baker's'' use of overset meshes, it is generally accepted that the Menter kW / S - - E SST model provides superior near-wall behavior (this model transitions to k-w in the near-wall region). The improved near-wall behavior over the k--E model may well do a better job of modeling the physics of the turbulent Coanda wall jet. 2) FRSM. These models appear to be better-suited for application to general CC foil problems. Mesh refinement studies are needed to explore fully the performance of these models, however. In addition, only the LRR FRSM was
FULL-REYNOLDS STRESS MODELING
465
exploited. There are other FRSM variants, such as the Launder-Shima12 FRSM, which are known to be less dissipative. Such models may offer improved predictive performance.
IV. Conclusions An extensive series of RANS calculations have been performed on twodimensional CC airfoils with circular arc and logarithmic-spiral TEs. It is shown that for a circular-arc TE, the full Reynolds stress turbulence closure can predict the Coanda jet detachment point fairly well for a range of angles of attack, jet slot heights, and jet blowing coefficients. For most cases the lift is low in comparison to experimental values. However, the trends in lift due to angle of attack and jet blowing coefficient are correctly predicted. The logarithmic-spiral TE is a much more challenging case; for higher jet blowing rates, the Coanda jet detaches upstream on the pressure (lower) side of the airfoil and the lift is overpredicted. For lower blowing rates, however, the correct detachment point is predicted. The k-w/k-E SST model is successful in predicting the detachment point for the circular TE, higher C, case, and in addition, is able to come closer to predicting the correct detachment point for the highest C, logarithmic-spiral case.
Acknowledgments This work was performed at the Naval Surface Warfare Center-Carderock Division (NSWCCD), West Bethesda, Maryland. It was sponsored by the Office of Naval Research, (Ronald D. Joslin, program manager) under work units 03-1-5400-616 and 04-1-5400-616. Computations were supported by a grant of High Performance Computing (HPC) time from the Department of Defense (DoD) HPC Shared Resource Centers, the U.S.Air Force’s Aeronautical Systems Center at Wright-Patterson Air Force Base, Ohio (Origin 3900, hpc11), and the U.S. Army’s Research Laboratory at Aberdeen Proving Ground, MD (IBM SP-4). The advice of Ernest Rogers is appreciated and duly noted. References ‘Englar, R., and Huson, G.,“Development of Advanced Circulation Control Wing High Lift Airfoils,” AIAA Aerospace Sciences Meeting, AIAA Paper 83-1847, Jan. 1983. ’Englar, R., Smith, M., Kelley, S., and Rover, R., “Development of Circulation Control Technology for Application to Advanced Subsonic Aircraft,” AIAA Aerospace Sciences Meeting, AIAA Paper 93-0644, Jan. 1993. 3Shrewsbury, G.,“Analysis of Circulation Control Airfoils Using an Implicit NavierStokes Solver,” AIAA Aerospace Sciences Meeting, AIAA Paper 85-0171, Jan. 1985. 4Shrewsbury, G., “Dynamic Stall of Circulation Control Airfoils,” Ph.D. Dissertation, Aviation and Surface Effects Department, Georgia Inst. of Technology, Atlanta, GA, Sept. 1990. ’Abramson, J., “Two-Dimensional Subsonic Wind Tunnel Evaluation of Two Related Cambered 15-Percent-Thick Circulation Control Airfoils,” Tech. Rept. ASED-373, DTNSRDC, Sept. 1977.
466 6
P. A. CHANG ET AL.
Liu, Y., Sankar, L., Englar, R., and Ahuja, K., “Numerical Simulations of the Steady and Unsteady Aerodynamic Characteristics of a Circulation Control Wing Airfoil,” 39th AIAA Aerospace Sciences Meeting, AIAA Paper 2001-0704, Jan. 2001. ’Baldwin, B. and Lomax, H., “Thin Layer Approximation and Algebraic Model for Separated Turbulent Flows,” AIAA Aerospace Sciences Meeting, AIAA Paper 78-0257, Jan. 1978. ‘Spalart, P., and Allmaras, S., “A One-Equation Turbulence Model for Aerodynamic Flows,” AIAA Paper 92-0439, Jan. 1992. ’Slomski, J. F., Gorski, J. J., Miller, R. W., and Marino, T. A., “Numerical Simulation of Circulation Control Airfoils as Affected by Different Turbulence Models,” 40th AIAA Aerospace Sciences Meeting & Exhibit, AIAA Paper 2002-0851, Jan. 2002. “Paterson, E. G. and Baker, W. J., “Simulation of Steady Circulation Control for Marine-Vehicle Control Surfaces,” 42nd AIAA Aerospace Sciences Meeting, AIAA Paper 2004-0748, Jan. 2004. “Launder, B., Reece, G.,and Rodi, W., “Progress in the Development of a ReynoldsStress Turbulence Closure,” Journal of Fluid Mechanics, Vol. 68, No. 3, 1975, pp. 537-566. ”Launder, B. and Shima, N., “Second-Moment Closure for the Near-Wall Sublayer: Development and Application,” AIM Journal, Vol. 27, No. 10, 1989, pp. 1319-1325.
1II.B. Tools for Predicting Circulation Control Performance: NCCR 103RE Airfoil Test Case
Chapter 18
Aspects of Numerical Simulation of Circulation Control Airfoils R. Charles Swanson,* Christopher L. Rumsey,+ and Scott G. Anders' NASA Langley Research Center, Hampton, Virginia
Nomenclature A = planform area, ft2 a = speed of sound, ft/s b = wing span, ft C , = section drag coefficient, D / ( q A ) C= ' surface skin friction coefficient, Tw/qoo CL = section lift coefficient, L / ( q . d ) C , = pressure coefficient, (p - poo)/qoo C, =jet momentum coefficient, (hV,)/(q,A) c = chord length, in. cr3= parameter for curvature effects h = slot height, in. k = turbulent kinetic energy per unit mass, ft . lb/slug M = Mach number, V / a m = mass flow rate, slug/s p = pressure, lb/ft2 q = dynamic pressure, 'pV2, lb/ft2 R = gas constant, ft . lbfslug . OR Re = Reynolds number, ( pV,c)/p T = Temperature, OR u, v = Cartesian velocity components, ft/s u, = friction velocity, ft/s V = velocity, ft/s
m,
*Senior Research Scientist, Computational AeroSciences Branch, Senior Member AIAA. 'Senior Research Scientist, Computational AeroSciences Branch, Associate Fellow AIAA. 'Research Engineer, Flow Physics and Control Branch, Senior Member AIAA. Copyright 02005 by the American Institute of Aeronautics and Astronautics, Inc. No copyright is asserted in the United States under Title 17, U.S. Code. The U.S. Government has a royalty-free license to exercise all rights under the copyright claimed herein for Governmental purposes. All other rights are reserved by the copyright owner.
469
R. C. SWANSON, C. L. RUMSEY, AND S. G. ANDERS
470
x, y = Cartesian coordinates, in. y+ = normalized coordinate, y ( u 7 / v ) a = angle of attack, deg y = specific heat ratio E = dissipation rate of k, ft . lb/slug/s p = coefficient of viscosity, lb . s/ft v = kinematic viscosity, ft2/s p = density, slug/ft3 T = shear stress, lb/ft2 w = specific dissipation rate of k, k / v t , s-'
Subscripts
c = based on chord length exp = refers to experiment j =jet condition ref = reference ( 0 0 ) condition t = turbulent flow quantity w = solid surface (wall) condition 0 = total condition 00 = freestream quantity I. Introduction ONVENTIONAL high-lift systems use slats and flaps to create the necessary airfoil camber to achieve the desired circulation, and thus lift. There is a weight penalty and increased maintenance associated with these systems. For a number of years,' aerodynamicists have been seeking alternative high-lift systems that can reduce the weight and complexity of the conventional systems. One such system for circulation control (CC) involves the Coanda effect. By controlling a jet discharged from a slot on the upper surface of the airfoil, the trailing edge (TE) stagnation point is moved toward the lower surface on a rounded TE, and the leading edge (LE) stagnation point is moved toward the lower surface as well. In this way the effective camber of the airfoil can be increased, resulting in the augmentation of lift. Previously, the weight and operational requirements of such systems have been unacceptable. The potential benefits of these CC systems in terms of reduced takeoff and landing speeds as well as increased maneuverability have encouraged aerodynamicists to reconsider such systems. Moreover, the benefits of using pulsed jets offer the genuine possibility of significantly mitigating the obstacles preventing the implementation of these CC systems.2 Computational methods will play a vital role in designing effective CC configurations. Certainly, detailed experimental data, such as velocity profiles and Reynolds stresses, will be absolutely essential for validating these prediction tools. Because of the cost of flow control experiments, design and parametric studies will strongly depend on accurate and efficient prediction methods. These methods must have the potential to treat pulsating jets, even multiple jets, for a broad range of flow conditions (e.g., Mach number, Reynolds
C
NUMERICAL SIMULATION OF CC AIRFOILS
47 1
number, angle of attack). In general, the numerical methods must be extendable to time-dependent and three-dimensional flows. A number of computational methods have been applied to CC airfoil flows. In 1985 Pulliam et al.3 used ARC2D: an implicit Navier-Stokes solver, to compute solutions for two of the CC configurations tested by Abramson and R ~ g e r sA .~ spiral grid that begins in the plenum and wraps around the airfoil several times was used for the computations. Turbulence modeling of the flow over the airfoil and Coanda surface was carried out by applying a modified form of the zeroequation model of Baldwin and lo ma^.^ A term was introduced in the model to account for streamline curvature effects. The modification includes a constant C,. This constant was modified for each set of experimental conditions, and a set is defined by Coanda geometry, freestream Mach number, angle of attack, and a range ofjet momentum coefficient C,. The C, was adjusted so that the computed C, matched the experimental value for one of the C, values. Then this C, was used in computing all of the cases for the given set of conditions. Certainly, this approach is not satisfactory in general for modeling the turbulence. Nevertheless, Pulliam et al. were able to obtain good comparisons with experimental data for all cases considered. This work demonstrated that accurate Navier-Stokes simulation of CC airfoil flows is possible, and turbulence modeling is the key issue. In 2002 Slomski et a1.' considered the effects of turbulence modeling on the prediction of CC airfoil flows. Calculations were performed for the NCCR 1510-7067 airfoil, which is a cambered, 15% thick, CC airfoil with a jet slot located on the upper surface just upstream of the TE. The airfoil was at 0 deg angle of attack. Two variations of a two-equation transport model ( k - ~model) and a Reynolds stress model were used for modeling turbulence. Predictions of surface pressures with the two-equation model compared favorably with the experimental data at low blowing rates. At high rates of blowing only the Reynolds stress model provided predictions that compared well with the data. A principal conclusion of Slomski et al. is that nonisotropic turbulence models are probably required for the simulation of CC airfoils or lifting surfaces. Recently, Paterson and Baker' used an incompressible Navier-Stokes code to calculate the flow over the same CC airfoil considered by Slomski et al. They obtained solutions for the high blowing rate case that Slomski et al. computed and a case with the same freestream conditions but an a of - 8 deg. The shear stress transport (SST) model of Menter" was used to model turbulence. Using this isotropic turbulence model, their predicted surface pressure distributions compared favorably with experiment, even though an incompressible simulation was performed. However, it should be pointed out that the variation in the ratio of the jet density to the freestream density for the a of zero degree case can vary roughly from 0.8 to 1.2. Thus, there are compressibility effects, and these may be quite important when attempting to predict the characteristics of the jet. In the current work various aspects of simulating CC airfoil flows are examined. These aspects include 1) flow conditions, 2) grid density, and 3) turbulence modeling. The primary purpose of this paper is to provide some guidelines for accurate solutions and to delineate improvements needed in numerical techniques to reliably predict CC flows, eventually including pulsed jets. The two-dimensional, compressible, mass-averaged Navier-Stokes equations are solved with a finite-volume approach for discretization. The equations are solved on a multiblock, patched grid, and a multigrid method with an implicit approximate
472
R. C. SWANSON, C. L. RUMSEY, AND S. G. ANDERS
factorization scheme is used to integrate the equations. Numerical solutions are obtained for flow over the CC geometry tested by Abramson and Rogers.’ Several turbulence models are considered, including models based on one transport equation and two transport equations. A two-equation explicit algebraic Reynolds stress model is also considered. The influence of turbulence modeling is revealed by comparing computed and experimental pressure distributions, as well as Coanda jet streamlines. The initial sections of this chapter concern the CC airfoil geometry and flow conditions, description of grids, numerical method, and boundary conditions. This is followed by a section on turbulence modeling, where particular emphasis is given to modifications introduced into the models, and also, implementation details of the models that can significantly affect their performance. In the final sections the numerical results are discussed and concluding remarks are given. 11. Geometry and Grid
The CC geometry for the 2004 Circulation Control Workshop” held at NASA Langley Research Center is the CC elliptical airfoil, which is designated NCCR 1510-7067 N. This airfoil has a chord of 8 in., thickness ratio of 15%, and a camber ratio of 1%. The jet slot height-to-chord ratio is 0.0030, which corresponds to a slot height of 0.024 in. Previously, we performed calculations for the CC airfoil that was tested by Abramson and Rogers5 (see also Wilkerson and Montana6). This airfoil, which is designated as 103RE (and also referred to as 103XW in the literature), has a chord of 18 in., thickness ratio of 16%, and a camber ratio of 1%. The jet slot height-to-chord ratio is 0.0021, which corresponds to a slot height of 0.0378 in. This CC airfoil is compared with the NCCR 1510-7067 N airfoil in Fig. 1. The most significant differences between the two configurations are the
:I; T I ..............................................
;
............... ........,....................... ................ ......., ......................
:
04
.~.......................................................................................................................
03
. j ..................
02
.;......................
0.1
-NCCR coarlirrates ------ 103RE coarlimtes
...............................................
...................:
1 , ~
.....................
~
~
~
~
------ i - - - -; * ....................... ........ ...~.......... ................ ......., ..........”... ..... ;*-I - - - - _ _ _ -*-;- ---- - - L - - - - _ -j ;
$ 0
,
-0.1
-0.2
......................
-0.4
~
~
.......................................................................................................................
1 1L
-0.3
~
i
L 0.2 0.6
0.4
xlc
Fig. 1 Geometry of airfoils.
0.8
1
NUMERICAL SIMULATION OF CC AIRFOILS
473
size of the plenum and the jet slot height. Because the computational grid for the 103RE airfoil was available, and this geometry is quite similar to the one of the workshop, we elected to use the 103RE airfoil in simulating the workshop cases. In order to compute solutions for the workshop cases, we applied the freestream conditions for these cases and matched the corresponding jet momentum coefficients. The coordinates defining the 103RE airfoil were provided by E. Rogers of the Naval Surface Warfare Center, Carderock Division (NSWCCD), and they are given in the Appendix of this chapter. These coordinates include the changes in the airfoil geometry caused when setting the jet slot height. In this chapter we consider CC airfoil flows for high and low freestream Mach numbers. The designated case numbers, which are associated with the experiments, and the flow conditions are given in Table 1. In addition to these primary cases, others at M , = 0.12 and a = 0 deg are computed at different C, levels. The definition of C, is given in the nomenclature, and some discussion of C, is given in a later section. For Case 302 the testing was done by Abramson and Rogers? and for Cases 283 and 321, the experimental data were obtained by Abramson.12 Surface pressure distributions are available from the experiments. There are no velocity profiles or Reynolds stresses to allow a detailed assessment of turbulence models. Nevertheless, pressure data provide an opportunity for initial evaluation of the models. The experimental lift coefficients were determined by integrating the surface pressures, and the drag coefficients were computed from wake survey data using a momentum deficit method. Thus, the experimental drag values include the propulsion effects due to the Coanda jet. There are no data available specifying the error bounds of the aerodynamic coefficients. Several sources of error in the experimental data were reported by Abramson.l 2 Although the experiments were generally two-dimensional, there were three-dimensional effects produced at the high blowing rates. Also, there were changes in the slot height caused by the higher pressures required for the high blowing rates. We have not accounted for these effects on the experimental data. For the numerical computations the domain surrounding the CC airfoil extended 20 chords away from the airfoil. This domain was partitioned with three blocks. At the interface boundary on the lower airfoil surface the grid is patched, as seen in Fig. 2, which displays the near-field of a medium-resolution grid with a total of 17,875 points. This grid includes 235 grid points around the entire airfoil and 49 points in the normal direction over the forward part of the airfoil. Over the aft part of the airfoil there are 101 points in the normal direction, and this number includes the points in the plenum for the jet. For the fine grid the number of cells in the medium grid is doubled in each coordinate direction,
Table 1 Flow conditions for CC airfoil flow Case
M,
Re,
a,deg
CP
302 283 321
0.6 0.12 0.12
5.2 x lo6 5.45 lo5 5.45 lo5
0 0 -8
0.0032 0.2090 0.1840
474
R. C. SWANSON, C. L. RUMSEY, AND S. G. ANDERS 0
1
0.5
0.5
0.5
0.4
0.4
0.3
0.3
0.2
0.2
0.1
0.1
u
s o
0
-0.1
-0.1
-0.2
-0.2
-0.3
-0.3
-0.4
-0.4
-0.5
-0.5 0
1
0.5 xlc
Fig. 2 Near field of medium grid for CC airfoil.
resulting in 70,563 points. The clustering of the grid at the airfoil LE and jet slot is clearly seen in Figs. 3 and 4. In the normal direction the grid is clustered at the surface so that the normalized distance y+ is less than one for the first point off the wall.
0.1
-0.05
0
0.05
0.1
0.1
0.08
0.08
0.06
0.06
0.04
0.04
0.02
0.02
-go
0
-0.02
-0.02
-0.04
-0.04
-0.06
-0.06
-0.08
-0.08
-0.1
-0.05
0
0.05
0.1
-0.1
XlC
Fig. 3 Leading-edge region of medium grid for CC airfoil.
NUMERICAL SIMULATION OF CC AIRFOILS 0.95
0
1
1.05
0.08
0.08
0.06
0.06
0.04
0.04
0.02
0.02
3
475
0
0
-0.02
-0.02
-0.04
-0.04
-0.06
-0.06 0.95
1
1.05
XlC
Fig. 4 Trailing-edge region of medium grid for CC airfoil.
111. Numerical Method
Numerical solutions were computed with CFL3D, a multizone mass-averaged Navier-Stokes code developed at NASA Langley. l 3 It solves the thin-layer form of the Navier-Stokes equations in each of the (selected) coordinate directions. It can use one-to-one, patched, or overset grids, and employs local time-step scaling, grid sequencing, and multigrid to accelerate convergence to steady state. In time-accurate mode, CFL3D has the option to employ dual-time stepping with subiterations and multigrid, and it achieves second-order temporal accuracy. Thus, this code has sufficient flexibility to solve either two-dimensional or threedimensional problems with multiple and/or pulsating jets. The code CFL3D is based on a finite-volume method. The convective terms are approximated with third-order upwind-biased spatial differencing, and both the pressure and viscous terms are discretized with second-order central differencing. The discrete scheme is globally second-order spatially accurate. The flux difference-splitting (FDS) method of Roe is employed to obtain fluxes at the cell faces. Advancement in time is accomplished with an implicit approximate factorization method (number of factors determined by number of dimensions). In CFL3D, the turbulence models are implemented uncoupled from the meanflow equations. The turbulent transport equations are solved with the same implicit approximate factorization approach used for the flow equations. The advection terms are discretized with first-order upwind differencing. The production source term is treated explicitly, while the advection, destruction, and diffusion terms are treated implicitly. For the explicit algebraic Reynolds stress (EASM-ko) model, the nonlinear terms are added to the Navier-Stokes equations explicitly.
R. C. SWANSON, C. L. RUMSEY, AND S. G. ANDERS
476
IV. Boundary and Initial Conditions Boundary conditions are required at the inflow (internal and external), outflow, and solid surface boundaries. For numerical computations the physical boundary conditions must be supplemented with numerical boundary conditions, which generally involve extrapolation of flow quantities or combinations of them (e.g., Riemann invariants) from the interior of the domain. Discussion of the numerical boundary conditions is given in the user’s manual for CFL3D.13 At the far-field inflow boundary a Riemann invariant, entropy, and flow inclination angle are specified. A Riemann invariant is specified at the far-field outflow boundary. For the plenum the mass flow rate and flow inclination angle are prescribed. If the mass flow rate is not known from the experiment, it is determined with an iterative process where it is changed until the experimental C, at the jet exit is matched. At the surface boundaries the no-slip and adiabatic wall conditions are specified. Boundary conditions for the various turbulence models considered herein are given in the CFL3D user’s manual. The initial solution is defined by the freestream conditions. V. Turbulence Modeling Several turbulence models for computing CC airfoil flows are considered. The three principal models are the one-equation Spalart-Allmaras (SA) model,14the Spalart- Allmaras rotation/curvature (SARC) and the two-equation shear-stress transport (SST) model of Menter’0”79’8. In addition, the zeroequation Baldwin-Lomax (BL) model’ and the explicit algebraic stress (EASM) model in k-w form (EASM-ko)” are used. The three primary models and the BL model are all linear eddy-viscosity models that make use of the Boussinesq eddy-viscosity hypothesis, whereas the EASM-ko model is a nonlinear model. The equations describing these four models can be found in their respective references. However, there are certain details concerning the implementation of the SARC and SST models that should be given here in order to facilitate the discussion of the numerical results. The SA model can be written in general form as
where V vt,and P, Vaiff,and Ddiss are the contributions associated with turbulence resulting from production, diffusion, and dissipation, respectively. The production term is given by N
P = C&l[l-523WV In the SARC model P is replaced by
.*
(2)
NUMERICAL SIMULATION OF CC AIRFOILS
477
where the function r* is the ratio of scalar measure of strain rate to the scalar measure of rotation, the function 7 depends on the Lagrangian derivative of the strain-rate tensor principal axes angle (see Ref. 16 for details), and crl = 1 , cr2 = 12, and cr3 = 0.6-1.0. As cr3 is increased, the turbulence production decreases near convex surfaces. Later, we will exploit this behavior to reduce the production of turbulence in the Coanda flow and, in so doing, explore its local and global effects. The production term Pk in the turbulent kinetic energy equation of the Menter SST model can be written as
where the stress tensor TU is defined as
and the partial derivatives are strain rates. The production term P, in the w equation of the SST model is proportional to Pk. Generally, in the computations with the SST model, the incompressible assumption is imposed, and the turbulent kinetic energy contribution is neglected. Thus,
where Sij is the strain-rate tensor, and S,S, represents the double dot product of two tensors. When the strain-rate tensor is used for Pk, the SST model will be designated SST(1994). In some versions of the SST model, also referenced as SST(base1ine) model herein, the vorticity is substituted for the strain rate. l7 In this case the production term is written as
Pk = 2ptWijwij = ( U t l f l 2
(8)
where is the magnitude of the vorticity vector. The vorticity is used with the default SST model in the CFL3D code. Certainly, one would not expect much difference in boundary-layer-type flows between using strain rate or vorticity in the production terms. The eddy viscosity determined with the SST model is defined as vt =
a1 k max (a1w;RF2)
(9)
where al is a constant, w is equal to the ratio of the turbulent dissipation rate to the turbulent kinetic energy, R = and F2 is a blending function. In a recent paper by Menter et a1.,20the R in Eq. (8) is replaced by S = In the default SST model in CFL3D the R is used. Attempts to use S instead of R in this work resulted in nonphysical behavior of the solution for high blowing rates.
,/m,
Jm.
478
R. C. SWANSON, C. L. RUMSEY, AND S. G. ANDERS
VI. Jet Momentum Coefficient A frequently used parameter in assessing the performance of CC devices is the jet momentum coefficient. This parameter is defined as
c,
l i j v j = pjvj2hb =-
q,A
$p,Vicb
where usually l i j is a measured quantity. In this definition the jet velocity vj is determined by isentropically expanding the plenum flow to the freestream static pressure. Thus, vj can be calculated from
In addition, C, can be rewritten as
If we assume fixed h / c and jet conditions,
Then for M , = 0.12 and M , = 0.6 (two freestream Mach numbers considered in this chapter)
Thus, for a given C, with M , = 0.12, the C, corresponding to M , = 0.6 is more than an order of magnitude smaller. One must keep this behavior in mind when considering C, as M , increases.
VII. Numerical Results The computational method described in previous sections was applied first to the high-Mach-number flow over the CC airfoil 103RE, which is Case 302 in Table 1. Calculations were performed on the medium grid. A comparison of the surface pressure distributions computed with the BL, SA, SST(baseline), and the anisotropic EASM-ko models is shown in Fig. 5 . There is a significant discrepancy between the calculated and experimental5 pressures for all of the turbulence models. Moreover, the predicted lift coefficient is about two times the experimental C, of 0.191 for all models. Because all of the models predict separation on the Coanda surface downstream of the location indicated by the
NUMERICAL SIMULATION OF CC AIRFOILS
479
-0.8
-0.4
0
0" 0.4
0.8
1.2
0
0.2
0.6
0.4
0.8
1
xlc
Fig. 5 Comparison of surface pressures computed with several turbulence models ( M , = 0.6, (Y = 0 deg, Re, = 5.2 X lo6, C, = 0.0032, medium grid).
experiment, this means that each model is producing near-wall eddy-viscosity values on the Coanda surface that are too high. Thus, too much high-momentum fluid is being transferred to the inner part of the shear layer. For the transport equation models this indicates that the production of turbulent kinetic energy (TKE) is too high. To determine the effect of reducing the TKE, we decided to use the curvature correction term in the SARC model as a vehicle for TKE reduction. As discussed in the turbulence modeling section, the cr3 parameter in the curvature correction term of the SARC model can provide a means to reduce the TKE in the Coanda flow. In Fig. 6 the influence of this parameter on the computed variations in pressure is displayed. With c3, = 9.6 there is good agreement with the experimental data. The calculated CLis 0.177, which underpredicts the experimental value by approximately 7%. This result is on the medium grid. Although the effect of grid density was not assessed for this case, the lower Mach number cases discussed below show little difference between the medium and fine grid results for the SARC model. Figure 7 shows the effect of cr3 on the variation in the turbulent viscosity pt in the direction normal to the airfoil trailing edge (x-axis). The dashed line represents c,3 = 0.6, which is the standard value for curvature correction, and the thin solid line refers to c,3 = 9.6. With cr3 = 9.6 there is a maximum reduction factor in pr of about 3 in the shear layer near the surface. Figures 8- 10 reveal the basic physics of the flow. In Fig. 8 the initial entrainment of the upper surface flow produced by the jet flow is discernible. A shear layer develops as a result of the entrainment. The early and later development of the shear layer is evident. The Mach contours (with an interval of 0.04) in
480
R. C. SWANSON, C. L. RUMSEY, AND S. G. ANDERS -0.8
-0.4
0 0"
0.4
0.8
1.2
0
0.4
0.2
0.6
0.8
1
xlc
Fig. 6 Effect of turbulence production parameter cr3 of SARC model on surface pressures ( M , = 0.6, (Y = 0 deg, Re, = 5.2 X lo6, C, = 0.0032, medium grid).
Fig. 9 indicate the rather thick boundary layers that develop on the upper and lower surfaces of the airfoil. They also suggest the separation of the Coanda jet. In Fig. 10 the separation of the jet flow is delineated by the streamline pattern. The flow over the blunt TE separates later with the jet than without
0.99262 0.9926 0.99258 0.99256
'
0.99254
0.99252 0.9925 0.99248 0.99246 0.992M0
10
20
30
I.lt
Fig. 7 Effect of turbulence production parameter cr3 of SARC model on turbulent viscosity ( M , = 0.6, a = 0 deg, Re, = 5.2 X lo6, C, = 0.0032, medium grid).
NUMERICAL SIMULATION OF CC AIRFOILS 0.968
0.969
0.97
0.971
48 1
0.972 0.036
0.035
0.034
0.033
0.032 0.968
0.969
0.97
0.971
0.972
X/C
Fig. 8 Velocity vectors near jet exit computed with SARC model and cr3 = 9.6 (Moo= 0.6, (Y = 0 deg, Re, = 5.2 X lo6, C, = 0.0032, medium grid).
the jet, but still upstream of the TE. This delay in separation results in one of the vortices normally appearing in the blunt TE region being eliminated. In the subsequent discussion we consider results for the same airfoil at low Mach number ( M , = 0.12), with several different blowing coefficients. For the 0.85
0.9
0.95
1
1.05
1.1
0.1
0.1
0.05
0.05
P* 0
0
-0.05
-0.1
-0.05
0.85
0.9
0.95
1
1.05
1.1
-0.1
X/C
Fig. 9 Mach contours at TE computed with SARC model and cr3 = 9.6 (Moo= 0.6, (Y = 0 deg, Re, = 5.2 X lo6, C, = 0.0032, medium grid).
482
R. C. SWANSON, C. L. RUMSEY, AND S. G. ANDERS 0.98
0.06
1
1.02
1.04
1.06
1.08 0.06
0.04
0.04
0.02
0.02
P*
0
0
-0.02
-0.02
-0.04
-0.04 0.98
1
1.02
1.04
1.06
1.08
XlC
Fig. 10 Streamline pattern at TE computed with SARC model and cr3 = 9.6 ( M , = 0.6, a = 0 deg, Re, = 5.2 X lo6, C , = 0.0032, medium grid).
first group of cases, solutions were obtained on the medium grid with the SA, SARC(c,3 = 9.6), and SST(base1ine) turbulence models for various C, values. Comparisons are made in Fig. 11 between the computed and experimental" pressure distributions for C, = 0.026. With the SA model there is significant disagreement with the data on the lower and upper surfaces of the airfoil. -4
-3 -2
0"
-1
0 1
2 o
0.2
0.4
0.6
0.8
1
X/C
Fig. 11 Surface pressures computed with SA, SARC(cr3= 9.6), and SST turbulence models (Moo= 0.12, a = 0 deg, Re, = 5.45 X lo5, C , = 0.026, medium grid).
NUMERICAL SIMULATION OF CC AIRFOILS 0.85
0.9
0.95
1
1.05
483
1.1
0.1
0.1
0.05
0.05
P* 0
0
-0.05
-0.1
-0.05
0.85
0.9
0.95
1
1.05
1.1
-0.1
XlC
Fig. 12 Jet streamlines computed with SARC(cr3 = 9.6) turbulence model ( M , = 0.12, a = 0 deg, Re, = 5.45 X lo5, C , = 0.026, medium grid).
There is improvement in the agreement with the SST(base1ine) model. The solution with the SARC model and c,3 = 9.6 exhibits relatively good agreement with the data. Figure 12 shows the Coanda jet streamlines for the SARC(c,3 = 9.6) model. The vortex pair usually occurring behind the blunt TE is conspicuously absent. 0
2500
-4 5 1 -5
5000
7500 -4.5
...................................
-5 -5.5 -6 -6.5 -7 -7.5 -8
Fig. 13 Residual histories with SA turbulence model, without and with preconditioning( M , = 0.12, (Y = 0 deg, Re, = 5.45 X lo5, C , = 0.026, medium grid).
484
R. C. SWANSON, C. L. RUMSEY, AND S. G. ANDERS
To provide some indication of convergence behavior of the computations, the variation with multigrid cycles in the L2 norm of the residual (for density equation) is presented in Fig. 13. Roughly 7500 cycles are required to reduce the residual four orders of magnitude. A major contribution to this slow convergence is the slowly converging plenum solution, which is a consequence of the very low-s eed flow in the plenum. The implementation of low-speed preconditioning,21- especially in the plenum, should result in a significant acceleration of convergence. Recently, we tested preconditioning for this particular case. Without any attempt to optimize the performance of the preconditioning, the number of cycles required to attain the same level of convergence obtained previously was reduced by a factor of two. It should be mentioned that the need for preconditionin to achieve accurate solutions in very low-speed regions has been demonstrated. 25 In Fig. 14 the computed pressures when C, = 0.093 are shown. Generally, the trends described for C, = 0.026 are exhibited here as well. For this case, solutions with both the SA and SST(base1ine) models indicate jet wraparound (i.e., Coanda jet moves onto the lower surface of the airfoil), as supported by the reduced pressures on the airfoil lower surface. These reduced pressures are associated with the occurrence of recirculation. The jet wraparound with the SA model is seen in Fig. 15. With the SARC(c,3 = 9.6) model there is generally good agreement with the data. However, a thin separation region (about 0.01 chord in maximum thickness) occurs just downstream of the airfoil LE. This separation results in a barely discernible plateauing effect on the calculated pressures in Fig. 14, which is not consistent with the experimental data. Figure 16 shows the jet streamlines for the SARC model and the stronger jet penetration (relative to that in Fig. 12) into the flowfield because of the increased C,.
2
-1 0
-8 -6
-4 0 "
-2
0 2
4 0
0.2
0.4
0.6
0.8
1
X/C
Fig. 14 Surface pressures computed with SA, SARC(cn = 9.6), and SST turbulence models ( M , = 0.12, (Y = 0 deg, Re, = 5.45 X lo5, C, = 0.093, medium grid).
NUMERICAL SIMULATION OF CC AIRFOILS 0.6
0.2
0.7
0.8
0.9
1
1.1
485 0.2
0.1
0.1
0
0
s -0.1
-0.1
-0.2
-0.2
-0.3
-0.3
-0.4
0.6
0.7
0.8
0.9
-0.4
1.1
1
X/C
Fig. 15 Jet streamlines computed with SA turbulence model ( M , = 0.12, (Y = 0 deg, Re, = 5.45 X lo5, C, = 0.093, medium grid).
The final two cases, Case 283 and Case 321, are those considered in the 2004 Circulation Control Workshop held at NASA Langley Research Center. Flow conditions for these cases are given in Table 1. For Case 283, where C, = 0.209, the computed pressure distributions on the medium grid are 0.85
0.9
0.95
1
1.05
1.1
0.1
0.1
0.05
0.05
Y>, 0
0
-0.05
-0.1
-0.05
0.85
0.9
0.95
1
1.05
1.1
-0.1
XlC
Fig. 16 Jet streamlines computed with SARC(cr3 = 9.6) turbulence model ( M , = 0.12, a = 0 deg, Re, = 5.45 X lo5, C, = 0.093, medium grid).
486
R. C. SWANSON, C. L. RUMSEY, AND S. G. ANDERS -1 0
-8 -6 -4
0
2 4
0
0.2
0.4
0.6
0.8
1
X/C
Fig. 17 Surface pressures computed with SA, SARC(cr3= 9.6), SARC(cr3= 0 - 9.6), and SST turbulence models (Moo= 0.12, (Y = 0 deg, Re, = 5.45 X lo5, C, = 0.209, medium grid).
compared with the experimental data in Fig. 17. There is considerable reduction in the computed lower surface pressures with the SA and SST(base1ine) models relative to the experimental values. Such behavior indicates extensive flow separation on the lower surface with these models. In fact, the Coanda jet in these cases wraps around the TE and moves even further upstream than shown in Fig. 15, a completely unphysical situation. The result with the SARC(cr3 = 9.6) model exhibits fairly good agreement with the data on the lower airfoil surface, but it shows a plateau behavior over more than 50% of the airfoil on the upper surface. Thus, there is a large separation bubble on the upper surface. Numerical tests confirmed that this is a consequence of the large cr3 value being used for the SARC model in the airfoil LE region. By simply setting cr3 = 9.6 on the Coanda surface and taking it to be zero elsewhere, relatively good agreement with the data is again obtained for the SARC(cr3 = 0 - 9.6) model. The jet streamlines for the SARC(cr3 = 0-9.6) model on the fine grid are presented in Fig. 18. In the Mach contours of Figs. 19 and 20 the rearward movement of the LE stagnation point, due to the Coanda effect, and the acceleration of the Coanda flow are seen. Details of the Mach contours at the jet exit, along with the corresponding fine grid, are displayed in Figs. 21 and 22. The jet flow is accelerated to a Mach number exceeding 0.9, indicating the compressible character of the jet. There is only a small effect of mesh refinement on the solution calculated with the SARC(cr3 = 0-9.6) model. Although not shown, the fine grid solution for the surface pressures nearly coincides with the medium grid solution. In addition, the velocity fields for the two grids are quite similar, as evident in
NUMERICAL SIMULATION OF CC AIRFOILS 0.85
0.9
0.95
1
1.05
487
1.1
0.1
0.1
0.05
0.05
Y* 0
0
-0.05
-0.05
-0'1 0.85
0.9
0.95
1
1.05
1.1
-0.1
XlC
Fig. 18 Jet streamlines computed with SARC(cr3 = 0 - 9.6) turbulence model ( M , = 0.12, a = 0 deg, Re, = 5.45 X lo5, C, = 0.209, fine grid).
the velocity profiles shown in Figs. 23 and 24. Table 2 compares the predicted lift and drag coefficients with the experimental values. In addition, the changes in aerodynamic coefficients with further increases in C, are indicated. There are two factors one should keep in mind regarding this table. First, as indicated previously, the experimental CDvalues include the thrust effects produced by the jet,
0.2°.2
-0.1
0
0.1
0.2
0.1
0.1
0
$ 0
-0.1
-0.2
0.2
-0.1
-0.2
-0.1
0
0.1
0.2
-0.2
XlC
Fig. 19 Mach contours computed at LE with SARC(cr3 = 0 - 9.6) turbulence model ( M , = 0.12, a = 0 deg, Re, = 5.45 X lo5, C, = 0.209, fine grid).
4aa
R. C. SWANSON, C. L. RUMSEY, AND S. G. ANDERS 0.85
0.9
0.95
1
1.05
1.1
0.1
0.1
0.05
0.05
P* 0
0
-0.05
-0.1
-0.05
0.85
0.9
1
0.95
1.05
1.1
-0.1
XlC
Fig. 20 Mach contours computed at TE with SARC(cn = 0 - 9.6) turbulence model (Moo= 0.12, a = 0 deg, Re, = 5.45 X lo5, C, = 0.209, fine grid).
whereas the computed CD values do not. Secondly, there is some effect, although it may be small, on these low-speed predictions because of the differences between the 103RE and the NCCR geometries. For Case 283 given in Table 2 the calculated CL is about 25% lower than the experimental CL.A rather large increase in the C, is needed to attain nearly the 0.966
0.968
0.97
0.972
0.038
0.038
0.036
0.036
P* 0.034
0.034
0.032
0.032
0.966
0.968
0.97
0.972
XlC
Fig. 21 Fine grid in jet exit region.
NUMERICAL SIMULATION OF CC AIRFOILS 0.966
0.968
0.97
489
0.972
0.038
0.038
0.036
0.036
Y* 0.034
0.034
0.032
0.032 0.966
0.968
0.97
0.972
XlC
Fig. 22 Mach contours in the vicinity of jet exit computed with SARC(cr3 = 0 - 9.6) turbulence model (Moo= 0.12, (Y = 0 deg, Re, = 5.45 X lo5, C , = 0.209, fine grid).
0.008
1
1
0 007
.........
0 006
.........!
0 005 L.......................................................
Y*
.............
jr
_
0 004 ........................................................ 0 003 I.......................... 0 002 I.......................... 0 001 -..........................
..............................
.............
i
..............................
.............................
i
..............................
i
;+...................
1 i
..............................
I
~
.............................
i
1 i
................
1
...............................................................................................
0 -0.001
.............................
i
1i.............................
medium grid fine grid
"
"
I
"
"
I
"
"
I
"
"
Fig. 23 Effect of mesh density on velocity profiles computed at jet exit with SARC(cr3 = 0 - 9.6) turbulence model (Moo= 0.12, a = 0 deg, Re, = 5.45 X lo5, C , = 0.209).
490
R. C. SWANSON, C. L. RUMSEY, AND S. G. ANDERS
medium grid fine grid
..........................................................
..........................................................
0
0.1
0.2 (u2
0.3
0.4
+ v2)1/*/aref
Fig. 24 Effect of mesh density on velocity profiles computed at TE with SARC(cr3 = 0 - 9.6) turbulence model (Moo= 0.12, (Y = 0 deg, Re, = 5.45 X lo5, C, = 0.209).
same CL as in the experiment. The small effects of mesh refinement result in the predicted lift and drag coefficients decreasing by 3.4% and 4.4% (relative to the medium mesh values), respectively. To provide an equitable assessment of the SST model, we consider a frequently used alternative implementation. As indicated in the section on turbulence modeling, Menter'* has considered two ways to define the turbulence production terms of the SST model. For all of the previous SST(base1ine) results that we have shown, the production term was computed with vorticity (see Eq. 8). The next results show the impact of evaluating the production term using the principal strain-rate tensor (Eq. 7). As mentioned earlier, we refer to this form of the SST model as SST( 1994). Table 2 Comparison of computed [with SARC(c,3 = 0 - 9.6) model] and experimental lift and drag coefficients for CC airfoil Case
c,
Grid
(CLIexp
CL
(CD)exp
CD
283 283
0.209 0.209 0.281 0.342 0.184 0.184
Medium Fine Medium Medium Medium Fine
4.20 4.20
3.26 3.15 3.62 4.05 2.17 2.03
- 0.050 - 0.050 -
0.1140 0.1090 0.1560 0.2100 0.0957 0.0922
321 321
-
3.10 3.10
- 0.080
- 0.080
NUMERICAL SIMULATION OF CC AIRFOILS
49 1
-1 0
-8 -6 -4 0"
-2 0
2 4
0
0.2
0.6
0.4
0.8
1
XlC
Fig. 25 Surface pressures computed with two versions of SST turbulence model (Moo= 0.12, a = 0 deg, Re, = 5.45 x lo5, C, = 0.209).
A comparison of the pressure distributions calculated with the SST(base1ine)and SST(1994) turbulence models is shown in Fig. 25 for Case 283. Both medium- and fine-grid results are given. There is relatively good agreement with the data when applying the SST( 1994) model, whereas the SST(base1ine) results exhibit poor 1 0.9 0.8 0.7 0.6 0.5
O' 0.4 0.3 0.2 0.1
0 -0.1
0.96
0.97
0.98
0.99
1
X/C
Fig. 26 Comparison of surface skin-friction distributions at the TE computed with SARC(cr3= 0 - 9.6) and SST(1994) turbulence models ( M , = 0.12, a = 0 deg, Re, = 5.45 X lo5, C, = 0.209).
492
R. C. SWANSON, C. L. RUMSEY, AND S. G. ANDERS 0.85
0.9
0.95
1
1.05
1.1
0.1
0.1
0.05
0.05
Y s
o
0
-0.05
-0.1
-0.05
0.85
0.9
0.95
1
1.05
1.1
-0.1
X/C
Fig. 27 Jet streamlines and Mach contours computed with SST(1994) turbulence model ( M , = 0.12, (Y = 0 deg, Re, = 5.45 X lo5, C, = 0.209, fine grid).
agreement. Although use of Eq. (8) for the SST model has proven to be satisfactory for many aerodynamic flows of interest, it does not appear to be appropriate for the Coanda jet flows being considered here; the SST( 1994) model performs better for these particular low-Mach-number Coanda flows. There is greater sensitivity to mesh refinement with the SST( 1994) model than that experienced with the SARC(cr3 = 0-9.6) model. The effect of mesh refinement on the Coanda surface skin-friction distributions calculated with these two models is shown in Fig. 26. Comparing Figs. 18 and 27, the jet streamlines with the SST( 1994) model exhibit less spreading than those with the SARC(cr3 = 0 - 9.6) model. Mesh refinement effect on the predicted CL and CD with the SST(1994) model is given in Table 3. On the fine grid, the predicted CL for Case 283 is 7.6% below that of the experiment. However, as shown in Fig. 28, the lift augmentation (slope of CL vs C,) appears to remain about the
Table 3 Comparison of computed, [with SST(1994) model] and experimental lift and drag coefficients for CC airfoil
283 283 321 321
0.209 0.209 0.184 0.184
Medium Fine Medium Fine
4.20 4.20 3.10 3.10
4.19 3.88 2.96 2.41
- 0.050
- 0.050 - 0.080 - 0.080
0.0966 0.0746 0.0655 0.0559
NUMERICAL SIMULATION OF CC AIRFOILS
493
4.5 4 3.5 3 2.5 ' 0
2 1.5 1
0.5
O O
0.05
0.15
0.1
0.2
0.25
c, Fig. 28 Variation of lift coefficientwith jet momentum coefficientusing SARC(cr3 = 0 - 9.6) and SST(1994) turbulence models ( M , = 0.12, a = 0 deg, Re, = 5.45 X lo5).
same for SST(1994) with mesh refinement. In the CL predictions with both models shown in Fig. 28, there is a monotonic increase in CL with increasing C,. The two-equation k--E models considered by Slomski et a1.' result in a nonphysical decrease in CLbeyond a C, of 0.093 (i.e., jet wraparound predicted).
-20 -1 6 -1 2
-4
0
4 0
0.2
0.6
0.4
0.8
1
XlC
Fig. 29 Surface pressures computed with SARC(cr3 = 0 - 9.6) turbulence model ( M , = 0.12, a = -8 deg, Re, = 5.45 X lo5, C, = 0.184).
494
R. C. SWANSON, C. L. RUMSEY, AND S. G. ANDERS
XlC
Fig. 30 Surface pressures computed with SST(1994) turbulence model ( M , = 0.12, a = -8 deg, Re, = 5.45 X lo5, C, = 0.184).
For the second case (Case 321, angle of attack of -8 deg) of the workshop, computed surface pressures for the medium and fine grids are presented in Figs. 29 and 30. Results with both the SST(1994) and SARC(c,3 = 0 - 9.6) models compare favorably with the experimental data. Nevertheless, the 0.85
0.9
1
0.95
1.05
1.1
0.1
0.1
0.05
0.05
P* 0
0
-0.05
-0.1
-0.05
0.85
0.9
0.95
1
1.05
1.1
-0.1
XlC
Fig. 31 Jet streamlines and Mach contours computed at TE with SARC(cr3 = 0 - 9.6) turbulence model ( M , = 0.12, a = -8 deg, Re, = 5.45 x lo5, C, = 0.184, fine grid).
NUMERICAL SIMULATION OF CC AIRFOILS
495
Fig. 32 Jet streamlines and Mach contours computed at TE with SST(1994) turbulence model ( M , = 0.12, (Y = -8 deg, Re, = 5.45 X lo5, C, = 0.184, fine grid).
experimental C, is underpredicted on the fine grid by more than 22% (see Tables 2 and 3). As in the previous case (Case 283) one of the effects of grid refinement seems to be reduced circulation, which results in the pressures on the airfoil suction surface increasing. This effect appears to be much greater for the current case because of the - 8 deg angle of attack. Paterson and Baker’ obtained approximately the same value for the CL of this case using the SST( 1994) model and performing an incompressible simulation for flow over the NCCR-15107067 N geometry. With the SARC(c,3 = 0 - 9.6) model there is again greater spreading of the jet than with SST(1994), as revealed by comparing Figs. 31 and 32, which depict the jet streamlines and Mach contours. There is an extremely small recirculation region, which occurs only for the SST( 1994) model, on the lower surface that centers near the 0.92 chord location, but it is not visible in Fig. 32.
VIII. Conclusions A computational method (CFL3D) has been applied to both low- and highsubsonic Mach number CC airfoil flows. Several turbulence models have been investigated. These models include the one-equation SA model with curvature correction (SARC) and two variations of the two-equation shear stress transport (SST) model of Menter. For the high-subsonic Mach number CC flow (Case 302), all models have predicted jet separation from the Coanda surface downstream of the experimental location, resulting in a significant overprediction of lift. In other words, all of the models have produced near-wall
496
R. C. SWANSON, C. L. RUMSEY, AND S. G. ANDERS
eddy-viscosity levels that are too high in the Coanda flow. A parameter (c,3) in the curvature correction term of the SARC model has been used as a vehicle to explore the effect of reducing the turbulent kinetic energy in the Coanda flow. In so doing, relatively good agreement with the experimental pressure distribution of Case 302 has been obtained, even though the required c,3 value is unrealistically high. In the simulation of low Mach number CC airfoil flows a set of calculations has been performed for a range of values of C., The two cases of the 2004 Circulation Control Workshop have also been considered. Relatively good agreement with experimental pressure data has been obtained when modeling turbulence with the SARC(c,3 = 0 - 9.6) and the SST(1994) models. The SST(1994) model uses principal strain rate for the shear stress in the modeling of the turbulence production. The SST(base1ine) model, which uses vorticity in the turbulence production term, has not been satisfactory when computing Coanda jet flows. An indication of the effects of grid refinement on the results computed with the turbulence models has been given. The SST( 1994) model has shown greater sensitivity to mesh refinement than the SARC(0 - 9.6) model. Lift and drag coefficients have also been determined in the calculations. Clearly, turbulence modeling is the major component in determining the success of a computational method for predicting CC airfoil flows. Most standard models, including SA, SARC (c,3 5 l.O), SST(baseline), and EASM-ko, have predicted jet separation too far around the Coanda surface. Accounting for streamline curvature effects has been shown to be important, although the SARC model required an artifically high level of its cr3 parameter in order to produce reasonable results when compared with these particular experiments. It is appropriate to note that in comparison to a different CC experimentz4 the SARC model with its recommended value (cr3 = 1.0) worked reasonably well, and the SST( 1994) model performed poorly. Further investigation of models is essential to achieving a reliable prediction technique that can be used for a broad range of flow conditions. In addition, improvements in computational efficiency must also be considered quite important if the prediction method is to be applied on a routine basis with a high degree of reliability. Some rather straightforward numerical algorithm features such as low-speed preconditioning should be included in the method. Potential benefits of this preconditioning have been indicated in this paper. Another possible improvement in computational performance can be achieved by full coupling of the fluid dynamic and turbulence transport equations, which is not done currently with the CFL3D code. These and other improvements in computational efficiency are especially important as the heiarchy (i.e., complexity) of the turbulence modeling is increased. For example, if a full Reynolds stress model is used instead of a two-equation model, such as SST( 1994), one must anticipate that there will be a reduction in computing efficiency, because of a lower degree of numerical compatibility of the more complex model. In the case of steady flows, numerical compatibility can be defined as a measure of the effect on solution convergence of the complete system of flow equations due to turbulence modeling.
NUMERICAL SIMULATION OF CC AIRFOILS
497
Acknowledgments The authors would like to thank E. Rogers of the Naval Surface Warfare Center, Carderock Division, for providing coordinates of the 103RE airfoil and experimental data. Appendix: Coordinates of 103RE Airfoil XI.
0.91346 0.91436 0.91762 0.92444 0.93516 0.94627 0.95247 0.95592 0.95892 0.96079 0.96232 0.96419 0.96527 0.96646 0.96776 0.96920 0.97078 0.97250 0.97439 0.97644 0.97866 0.98108 0.98367 0.98643 0.98933 0.99231 0.99512 0.99753 0.99925
Y I.
0.010565 0.016505 0.023061 0.028929 0.031569 0.032378 0.032618 0.032815 0.033013 0.033051 0.033040 0.032955 0.032875 0.032761 0.032607 0.032403 0.032117 0.031733 0.031228 0.030569 0.029731 0.028721 0.027438 0.025852 0.023862 0.021385 0.018159 0.014103 0.0091942
XI.
0.99999 0.99939 0.99700 0.99243 0.98563 0.97807 0.96976 0.96123 0.95121 0.93948 0.92582 0.90993 0.89147 0.87004 0.84518 0.81636 0.78295 0.74424 0.69939 0.64742 0.58721 0.5 1748 0.43673 0.35601 0.28917 0.23385 0.18811 0.15032 0.11915
YlC
XI.
0.0035035 -0.0027953 -0.0093482 -0.015540 -0.020621 -0.024547 -0.027987 -0.030852 -0.033717 -0.036499 -0.039348 -0.042273 -0.045275 -0.048349 -0.051494 -0.05471 1 -0.058003 -0.061375 -0.064825 -0.068009 -0.070608 -0.072286 -0.072472 -0.070647 -0.067442 -0.063413 -0.058917 -0.054098 -0.049115
0.093469 0.072353 0.055038 0.040894 0.029411 0.020176 0.012873 0.0072776 0.0032691 0.00083659 0.00027135 0 0.00022742 0.00075518 0.0031043 0.0070381 0.012574 0.019838 0.029062 0.040572 0.054791 0.072250 0.093598 0.11963 0.15131 0.18980 0.23654 0.29326 0.36203
Y I.
-0.0441 14 -0.039186 -0.034379 -0.029713 -0.025191 -0.020794 -0.016494 -0.012252 -0.0080402 -0.0038889 -0.0021 13 0 0.0021185 0.0039056 0.0081157 0.012430 0.016818 0.021317 0.025973 0.030832 0.035930 0.041298 0.046950 0.052894 0.059114 0.065569 0.072089 0.078324 0.083837
XIC
Y I.
0.44534 0.087908 0.52873 0.089357 0.60007 0.08861 1 0.66105 0.086478 0.71316 0.083469 0.75764 0.079735 0.79558 0.075445 0.82791 0.070840 0.85543 0.066064 0.87885 0.061361 0.89877 0.056818 0.91570 0.052486 0.93008 0.048398 0.94228 0.044569 0.95262 0.041005 0.96137 0.037705 0.96877 0.034657 0.96877 0.034491 0.94527 0.038834 0.89801 0.038252 0.83406 0.030760 0.50044 0.030760 0.50044 - 0.041077 0.73553 -0.041077 0 0.91346 0.91346 0.010565
References ‘Englar, R. J., and Williams, R. M., “Test Techniques for High Lift Two-Dimensional Airfoils with Boundary Layer and Circulation Control for Application to Rotary Wing Aircraft,” Canadian Aeronautics and Space Journal, Vol. 19, No. 3, 1973, pp. 93-108. ’Jones, G. S., Viken, S. A, Washburn, A. E., Jenkins, L. N., and Cagle, C. M., “An Active Flow Circulation Controlled Flap Concept for General Aviation Aircraft Applications,” AIAA Paper 2002-3157, June 2002. 3Pulliam, T. H., Jespersen, D. C, and Barth, T. J., “Navier-Stokes Computations for Circulation Control Airfoils,” AIAA Paper 85-1587, July 1985. 4Pulliam, T. H., “Euler and Thin Layer Navier-Stokes Codes: ARC2D, ARC3D,” Notes for Computational Fluid Dynamics User’s Workshop, Univ. of Tennessee Space Inst., Tullahoma, TN, March 1984.
498
R. C. SWANSON, C. L. RUMSEY, AND S. G. ANDERS
’Abramson, J., and Rogers, E., “High-speed Characteristics of Circulation Control Airfoils,” AIAA Paper 83-0265, Jan. 1983. 6Wilkerson, J. B., and Montana, P. S., “Transonic Wind Tunnel Test of a 16 Percent Thick Circulation Control Airfoil with 1 Percent Asymmetric Camber,” DTNSRDC ASED 82/03, April 1982. ’Baldwin, B. S., and Lomax, H., “Thin Layer Approximation and Algebraic Model for Separated Flows,” AIAA Paper 78-257, Jan. 1978. ‘Slomski, J. F., Gorski, J. J., Miller, R. W., and Marino, T. A., “Numerical Simulation of Circulation Control Airfoils as Affected by Turbulence Models,” AIAA Paper 20020851, Jan. 2002. ’Paterson, E. G.,and Baker, W. J., “Simulation of Steady Circulation Control for Marine-Vehicle Control Surfaces,” AIAA Paper 2004-0748, Jan. 2004. “Menter, F. R., “Two-Equation Eddy-Viscosity Turbulence Models for Engineering Applications,” AIAA Journal, Vol. 32, No. 8, 1994, pp. 1598-1605. “Jones, G.S., and Joslin, R. D. (ed.), Proceedings of the 2004 NASA/ONR Circulation Control Workshop, NASA/CP 2005-213509, March 2004. ”Abramson, J., “Two-Dimensional Subsonic Wind Tunnel Evaluation of Two Related Cambered 15-Percent Circulation Control Airfoils,” DTNSRDC ASED-373, Sept. 1977. 13Krist,S. L., Biedron R. T., and Rumsey, C. L., “CFL3D User’s Manual,” NASA TM 1998-208444, June 1998. 14Spalart,P. R., and Allmaras, S. R., “A One-Equation Turbulence Model for Aerodynamic Flows,” La Recherche Aerospatiale, Vol. 1, 1994, pp. 5-21. ”Spalart, P. R., and Shur, M., “On the Sensitization of Turbulence Models to Rotation and Curvature,” Aerospace Science and Technology, Vol. 5, 1997, pp. 297-302. 16Rumsey,C. L., Gatski, T. B., Anderson, W. K., and Nielsen, E. J., “Isolating Curvature Effects in Computing Wall-Bounded Turbulent Flows,” International Journal of Heat and Fluid Flow, Vol. 22, 2001, pp. 573-582. ”Menter, F. R., “Improved Two-Equation k - o Turbulence Model for Aerodynamic Flows,” NASA TM 103975, Oct. 1992. 18Menter,F. R., “Zonal Two Equation k - o Turbulence Model for Aerodynamic Flows,” AIAA Paper 93-2906, July 1993. ‘’Rumsey, C. L., and Gatski, T. B., “Summary of EASM Turbulence Models in CFL3D with Validation Test Cases,” NASA/TM-2003-212431, June 2003. ”Menter, F. R., Kuntz, M., and Langtry, R., “Ten Years of Industrial Experience with the SST Turbulence Model,” Turbulence, Heat and Mass Transfer 4, edited by K. Hanjalic, Y. Nagano, and M. Tummers, Begell House, Redding, CT, 2003, pp. 625-632. ’lTurke1, T., Vatsa, V. N., and Radespiel, R., “Preconditioning Methods for Low-Speed Flow,” AIAA Paper 96-2460, June 1996. ”Turkel, T., Radespiel, R., and Kroll, N., “Assessment of Two Preconditioning Methods for Aerodynamic Problems,” Computers and Fluids, Vol. 26, No. 6, 1997, pp. 613-634. 23Turkel, T., “Preconditioning Techniques in Computational Fluid Dynamics,” Annual Review of Fluid Mechanics, Vol. 31, 1999, pp. 385-416. 24Swanson,R. C., Rumsey, C. L., and Anders, S. G.“Progress Towards Computational Method for Circulation Control Airfoils,’’ AIAA Paper 2005-0089, Jan. 2005.
Chapter 19
Role of Turbulence Modeling in Flow Prediction of Circulation Control Airfoils Gregory McGowan,* Ashok Gopalarathnam,+ Xudong Xiao,* and Hassan Hassans North Carolina State University, Raleigh, North Carolina
Nomenclature c = airfoil chord C, = pressure coefficient Cf = skin friction coefficient C, =jet momentum coefficients h = slot height k = turbulence kinetic energy riz = mass flow rate M = Mach number V = velocity aeff= effective angle of attack p = density w = turbulence frequency C = enstrophy Subscripts j =jet 00 = freestream conditions
*Research Assistant, Department of Mechanical and Aerospace Engineering. Student Member AIAA. 'Associate Professor, Department of Mechanical and Aerospace Engineering. Senior Member AIAA. 'Research Assistant, Professor, Department of Mechanical and Aerospace Engineering. Member AIAA. %Professor, Department of Mechanical and Aerospace Engineering. Fellow AIAA. Copyright 02005 by the authors. Published by the American Institute of Aeronautics and Astronautics, Inc., with permission.
499
500
G. McGOWAN ET AL.
I. Introduction HERE IS a continuing need to pursue technologies that can address longterm aerodynamic goals for future aircraft. ' For example, long-term visions for aeronautics predict the need and potential for a dramatic 50% reduction in fuel bum over the next 20 years, 36% of which is expected to come from improved aerodynamics. Other system studies by NASA Langley and Boeing2 have identified that the benefits of flow control are best realized by the development of simplified high-lift systems. Thus there is a long-term need for technology that can integrate the achievement of significant drag reduction (36%) at cruise with the achievement of very high lift for short-field operations. Circulation control (CC) is one type of flow control that has received considerable attention in recent years. This is because these systems offer the possibility of reduced takeoff and landing speeds, as well as increased maneuverability. The flow control is implemented by tangentially injecting a jet over a rounded wing trailing edge (TE). As a result of the balance between the pressure and the centrifugal force (the Coanda effect), the jet remains attached along the surface of the wing. Thus, the TE stagnation point is moved towards the lower surface, whereas the leading edge (LE) stagnation point is moved rearward, resulting in increased effective camber. This important area of research was the subject of a wellattended 2004 NASA/ONR workshop on circulation control3 that was held at NASA Langley Research Center in March 2004. A number of contributors used different turbulence models including algebraic, one-equation, two-equation, and stress models to try to predict flow characteristics of various CC airfoils. None of the models employed performed well for all jet momentum coefficients C, considered. The only exception is the Spalart-Allmaras model that includes curvature effects (SARC).4 However, the success of this model came as a result of adjusting5 one of the model constants, (c,3), which typically lies in the range of 0.6- 1.0, to 9.6. This adjustment has the effect of reducing the eddy viscosity throughout the flowfield and may change the character of the flow from turbulent to laminar or transitional flow over a large portion of the airfoil. The goal of this investigation is to consider the flow over the 103RE(103XW) CC airfoil tested by Abramson and Rogers.6 The tests were conducted to determine the performance characteristics of CC airfoils at transonic speeds. This airfoil was considered in Ref. 5 . Two turbulence models are employed in this investigation: the k - 5 model of Robinson and Hassan7 and the k-w model of Wilcox.* The latter model is included for comparison purposes because it yields results similar to the other turbulence models (other than SARC) in CFL3D.9 Both models were implemented in CFL3D (Version 5 ) . This version of CFL3D was modified to incorporate the k - 5 transitional/turbulence model of Warren and Hassan." The k - 5 model7," differs from other turbulence models used in Ref. 9 by the fact that it is derived by modeling the exact equations that govern the variance of velocity, or turbulence kinetic energy, k, and the variance of vorticity, or enstrophy, 5.As a result, the k - 5 model contains all the relevant physics in the k and 5 equations, is tensorially consistent, Galilean invariant, coordinate-system independent, and is free of wall or damping functions. It correctly predicts wall-bounded shear flows and the growth of all free shear layers" (jets, wakes
T
TURBULENCE MODELING IN FLOW PREDICTION
501
and mixing layers). According to Wilcox,* this is a minimum requirement for any turbulence model that is proposed for use in complex flows. It is to be noted that none of the turbulence models used in Ref. 9 satisfies the requirements suggested by Wilcox. The k - j transitional/turbulence model has the option to treat the flow in each block as laminar, transitional, or turbulent. The model requires that the transitional mechanism and freestream turbulent intensity be specified and is capable of predicting the onset and extent of transition. In this work, the transition over the external surface of the airfoil is deemed to be a result of the growth of Tollmien-Schlichting waves. The code has no transitional mechanism suited for internal flows, such as the cavity flow or subsonic nozzle employed here.
11. Formulation of the Problem A. Turbulence Models Most of the results obtained here employ the k - j turbulence model7 and the k- j transitional/turbulence model.’ The governing equations and boundary conditions are detailed in the cited references. Calculations are also presented for the k - w model.
B. Geometry and Grid The airfoil under consideration is elliptical in shape, has a chord of 1.5 ft., thickness ratio of 16%, and a camber ratio of 1%. The jet slot-height-to-chord ratio ( h / c )is 0.0021. The near-field of the medium-resolution grid is shown in Fig. 1. The fine grid has 235 points around the airfoil and 49 points in the normal direction over the forward part (block 2) and 101 points in the aft part (block 3) including points in the cavity (block 1). In the normal direction, the grid is clustered at the surface and y+ there is less than one. The total number of grid points is 70,563. For the medium grid the number of cells is halved in each coordinate direction. The grid is patched at the lower airfoil surface.
C. Numerical Procedure The numerical solution was computed using the code CFL3D.’ It is based on a finite volume method. The convective terms are approximated by upwind-biased spatial differences, and the viscous terms are discretized using central differences. In this work, the flux difference splitting of Roe is employed. Time integration is accomplished with an implicit approximate factorization scheme. The turbulence models are uncoupled from the mean flow equations. Their advection terms are discretized with first-order upwind differencing, whereas the source terms were treated implicitly. Characteristic-type boundary conditions are employed at inflow and outflow boundaries. For the plenum the mass flow rate and flow inclination angle are prescribed. At the surface of the airfoil, no-slip and adiabatic wall conditions are employed.
G. McGOWAN ET AL.
502
Fig. 1 Close-up of the grid employed.
111. Results and Discussion
A summary of the flow condition employed is given in Table 1, with C, defined as mj vj c -- 1/2pmv:c
where hj is the jet mass flux per unit span, vj is the jet velocity, pooand Vm are the freestream density and velocity, and Mj is the jet Mach number. The effective angle of attack a , was ~ determined by matching pressure coefficient distribution forward of midchord with a potential code that used CL and angle of attack as input^.^ Because the freestream temperature is constant, it is seen that C, is inversely proportional to the square of the freestream Mach number. This is why the C, values appear to be small for this case. Table 1 Summary of flow conditions employed for each case
301 302 306
0.0 0.0032 0.0110
0.0 0.519 0.979
- 0.0540 - 0.2865 - 0.7980
TURBULENCE MODELING IN FLOW PREDICTION
503
-0.8 -0.6
-0.4 -0.2
0
0"
0.2
0.4 0.6
0.8 1
0
0.25
0.5
0.75
1
XlC
Fig. 2 Comparison of C, using k-land k - o for Case 301, C, = 0.
As indicated below, grid refinement shows that results for the medium grid are identical to those for the fine grid. In spite of this, all results presented here employ the fine grid. Figure 2 compares calculated and measured pressure distribution in the absence of injection (Case 301). As is seen from the figure, both model predictions are in good agreement with experiment. Figure 3 compares predictions with experiment for Case 302. As is seen from the figure, the k - 5 turbulence model predictions are in better agreement with experiment than those given by the k-w model. The reason for this may be observed in Figs. 4 and 5 , which compare the streamline patterns in the injection region. As may be seen from the figures, the flow separation for the k-w model is delayed resulting in higher lift. Figure 6 presents calculated skin-friction coefficients using the k-5 model. The transitional behavior indicated in the figure is a result of a numerical transition. This is typical of all turbulence models. Figure 7 compares the pressure distribution for Case 302 using the k- 5 model on the fine and intermediate grids. It is seen that the solutions are grid independent. Figure 8 shows the calculated pressure distribution for Case 306 using the k-w model. We were unable to obtain a steady-state solution using the k - 5 model for this case. This can be seen from a plot of the residual indicated in Fig. 9. As may be seen from this figure, one can stop the solution earlier and obtain a rather reasonable solution or any solution desired depending on when the calculation is terminated. Because of the above behavior, a time-accurate solution was attempted. We were unable to detect a statistically steady solution even after
G. McGOWAN ET AL.
504 -0.8 -0.6
-0.4 -0.2
0
0"
0.2
0.4 0.6 0.8
1 0
0.25
0.75
0.5 X/C
1
Fig. 3 Comparison of C, using k-land k - o for Case 302, C , = 0.0032.
"
\
*
0.06 0.05 0.04 0.03 0.02 0.01 0 -0.01 -0.02 -0.03 -0.04 -0.05 -0.06 -0.07 -0.08
-0.09 0.95
1
1.05
X/C
Fig. 4 Streamline pattern around separation point ( k - 0 for Case 302, C, = 0.0032.
TURBULENCE MODELING IN FLOW PREDICTION
505
0.06 0.05
0.04 0.03 0.02 0.01
" . %
-0.01 -0.02 -0.03 -0.04 -0.05 -0.06 -0.07 -0.08 -0.09 0.95
1
1.05
X/C
Fig. 5 Streamline pattern around separation point ( k - o ) for Case 302, C, = 0.0032.
0.007 0.007
Case 302 upper surface lower surface
0.006 0.006
0.005
Cf
0.004 0.004
u-
0.003 0.003
0.002 0.001 0 0 0 0
0.2
0.4
0.6
0.8
11
x/c XlC
Fig. 6 Calculated skin friction using k-jmodel for Case 302, C, = 0.0032.
G. McGOWAN ET AL.
506 -0.8 -0.6 -0.4 -0.2
0.4 0.6 0.8 1
0
0.2
0.4
0.6
0.8
1
XlC
Fig. 7 Comparison of C, for k-jon fine and coarse grid for Case 302, C, = 0.0032.
-1.5
-1
-0.5
e 0
0.5
1
0
0.2
0.6
0.4
0.8
1
XlC
Fig. 8 Prediction of C, using k - o model for Case 306, C, = 0.011.
TURBULENCE MODELING IN FLOW PREDICTION
t
507
Case306
-4
-4.5
-5.5
-6 0
5000
10000
15000
20000
25000
Iteration
Fig. 9 Convergence history using k-jmodel for Case 306, C, = 0.011.
running the code for 28 periods. This suggests that a more elaborate approach, such as a large eddy simulation (LES)/Reynolds averaged Navier-Stokes (RANS), would be required. All of the above calculations assumed that the flow is fully turbulent. This is not necessarily the case. Attention was focused next on the use of the k- 5 transitional/turbulence model to analyze the flow for Case 306 because such a model will result in reduced eddy viscosity and earlier separation. The model, as coded in CFL3D, allows the user to specify laminar, transitional, or turbulent flow in each block. Further, it requires the user to specify the transitional mechanism and the freestream turbulence intensity. The transitional mechanism considered in the code is a result of the growth of Tollmien-Schlichting waves. This mechanism is not the correct mechanism for triggering transition in cavities. As a result, two cases were run. In the first, the flows in blocks 2 and 3 were specified transitional and turbulent, respectively, whereas the flow in the cavity was specified to be laminar. In the second case the flow in the cavity was assumed to be turbulent. It is seen from Figs. 10 and 11 that the results are dependent on whether the flow in the cavity is laminar or turbulent. Figures 12 and 13 show the streamline patterns in the injection region. They show that flow separation takes place earlier for the case where the flow in the cavity is laminar. This result explains why an increased value for the curvature parameter employed in Ref. 5 , which resulted in reduced eddy viscosity, gave good agreement with experiment.
508
G. McGOWAN ET AL.
Fig. 10 Prediction of C, using transitional k - j for Case 306, C, = 0.011, with laminar cavity.
X/C
Fig. 11 Prediction of C, using transitional k - j for Case 306, C, = 0.011, with turbulent cavity.
TURBULENCE MODELING IN FLOW PREDICTION
509
Y*
Fig. 12 Streamline pattern around separation point for Case 306, C, = 0.011, with laminar cavity. Case 306 turbulent cavity
Y*
I
I
I
1
I
I
I
I
I
I
I
L
1.05
XlC
Fig. 13 Streamline pattern around separation point for Case 306, C, = 0.011, with turbulent cavity.
510
G. McGOWAN ET AL.
IV. Conclusions The results presented in this work seem to suggest that the reason why existing turbulence models cannot predict flows over CC airfoils for high C , values is because the flow is not fully turbulent throughout. Use of a transitional flow is a more realistic representation of actual flows. The results further suggest that the flow in the cavity can have a major influence on the performance of CC airfoils. There is a need to develop new approaches to determine the effective Mach number and effective angle of attack for such flows. In addition, measurements other than the pressure distribution, such a velocity profiles and turbulent stresses are needed to further validate turbulence models. Acknowledgments The authors would like to acknowledge the assistance of Chris Rumsey of NASA Langley for providing us with the 103 RE (103 XW) airfoil, grid, input data, and advice. Thanks also to Charlie Swanson of NASA Langley for sharing with us the results presented at the October 2003 workshop. Further, the authors would like to acknowledge many helpful discussions during and after the workshop with Jane Abramson of the David Taylor Naval Ship Research and Development Center. References ‘Sellers, W. L., III, Singer, B. A., and Leavitt, L. D., “Aerodynamics for Revolutionary Air Vehicles,” AIAA Paper 2033-3785, June 2003. ’McLean, J. D., Crouch, J. D., Stoner, R. C., Sakurai, S., Seidel, G. E., Feifel, W. M., and Rush, H. M., “Study of the Application of Separation Control by Unsteady Excitation to Civil Transport Aircraft,” NASA CR 209338, June 1999. 3Jones, G. S., and Joslin, R. D. (eds.), Proceedings of the NASA/ONR Circulation Control Workshop, NASA CP-2005-213509, March 2004. 4Spalart, P. R., and Shur, M., “On the Sensitization of Turbulence Models to Rotation and Curvature,” Aerospace Science and Technology, Vol. 5, 1997, pp. 297-302. ’Swanson, R. C., Rumsey, C. L., and Anders, S. G., “Aspects of Numerical Simulation of Circulation Control Airfoils,” NASA/ONR Circulation Control Workshop, March 2004. 6Abramson, J., and Rogers, E. O., “High-speed Characteristics of Circulation Control Airfoils,’’ AIAA Paper 83-0265, Jan. 1983. Robinson, D. F., and Hassan, H. A., “Further Development of the k - l (Enstrophy) Turbulence Closure Model,” AZAA Journal, Vol. 36, No. 10 1998, pp. 1825-1833. 8Wilcox, D. C., Turbulence Modeling for CFD, 2nd ed., DCW Industries, Inc., La Canada, CA, 1998. ’ f i s t , S. L., Biedron, R. L., and Rumsey, C. L., CFL3D User’s Manual, NASA TM 1998-20844, June 1998. ‘%arren, E. S., and Hassan, H. A., “Transition Closure Model for Predicting Transition Onset,” Journal of Aircraft, Vol. 35, No. 5, 1998, pp. 769-775. “Robinson, D. F., Harris, J. E., and Hassan, H. A., “Unified Turbulence Closure Model for Axisymmetric and Planar Free Shear Flows,” AZAA Journal, Vol. 33, No. 12, 1995, pp. 2324-233 1.
’
1II.C. Tools for Predicting Circulation Control Performance: GACC Airfoil Test Case
Chapter 20
Simulation of Steady Circulation Control for the General Aviation Circulation Control (GACC) Wing Warren J. Baker* and Eric G. Patersont Pennsylvania State University, University Park, Pennsylvania
Nomenclature a = speed of sound, ft/sec CD = section drag coefficient, F d / ( l / 2 ) p U k S CL = section lift coefficient, F J ( 1/2)pUkS C, = pressure coefficient, ( p - p , ) / ( 1 / 2 ) p ~ k C , = jet momentum coefficient, ~ U ~ / ( I / ~ ) P U ; S c = chord length, in. h = slot height, in. k = turbulent kinetic energy (TKE), ft2/s2 1 = reference length used in defining velocity boundary condition riz = mass flow rate, lbm/sec M = Mach number, U / a p = pressure, lbm/ft2 ramp = Cubic polynomial used to accelerate the velocity amplitude from 0 to the final value after a nondimensional time of 1.0 Re = Reynolds number, pU,c/p s = planform area, ft2 U,V,W = velocity component in Cartesian coordinates, ft/s Upoly= tenth-order polynomial curve fit for defining velocity boundary conditions vjet= steady blowing jet amplitude, ft/s *Graduate Research Assistant, Department of Aerospace Engineering. Member AIAA. 'Senior Research Associate, Applied Research Laboratory, and Associate Professor of Mechanical and Nuclear Engineering. Member A I M . Copyright 02005 by Warren J. Baker and Eric G. Paterson. Published by the American Institute of Aeronautics and Astronautics, Inc., with permission.
513
W. J. BAKER AND E. G. PATERSON
514
x,y,z = Cartesian coordinates p = density, lbm/ft3
0, =jet angle applied for velocity boundary conditions at the jet-slot exit, deg w = specific dissipation rate, ft2/s3 Subscripts 03
= freestream
j = at jet-slot exit
I. Introduction HE CONCEPT of circulation control (CC) using the Coanda effect is a phenomenon involving a two-dimensional wall bounded jet passing along a curved surface. The jet itself is introduced via a slot, which expels the jet, typically, tangentially to the curved surface. This jet adds momentum to the boundary layer close to the curved surface. With the curved surface, the Kutta condition is not applicable, and the rear stagnation point is free to move. The resultant is a net change in the circulation, and the flow turning and separation location are altered based on the rate of mass addition. Accompanying the change in circulation are changes in certain aerodynamic values such as lift, total drag, and local skin friction coefficient. Figure 1 shows an example of a Coanda jet CC setup with a single slot. The performance benefits of CC have been shown in many experiments since the early 1 9 7 0 ~ . l Increases -~ in lift of as much as 10 times the typical flap system
T
Fig. 1 Trailing-edge Coanda jet.
STEADY CC SIMULATION FOR GACC WING
515
have been reported. Other possible benefits of the use of circulation control include elimination of moving parts, part/card decrease, significant weight decrease, and a less complex high-lift system. Circulation control is very attractive for certain naval applications, in particular the replacement of current actuation techniques on surface ship and submarine control surfaces with CC schemes. The current schemes, although robust and efficient, particularly for high-speed operation, have drawbacks. The force generated by a control surface is a function of lift coefficient, which in turn is a function of foil geometry, angle of attack, the square of the relative velocity of fluid over the control surface, and the fluid density. At very low vehicle speeds, the control surfaces may not provide sufficient control authority. Also, for marine applications, the density of water means that very large actuation forces and therefore complicated mechanisms must be created to move the control surfaces. Because of these drawbacks, and the desire in the submarine community for effective and safe low-speed littoral operations, there is motivation to develop alternative technologies for creating maneuvering forces. Circulation control schemes would provide very high lift at very low speeds, for example, in littoral operation or for evasive maneuvering, where the current control surface technologies are insufficient. The placement of a fixed control surface would increase shock resistance, allow placement of sensors or payload on the control surface, or even allow for the placement of the control surface in nontraditional areas previously restricted by the need for moving surfaces, such as on the outside of the propulsor duct. The long term objective of the present research is to develop validated simulation tools using multiple data sets. These data sets include a two-dimensional CC experiment using the NCCR 1510-7067N,* a low-aspect-ratio, tapered, control surface for marine applications, CCFOIL,3 and the General Aviation Circulation Control (GACC) wing4 the latter two of which are three-dimensional configurations. The work presented herein is the initial effort to investigate steady blowing CC of the GACC wing using the Reynolds-averaged NavierStokes (RANS) equations, and knowledge gained here will be combined with that from previous studies of the NCCR foil5 to continue to develop, validate, and verify our simulation tools for CC. The GACC was selected as a validation benchmark because it provides a modem experiment with computational fluid dynamics (CFD) validation in mind. Also, other CFD efforts have been initiated for the GACC, and both steady and pulsed actuation were used in experiment. The geometry itself has two slots (upper and lower) and has multiple trailing edge (TE) variants. 11. Geometry, Conditions, and Data
The GACC was tested in the Basic Aerodynamics Research Tunnel at NASA Langley Research Center. The GACC section is a modified General Aviation Wing-1, and is a supercritical 17% thick airfoil, with two slots. The chord length is 9.40 in. and the freestream velocity for experimentation is 110 fps yielding a chord Reynolds number of 5.33 x lo5 and a freestream Mach number M, of approximately 0.10. The upper slot is located at x/c =0.985 and the lower slot is located at x/c = 0.975. The slot height-to-chord ratio h/c, is approximately 0.00106. The circular TE has a radius-to-chord ratio Y/C of 2.00%. A crosssection of the model is shown in Fig. 2.
516
W. J. BAKER AND E. G. PATERSON UPPER STEADY MAMFoLo
ACTUAtOA 0mSm
UPPER SLOT
Fig. 2 Cross-section of the GACC wing.
A range of blowing coefficients were investigated, the highest being 0.162. Equation (1) provides a relation between the jet velocity and the nondimensional blowing coefficient. The code used for the current work is an incompressible code. To adjust for this, the density relations during experiment were acquired in order to obtain the proper jet velocity at the jet-slot exit. Therefore the maximum jet velocity corresponding to a C, = 0.162 is 917 fps and the nondimensional jet velocity Uj/UW = 8.34:
For all cases studied, the angle of attack was 0 deg. Experimental data are available for u er slot steady blowing, lower-slot steady blowing, and dualassist blowing. BPThe- most recent experimentation completed focuses on pulsed actuation, and initial data from pulsed testing are available.6 Table 1 summarizes the experimental data available. Experimental uncertainty has not yet been provided. Previous results from CFD simulations using the NASA Fully Unstructured Navier-Stokes 2D code (FLJN~D)’have been published? FUN2D uses the Spalart- Allmaras turbulence model, and all simulations completed assumed fully turbulent flow. A comparison to experiment of lift and drag data for a range of steady blowing coefficients has been presented? Two slot heights were used in simulations, 0.01 and 0.02 in. and results showed good trend agreement for the smaller of the two heights. Figures 3 and 4 show the lift vs. blowing coefficient curve and the drag polar for FUN2D simulations and experiment, respectively? 111. Computational Methods
The flow code used for the current work, CFDSHIP, is a general-purpose, parallel, unsteady, incompressible, RANS CFD code. The computational approach is
STEADY CC SIMULATION FOR GACC WING
517
Table 1 Available data from GACC experimentation Baseline (no jet actuation)
1) Surface pressure distribution 2) Lift-curve slope 3) Drag polar
Steady upper slot blowing
1) Surface pressure distribution (C, = 0.059 and 0.162) 2) Lift-curve slope ( C , = 0.007, 0.015, 0.025, 0.041, and 0.060) 3) Lift vs blowing coefficient (slot height = 0.01 in. and 0.02 in.) 4) Drag polar 5) Jet exit Mach number profiles (C, = 0-0.162) 6) Lift vs mass flow rate
Pulsed upper-slot blowing
1) Surface pressure distribution (CL= 1.2) 2) Lift vs mass flow rate
Steady lower-slot blowing
1) “Negative lift configuration,” lift vs blowing coefficient 2) “Negative lift configuration,” drag polar
Dual-slot assist steady blowing
1) Drag polar (slot height = 0.01 in. and 0.02 in.) 2) Drag polar (matched slot C , = 0.0, 0.004, 0.005, 0.009, 0.021, and 0.0041) 3) Drag vs angle of attack 4) Angle of attack vs L I D
Fig. 3 C, vs C , for previous CFD simulations and experiment!
518
W. J. BAKER AND E. G. PATERSON
Fig. 4 Drag polar for previous CFD simulations and experiment!
based upon structured, overset-grid, higher-order finite-difference, and pressureimplicit split-operator (PISO) numerical methods. Production turbulence model uses a linear closure and the blended k-w/k-E SST two-equation model.* Efficient parallel computing is achieved using coarse-grain parallelism via MPI distributed computing. For time-accurate unsteady simulations, global solution of the pressure-Poisson equation is achieved using preconditioned GMRES and the PETSc libraries. IV. Grid Generation Overset grids are generated primarily using hyperbolic extrusion and orthogonal box grids, although transfinite interpolation and elliptic smoothing of blocks can be used when needed. Overset interpolation coefficients are calculated and holes are cut using PEGASUS 5.1.9 CFDSHIP employs double-fringe outer and hole boundaries so that the five-point discretization stencil (i.e., in each curvilinear coordinate direction) and order of accuracy does not have to be reproduced near overset boundaries. The level-2 interpolation capability of PEGASUS 5.1 is used to achieve an optimal match between donor and interpolated meshes. Two grids were created initially for simulations. One grid included the upper plenum for modeling of the jet at the diffuser nozzle, whereas the second grid did not contain the plenum grid and modeled the jet at the orifice. The former of the grids is shown in Fig. 5, with block numbers noted. The domain size, as marked by the outermost boundaries of a nested orthogonal box grid shown as block 1, ranged from - 3 < x / c < 4, - 3 < y/c < 3, and 0 < z / c < 0.1. Near-wall
STEADY CC SIMULATION FOR GACC WING
519
a)
Fig. 5 Overset computational domain including the plenum: a) Overall view; b) foil view; c) plenum view.
W. J. BAKER AND E. G. PATERSON
520
spacing ranged between 2.00 x and 2.00 x The finer spacing was applied to all external surfaces to obtain proper resolution of the sublayer region of the turbulent boundary layer. The larger spacing was applied for the internal surfaces of the plenum, such that the boundary layers could be resolved properly. Two elliptically smoothed blocks span along the TE from upper to lower slot, denoted as blocks 6 and 7. Then, an O-grid was hyperbolically extruded around the body and split into four blocks, 2-5. A plenum block was created, block 8, and finally, an overset grid was placed along the knife edge of the upper slot, block 9, for investigation of the slot-lip interaction. The RANS simulations were performed in a pseudo-two-dimensional fashion, which requires five points in the spanwise direction. The grid consists of nine blocks containing a total of 394,665 points. Block sizes ranged from 31,000 to 61,000 points, with the plenum block having 33,000 points. The second grid, which does not include the plenum, totals eight blocks with 381,810 points. Only the TE view is shown in Fig. 6, because the computational domain is very similar to that shown in Fig. 5 in all regions except the TE. The difference in grid point number between the two grids is a result of the removal of a block and modifications to the near-wall spacing at the jet-slot exit to facilitate the applied boundary conditions. The block numbers coincide with those shown in Fig. 5 , excluding the plenum block. A three-point grid study was completed for uncertainty assessment. The previous grid without the plenum was used as the fine grid for the study. A
0.M 0
4.M
0.96
0.97
0.a
0.99 XlC
1
1.01
Fig. 6 Trailing-edge view of grid without plenum.
1.m
STEADY CC SIMULATION FOR GACC WING
52 1
Table 2 Total grid points for the fine, medium, and coarse grids Grid
Fine Medium
Coarse
Total grid points
381,810 193,980 97,575
J2 refinement process was completed to create a medium and coarse grid. This process was completed by decreasing the number of grid points by J2 in each of the x-and y-directions of the finest grid to create the medium grid. Therefore, the near-wall spacing applied for each of the fine, medium, and coarse grids was 2.00 x lop6, 2.83 x lop6, and 4.00 x lop6, respectively. Because of smoothing of some of the computational domain during grid creation, larger near-wall spacing occurred. This larger near-wall spacing occurred at the bottom slot and was 3.48 x lop6, 4.44 x lop6, and 5.67 x lop6 for the fine, medium, and coarse grids, respectively. The result of the refinement process is a reduction of grid points by a factor of approximately 1/2 from fine to medium grids. The same process is carried out to create the coarse grid from the medium. The coarse grid has approximately 1/2 the total grid points as the medium grid and approximately 1/4 the total points of the fine grid. Thus, from the fine to coarse grid, we have what is called “grid halving.” Table 2 shows the total number of grid points for the fine, medium, and coarse computational domains.
V. Initial and Boundary Conditions Initial conditions for the steady-state RANS simulations were prescribed to be equal to the freestream velocity, turbulence, and pressure:
where the subscript 00 refers to freestream conditions. No-slip boundary conditions were applied to the upper and lower surface of the airfoil, the round TE region, and the upper and lower surfaces of the plenum. For each grid, a different boundary condition was specified for the steady blowing. For all cases, the angle of attack was zero degrees. Figure 7 shows the location of the steady blowing boundary condition for the grid without the plenum. This occurs along the bottom portion of the jet slot. A no-slip condition is applied to the top portion of the jet slot. A velocity boundary condition is prescribed, and the velocity profile Upolyis a tenth-order polynomial curve fit of a typical CC jet profile seen in a previous RANS results for the GACC
W. J. BAKER AND E. G. PATERSON
522
OOSO
0-
0-
Dose
09s7
0.99s
Om
Id& Fig. 7 Boundary condition for grid without the plenum.
airfoil4 and is given by Eq. (3): Upoly= (- 1.2222 x 102*y/1'o)- (1.7043 x 102*y/Z9)
+ (1.8036 x 103*y/Zs)- (3.4603 x 103*y/17) + (2.9482 x 103*y/Z6)- (1.0602 x 103*y/15) - (9.7236 x 10'*y/Z4)+ (2.2944 x 102*y/13) - (8.5386 x 10'*y/Z2)+ (1.4472 x lO'*y/l) + 0.0036
(3)
where y/1 is the nondimensional distance along the boundary. To acquire tangential flow to the round TE, an initial angle of 6, = 18 deg was enforced. It was necessary to include this angle because the flow was modeled at the jet-slot exit. If the plenum flow had been modeled, proper jet attachment would have already been established at the location of the jet-slot exit. The velocity boundary condition for the grid without the plenum is given as U = vjet x ramp x cos (6) x UpOly V = vjet * ramp x sin (6) x Upoly
w=o where vjet is the velocity amplitude based on the blowing coefficient and Eq. (l), and ramp is a cubic polynomial used to accelerate the velocity amplitude from 0
STEADY CC SIMULATION FOR GACC WING
523
t
0.001
s t
0
~
.
'
.
0.5
~
.
'
U
'
'
1
~
.
'
'
"
.
L
15
Fig. 8 Velocity profile prescribed for steady blowing boundary condition at the jetslot exit.
to the final value after a nondimensional time of 1.0. The U-velocity profile for the boundary condition is shown in Fig. 8. The boundary condition for the grid with the plenum is less complex. Figure 9 shows the upstream face of the plenum where the steady blowing boundary condition is applied. In this case, a top-hat velocity distribution is used. Also, no additional flow angle is required to obtain tangential flow. The velocity boundary condition for steady blowing with the grid including the plenum is given in Eq. (7): U = yetx ramp
(7)
VI. Computational Resources All simulations were executed on an IBM SP Power 3 machine with 64 nodes. Each node contains sixteen, 375 MHz Power 3 processors. Each CPU has 64 kB level-1 cache and 8 MB level-2 cache memory along with 1 GB RAM.Each processor has a maximum sustainable performance of 1.5 GFLOPS, giving each node 24 GFLOPS peak performance. Scratch space available to users totals 3.2 TB (from ARL MSRC IBM SP Information, http://www.arl.hpc.mil/ userservices/ibm.html). As a reference point, a fine grid without the plenum completed 10,000 iterations (well past convergence for most simulations completed) in 16.7 wall-clock hours or 133.7 CPU hours. VII. Results This section presents the results from three separate studies. The first details the effects of modeling the Coanda jet vs resolving the internal plenum
W. J. BAKER AND E. G. PATERSON
524 om
om
OD1
0
am
om
M
3dc Fig. 9 Boundary condition for grid with plenum.
geometry. The second focuses on a performance assessment over a range of blowing coefficients. Thirdly, the results of a grid study to assess numerical uncertainty are reported.
Plenum vs No Plenum Steady RANS simulations of a baseline case at zero degrees angle of attack were initially completed for the two grids, with and without plenum. The goal was to determine the efficiency and accuracy for the no-blowing case, so as to choose the method to complete all following simulations. When both simulations were run to convergence (note that the plenum case is not shown to convergence for plotting purposes), results showed good agreement, as can be seen in Fig. 10, which shows the drag coefficient vs time-step number. To further illustrate the similarity in both solutions, total velocity contours with streamlines for the grid without the plenum and the grid with the plenum are shown in Fig. 11. Although both grids converge to a similar value of lift and drag, what is of importance is the total time to reach convergence. The case without the grid obtained a converged solution at around 5000 iterations, whereas the grid with the plenum is not yet completely converged at 20,000 iterations. Performance parameters such as drag are considered converged when the values differ by less than 0.02% of the previous value. Both grids had similar runtimes per iteration; thus, when calculating the computational costs, one sees at least four times the CPU runtime, and one extra CPU per simulation as a result of the A.
STEADY CC SIMULATION FOR GACC WING
525
Fig. 10 Convergence comparison of drag coefficient for grids with and without plenum.
added plenum block. The long time to reach convergence for the grid with the plenum is caused by a lengthy pressure transient inside the plenum along with continued slow pressure convergence throughout the simulation, even after the initial transients.
B. Performance Assessment for Varying Blowing Coefficient The fine grid without the plenum was chosen for further simulations. A wide range of blowing coefficients was studied, and results were compared to experiment and FUN2D simulations. Experimental data included the surface pressure distribution for C, = 0.059. The corresponding results from CFDSHIP are compared to experiment, and are shown in Fig. 12. The simulation compares well to experiment over the leading 95% of the airfoil. Simulation underpredicts the magnitude of the maximum positive pressure by a factor of 2 and over predicts the maximum negative pressure by a factor of 1.5. These locations correspond to the two slot locations at x / c = 0.975 and 0.985, respectively. More investigation needs to be carried out to further understand the discrepancy, and it must be noted that experimental uncertainty is high in these regions because of slow pressure leaks during e~perimentation.~ A plot of mean lift coefficient vs blowing coefficient is shown in Fig. 13. CFDSHIP fine grid results are compared to experiment and FUN2D solutions. The plot shows very good agreement of CFDSHIP results with experiment and FUN2D results for C, I0.091. At higher values of C, where no experimental data have been recorded, the results vary from FUN2D solutions. The variations in FUN2D and CFDSHIP resuls at the highest blowing coefficient are observable
526
W. J. BAKER AND E. G. PATERSON
XlC
xk
Fig. 11 Total velocity contours and streamlines of the baseline case for computational domains a) without the plenum and b) with the plenum.
by investigating the total velocity contours, shown in Figs. 14 and 15, respectively. FUN2D simulations predict the separation at the lower slot, whereas CFDSHIP predicts the location of separation on the bottom side of the airfoil back upstream at about 50% chord, as shown by the streamtraces in Fig. 15. Initially it may seem that the CFDSHIP results are “unphysical.” Yet, the phenomenon in which the jet reattaches and travels further up towards the leading edge (LE) has been observed in experiment, and has been called the “drawdown e f f e ~ t ” Until . ~ more experimental data are obtained, it is difficult to know which of the FUN2D and CFDSHIP simulations is more accurate.
STEADY CC SIMULATION FOR GACC WING
527
X Fig. 12 Surface pressure distribution for experiment and simulation, C, = 0.059.
Fig. 13 Lift vs C, for experiment and simulations.
528
W. J. BAKER AND E. G. PATERSON
Fig. 14 Mach contours for FUN2D simulations with C, = 0.162:
Fig. 15 Total velocity contours for CFDSHIP simulations with C, = 0.162.
STEADY CC SIMULATION FOR GACC WING
529
Figure 16 shows the time history of the lift and drag coefficient for a wide range of blowing coefficients. For C, 5 0.031, forces converge to a single value. For larger blowing coefficients, forces begin to oscillate. As the blowing coefficient increases, the amplitude of the oscillations increases, and the wavelength of the oscillation increases. Figure 17 shows that the surface pressure
NarrdhrerrpknalThre
Fig. 16 a) Lift force and b) drag force histories for a wide range of C,.
530
W. J. BAKER AND E. G. PATERSON
x/c Fig. 17 Surface pressure at different intervals over one oscillation for C, = 0.162.
changes quite a bit along the TE over one oscillation for C, = 0.162. Other blowing coefficients not illustrated in this work, C, 2 0.041, show similar trends. This may explain the significant changes in the forces. The turbulent kinetic energy is shown in Fig. 18 for low, moderate, and high blowing coefficients. For the lowest blowing coefficient, C, = 0.021, there exist two definitive regions of increased turbulent kinetic energy (TKE). The first, denoted as a) in Fig. 18 is the interaction of the jet shear layer and the incoming boundary layer from the top half of the airfoil beginning just aft of the jet orifice and terminating at the jet separation. The second region of high TKE denoted as b) in Fig. 18, originates near the jet separation and protrudes into the wake. At the moderate blowing coefficient, C, = 0.059, the same interaction of the jet shear layer and shear layer from the top half of the airfoil is observed, a) a smaller second region of high TKE (hard to see in the figure) arises from the interaction of the jet passing around the bottom corner of the slot and the recirculation zone located along the inside comer of the bottom slot b). For the highest blowing coefficient, C, = 0.091, a) is the same as the previous two blowing coefficients, and the second region of high TKE originates at the location of jet reattachment past the bottom slot b).
C. Grid Study A three-point grid study was completed for verification of results. Table 3 shows grid size and runtimes for each of the three grids used in the study. These values coincide with non-time-accurate RANS simulations of 10,000 iterations for each grid. The blowing coefficients used in the earlier work were now
STEADY CC SIMULATION FOR GACC WING
531
i)
ii)
iii)
Fig. 18 Turbulent kinetic energy for selected C,: i) C, = 0.021; ii) C, = 0.059; iii) C, = 0.091.
W. J. BAKER AND E. G. PATERSON
532
Table 3 Grid size and runtime characteristics for grid study
Grid points Seconds/time step Wall-clock hours CPU hours
Coarse
Medium
Fine
97,575 1.o 3.6 29.1
193,980 2.8 9.9 79.1
381,810 6.5 16.7 133.7
investigated using the coarse and medium grids, and results were compared to each other and experiment. Figure 19 shows a plot of the mean lift coefficient vs. blowing coefficient for the three grids studied. All three grids show agreement to experiment for lower values of lift increment gain. At higher lift gain, the coarse and medium results differ from the fine-grid results. It was determined that coarse and medium grids were of inadequate fidelity to capture the Coanda jet physics properly, in particular, the location of separation of the Coanda jet because of insufficient near-wall spacing, which caused inaccuracies in the prediction of the TKE in the buffer layer. To illustrate this point, surface pressure plots for three blowing coefficients, C, = 0.021, 0.059, and 0.091, are shown in Fig. 20. These three cases coincide with instances in which all three results show similar lift values (C, = 0.021), when the coarse result differs from the fine and medium results (C, = 0.059), and when the coarse and medium results differ from the fine result (C, = 0.091). The surface pressure distributions look similar for all three grids for C, = 0.021, and thus the similar lift predicition is feasible. For C, = 0.059, the “drawdown effect” introduced
d
c, Fig. 19 Lift vs C, for experiment and simulations for grid study.
STEADY CC SIMULATION FOR GACC WING
533
a)
Fig. 20 Surface pressure plots for coarse, medium, and fine grids at varying blowing coefficients: a) C, = 0.021; b) C, = 0.059; c) C, = 0.091.
W. J. BAKER AND E. G. PATERSON
534 a)
Fig. 21 Plots of y + for coarse, medium, and fine grids at varying blowing coefficients: a) C , = 0.021; b) C , = 0.059; c) C , = 0.091.
STEADY CC SIMULATION FOR GACC WING
535
a)
Fig. 22 Velocity contours for a) coarse, b) medium, and c) fine grids with C, = 0.59.
536
W. J. BAKER AND E. G. PATERSON
previously is visible for the coarse grid, distinguishable by the pressure drop along the pressure side of the wing along the aft 20% and the effects on the LE of the airfoil. The same characteristics are seen for the coarse and medium grids when C, = 0.091. This drawdown effect explains the underprediction of the lift forces. Figure 21 shows plots of y+ for the three grids and the previous values of blowing coefficient, C, = 0.021, 0.059, and 0.091. For the plot with C, = 0.021, all three grids show acceptable near wall resolution, y+ of approximately 1.00. For C, = 0.059, the coarse grid shows a y+ value much larger than 1.00 at both the upper and lower slots. For C, = 0.091, both the coarse and medium grids have y+ values much larger than 1.00 at the lower slot. The lower slot is an important location on which to focus, because the flow can either reattach aft of the slot or stay separated. The importance of the aft slot is demonstrated in Fig. 22, which shows total velocity contours and streamlines for the three grids with C, = 0.059. Here, the “drawdown effect” is visible for the coarse grid, marked by the reattachment of the flow ahead of the lower slot. The medium and fine grids do not show the drawdown. Recall that it was only the coarse grid in which the y+ value was greater than 1.00. It is not presented in this work, but for C, = 0.091, both the coarse and medium grids show this jet reattachment ahead of the lower slot, whereas the fine grid does not. To sum up the results from the grid study, the coarse, medium, and fine solutions show monotonic divergence. Determining the proper near-wall spacing for CC problems is an issue. Typically a flat plate approximation is used when determining near-wall spacing during grid creation. Adjustments need to be made to account for the highly curved surfaces. In this case, the flat plate approximation based on Reynolds number yielded a near-wall spacing of 2.00 x For the fine grid, near-wall spacing was set at 2.00 x lop6, with the coarsegrid near-wall spacing set at 4.00 x lop6.Even with the increased near-wall fidelity chosen, the medium and coarse grids proved to be deficient at higher blowing coefficients. Without a method to determine proper near-wall spacing requirements for CC applications, result validation becomes laborious and ineffective with time and computational resources. A better technique needs to be developed to determine CFD uncertainty for CC problems, ideally a single grid error estimation process.
VIII. Conclusions The GACC wing was studied using non-time-accurate, RANS CFD. With careful consideration, computational runtime could be decreased by modeling the jet at the orifice instead of including the plenum and modeling the jet at the diffuser nozzle exit, as shown in Figs. 7 and 9, respectively. After choosing the most efficient and accurate grid, a study of the mean forces on the airfoil for a wide range of blowing coefficients was completed, and results showed good agreement with experiment and previous RANS efforts using FUN2D for blowing coefficients C, 5 0.091. For higher blowing coefficients, where no experimental data are provided, CFDSHIP results differed from FUN2D results. CFDSHIP simulations showed the presence of unsteady flow, perhaps caused by the jet separation and interaction with the wake. A grid study was performed to verify results, but showed monotonic divergence from the coarse
STEADY CC SIMULATION FOR GACC WING
537
to fine grid solutions. Both the medium and coarse grids had insufficient near-wall spacing along the lower jet-slot, which affected the separation characteristics. Future work includes recreating the grid to add in the tunnel walls and optimizing the near-wall spacing. This will determine what effects the interaction between the wake and the tunnel walls have on the source of unsteadiness. Some early indication from experiment is that there was interaction between the wake and tunnel walls, but no quantitative value could be given yet. Other means to address this include using time-accurate RANS to investigate whether the oscillations shown are a product of the computational model, that is, the large domain, or a result of non-time-accurate simulations.
Acknowledgments The authors acknowledge the support of the Advanced Submarine Systems Development Office of the Naval Sea Systems Command, SEA 073R (Program Manager; Meg Stout) in the form of a graduate student fellowship for the first author, and the Office of Naval Research through Grant Number N00014-03-1-0122 (Program Officer; Ron Joslin) for the second author. Also, the authors would like to acknowledge the DoD High Performance Computing Modernization Office (HPCMO) and Army Research Laboratory-Major Shared Resource Center (ARL-MSRC) for providing computing resources through a DoD HPCMO Challenge Project. References ‘Englar, R. J., Stone, M. B., and Hall, M, “Circulation Control-An Updated Bibliography of DTNSRDC Research and Selected Outside References,” DTNSRDC Rept. 77-0076, Sep. 1977. ’Abramson, J., “Two-Dimensional Subsonic Wind Tunnel Evaluation of Two Related Cambered 15-Percent Circulation Control Airfoils,” DTNSRDC ASED-373, Sept. 1977. 3Rogers, E. O., and Donnelly, M. J., “Characteristics of a Dual-Slotted Circulation Control Wing of Low Aspect Ratio Intended for Naval Hydrodynamic Applications,” 42nd AIAA Aerospace Sciences Meeting & Exhibit, AIAA Paper 2004-1244, Jan. 2004. 4Jones, G. S., Viken, S. A., Washburn, L. N., Jenkins, L. N., and Cagle, C. M., “An Active Flow Circulation Controlled Flap Concept for General Aviation Aircraft Applications,” AIAA Paper 2002-3157, Jan. 2002. ’Paterson, E. G., and Baker, W. J., “Simulation of Steady Circulation Control for Marine-Vehicle Control Surfaces,” 42nd AIAA Aerospace Sciences Meeting and Exhibit, AIAA Paper 2004-0748, Jan. 2004. 6Jones, G. S., and Engle, R. J., “Advances in Pneumatic-Controlled High-Lift Systems Through Pulsed Blowing,” 21st Applied Aerodynamics Conference, AIAA Paper 2003341 1, June 2003. ’Anderson, W. K., and Bonhaus, D. L., “An Implicit Upwind Algorithm for Computing Turbulent Flows on Unstructured Grids,” Computers Fluids,Vol. 23, No. 1, 1994, pp. 1-21. ‘Menter, F., “Two-Equation Eddy Viscosity Model for Engineering Applications,” AIAA Journal, Vol. 32, No. 8, 1994, pp. 1598-1605. ’Suhs, N., Dietz, W., Rogers, S., Nash, S., and Onufer, J. T., “PEGASUS User’s Guide Version 5.lg,” Tech. Rept., NASA, May 2000.
Chapter 21
Computational Study of a Circulation Control Airfoil Using FLUENT Gregory McGowan* and Ashok Gopalarathnamt North Carolina State University, Raleigh, North Carolina
Nomenclature A = area b = wing span c = chord Cd = drag coefficient Cl = lift coefficient C, = pitching moment coefficient about quarter chord C, = momentum coefficient h = slot height M = Mach number riz = mass flow rate P = pressure q = dynamic pressure R = gas constant for air r = radius of Coanda surface Re = Reynolds number s = arc length, measured from the slot exit around the upper surface of the airfoil T = temperature U = velocity magnitude w = slot width, equal to b for two-dimensional flows a = angle of attack y = ratio of specific heats p = viscosity coefficient *Graduate Research Assistant, Department of Mechanical and Aerospace Engineering. 'Associate Professor, Department of Mechanical and Aerospace Engineering. Copyright 02005 by the authors. Published by the American Institute of Aeronautics and Astronautics, Inc., with permission.
539
540
G. McGOWAN AND A. GOPALARATHNAM
p = density
Subscripts duct = stagnation conditions inside plenum fc = conditions at flow-control boundary 03 = freestream conditions J = slot-exit conditions
I. Introduction ECENT research in the Applied Aerodynamics Group at the North Carolina State University (NCSU) has led to the development of an automated cruiseflap system.lq2The cruise flap, introduced by P f e n n i ~ ~ g eisr ,a~ small ~ ~ trailingedge (TE) flap that can be used to adapt an airfoil and increase the effective size of the low-drag range of natural-laminar-flow (NLF) airfoils. The automation is achieved by indirectly sensing the leading-edge (LE) stagnation-point location using surface pressure measurements and deflecting the flap so that the stagnation-point location is maintained at the optimum location near the LE of the airfoil. Maintaining the stagnation point at the optimum location results in favorable pressure gradients on both the upper and lower surfaces of the airfoil. With such a cruise-flap system, the airfoil is automatically adapted for a wide speed range. This automated cruise-flap system was successfully demonstrated in the subsonic wind-tunnel at NCSU.2 Although the use of a cruise flap on an NLF airfoil results in low drag over a large range of flight speeds, there is a need for a revolutionary approach that integrates the achievement of significantly lower drag over a large range of operating speeds with the capability for generating very high lift at takeoff and landing conditions. Toward this objective, it is of interest to study an approach that integrates aerodynamic adaptation with the well-established high-lift capability of circulation control (CC) aerodynamics. Circulation control is not a new concept; it has been around since the late 1930s. The majority of research efforts have focused on blowing in a positive, or downward, direction at the TE of the airfoil. Early efforts accomplished this downward inclination using a jet of high-velocity air blown straight out of the TE at the desired angle.5 This pneumatic-flap concept has been studied theoretically and experimentally by several researchers over the past several decades?-'' As time has progressed, more researchers have begun to take advantage of the Coanda by blowing over a round TE. This Coandabased CC is currently attracting significant interest as a means of achieving high lift. This aerodynamic adaptation, when achieved using a blown cruise flap, carries with it the potential for significant skin-friction drag reductions through extensive laminar flow in addition to the high-lift benefits of CC aerodynamics. Figure 1 illustrates the overall concept. In a manner similar to that of a cruise flap, it is believed that by utilizing this stagnation-point sensing scheme, an adaptive CC airfoil, with a blown cruise flap, can achieve extensive laminar flow over a large lift-coefficient range. As a first step toward the long-term goal of studying an adaptive CC airfoil, the current effort was undertaken for establishing and validating computational
R
STUDY OF CC AIRFOIL USING FLUENT
541 high-speed cruise condition
Fig. 1 Illustration of the NCSU concept of an adaptive CC airfoil.
fluid dynamics (CFD) analysis procedures for blown-TE airfoils. The CFD package used for this work was the FLUENT flow solver. The results are compared to CFD and experimental data obtained from a recent study by Jones et al? of a General Aviation CC (GACC) airfoil conducted at the NASA Langley Research Center. Because previous CFD studies on this airfoil did not include tunnel walls, the current CFD study also includes an investigation of the effect of tunnel walls on the solution. To provide a foundation for the adaptive CC airfoil concept, the effects of CC on the LE stagnation-point location were also examined in the current work. The following section gives an explanation of the geometry under examination and information about the experimental setup. Then a description of the numerical approach is presented, including grid details, solver settings, and boundary conditions. Results are then presented for two cases: 1) solution in free-air or the infinite domain, and 2) solution with the presence of windtunnel walls. Results are also shown for a stagnation point study, in an effort to show how the stagnation point moves with changing blowing rates.
11. Configurations and Experiments The geometry chosen for the current research was the GACC airfoil, designed by Jones.15 The GACC airfoil was derived from a 17% GAW( 1) airfoil by modifying the TE to incorporate a 2% r / c Coanda surface and is shown in Fig. 2.
Fig. 2 General Aviation Circulation Control (GACC) airfoil geometry used in the current research.
542
G. McGOWAN AND A. GOPALARATHNAM
The wind-tunnel experiments were conducted by Jones et al.15 in the Basic Aerodynamic Research Tunnel (BART),which is located at the NASA Langley Research Center in Hampton, Virginia. The BART tunnel has a physical testsection size of 28 x 40 x 120 in. The GACC model chord length was 9.4 in., with angle of attack changes made about the half-chord location. Further details of the experimental setup are given in Ref. 16. 111. Numerical Approach
The commercial flow-solver code FLUENT version 6.1 was used in the current research. Grid generation was performed using GAMBIT, which is the preprocessor packaged with the FLUENT code. These codes were used to study two cases. The first case involves the examination of the GACC airfoil in free air with the objective of comparing the FLUENT two-dimensional results to CFD and wind-tunnel results presented in Ref. 15. It should be noted that the CFD solutions obtained in Ref. 15 did not include the effect of windtunnel walls. The second case involves two-dimensional simulations of the GACC airfoil in the BART facility to examine the influence of tunnel walls on this particular airfoil. Results from FLUENT were obtained for a matrix of 15 data points for each of the two cases.
A. Grid Details For the first study, a circular computational domain (Fig. 3) was generated that extends to approximately 20 chord lengths in all directions and is composed of 132,762 cells. For the study of wall effects, a second two-dimensional grid was generated to include the wind-tunnel upper and lower walls and is shown in Fig. 4. For the computation with walls, a separate grid was generated for each angle of attack, each of which comprises 123,602 cells and extends to 20 chord lengths upstream and downstream of the airfoil. The grids for all of the analyses are hybrid unstructured grids. The domains consist of an unstructured grid far from the airfoil in order to reduce the number of cells and a structured grid near the airfoil to maintain good
Fig. 3 Grid generated for the free-air analyses using FLUENT.
STUDY OF CC AIRFOIL USING FLUENT
543
Fig. 4 Grid generated for FLUENT study of wall effects.
resolution through the boundary and shear layers. For both cases, minimum wall spacing was chosen such that y+ < 1 at the wall.
B. Solver Settings For the current study the solution is assumed to be steady and is not run timeaccurate. The coupled-implicit solver was chosen with second-order upwind node-based discretization for both the flowfield and turbulence equations. The coupled solver was chosen for two reasons. First, compressibility effects need to be modeled, because the Mach number at the slot exit can often approach the sonic condition as the blowing rate is increased. Secondly, the FUN2D1' code has a compressible solver, and because the results from the current study were compared with FUN2D results, a compressible solver was also used for the FLUENT analysis. There was an attempt to run these problems with the segregated (decoupled) solver using very low relaxation factors; however, it was found that for the cases with larger blowing rates, the solution began to exhibit an unsteady effect after a few thousand iterations. In order to compare with the FUN2D results of Ref. 15, the one-equation Spalart- Allmaras turbulence model was chosen for the current work. Wall functions were not used in the FLUENT calculations. C. Boundary Conditions FLUENT does not allow the user to input the freestream Mach number and Reynolds number directly. Instead, the freestream velocity and operating pressure were calculated using Eqs. (1-3) and provided as inputs for the analyses. The Mach and Reynolds numbers were set to 0. l and 533,000, respectively, to match those used in Ref. 15. The results were used for both cases, with and without tunnel walls: Uw = MwJ3/RTm RePW Pw = uwc
An approximate method was developed to estimate the velocity required at the This method flow control boundary ( U f c )to achieve a desired C,, CPdesired. assumes incompressible flow throughout the duct, and was derived by solving the continuity equation. The equation for Ufc from this approximate method is
544
G. McGOWAN AND A. GOPALARATHNAM
given in Eq. (4):
Once FLUENT converged, an integration was performed across the slot exit as shown in Eq. ( 5 ) to obtain the actual C, of the jet at the slot. This C,, however, is because the Ufc for the latter is set using an approximate different from CPdeslred method.
Furthermore, to be consistent with the methods used for calculating C, in Ref. 15, all of the C values presented in this paper were calculated using isentropic flow relations? The equations for this procedure are given in Eqs. (6-8). To determine how close the isentropic C, is to the integrated C,, the two values are compared in Fig. 5 for several cases. The C, values indicated along the horizontal axis are values calculated using the isentropic relations. Values for C, on the vertical axis were computed by integrating the flow across the slot exit. The solid line in Fig. 5 indicates where the data points would lie if the two methods generated the same values for C,. The symbols are representative of the actual values calculated using FLUENT and isentropic relations. Although the differences are very small, approximately 3% at the highest blowing coefficient, care must be taken to ensure consistency in the CFD solutions and experiments: riZ = PJUJAJ
(6)
Fig. 5 Comparison of Cphkgm,d with CpbeotroPic for a = 0; the straight line is included to indicate deviation from a perfect correlation.
STUDY OF CC AIRFOIL USING FLUENT
545
IV. Results The results from FLUENT predictions for the GACC airfoil are presented in three parts. In the first part, the prediction for the GACC airfoil in freeair conditions is compared with the results presented in Ref. 15. In the second part, the predicted results for the GACC airfoil with tunnel walls are presented and compared with the free-air results. In the third part, the effects of a and C, on the LE stagnation-point location are presented and discussed. A. Results for Free-Air Conditions In this part of the study, FLUENT results for free-air conditions are compared with CFD and experimental results from Ref. 15. The overall comparison between the FLUENT results and experimental results is illustrated using Cl-a curves in Fig. 6. The results from FLUENT analyses consist of a matrix of 15 data points for a = - 5 , 0, and 5 deg and C, = 0, 0.008, 0.024, 4 Experimental Results Curve Fit to Experimental Data
Fluent Calculations
3.5 3.5
←Cµ= 0.060
3 3
Cµ= 0.078 → ←Cµ= 0.041
2.5 2.5 C CIl
←Cµ= 0.025
2 Cµ= 0.047→ ←Cµ= 0.015 ←Cµ= 0.007 ←Cµ= 0.0
1.5 1.5
11
0.5 0.5
Cµ= 0.024→
Cµ= 0.008 →
0 0 −10 -1 0
← Cµ= 0.0
−5 -5
0 5 0 5 α a (degrees)
10 10
15 15
Fig. 6 Comparison of NCSU FLUENT results from the current work with Langley experimental results from Ref. 15 (data points and curve fits for each Cp).
546
G. McGOWAN AND A. GOPALARATHNAM
Table 1 FLUENT results for the free-air cases Lift coefficient, CL Blowing coefficient (C,)
0.000 0.008 0.024 0.047 0.078
a = -5 deg
a=Odeg
a = 5 deg
0.090 0.382 1.082 1.979 3.045
0.666 1.009 1.646 2.544 3.206
1.193 1.486 2.080 2.7 19 3.296
0.047, and 0.078, and are presented in Fig. 6 using solid lines and square markers. The FLUENT data used to generate Fig. 6 are given in Table 1. The wind-tunnel results from Ref. 15. are presented as circular markers with the dashed lines in Fig. 6 representing best-fit curves for several angles of attack and for C, = 0, 0.007, 0.015, 0.025, 0.041, and 0.060. The values of C, for the FLUENT results differ from those for the results of Ref. 15 because of the difference between the actual C, and the desired C? when using the approximate method in Eq. (4) for estimating the Ufc using mcompressible-flow equations. Although the values of C, for the FLUENT results do not match those for the results of Ref. 15, it is clear that the trends and most of the predictions for the Cl are close to those from Ref. 15. In particular, the FLUENT predictions for C, = 0, 0.008, and 0.047 agree quite well with the results for similar values of C, from Ref. 15. Two discrepancies between the FLUENT predictions and those from Ref. 15 are apparent: 1) for C, = 0.024 and 2) for C, = 0.078. The reason for the first discrepancy in the results is attributed to the incorrect prediction of the jet-separation location on the Coanda surface for C, = 0.024. The apparent discrepancy in the results for C, = 0.078 is attributed to nonlinear effects at the high blowing rates and the fact that the highest blowing rate in the results of Ref. 15 is for C, = 0.060. The flowfield data for the FLUENT results are presented in two parts. In the first part, the effects of increasing C, for a constant angle of attack are presented. The second part examines the effects of angle-of-attack changes and their influence on the CC airfoil for a constant C,. The flowfield data are presented as streamline plots; these serve as visual aids in the understanding of the effects of CC on the flow over the airfoil. The first part of the flowfield data is shown in Figs. 7a-7c. It can be seen that as the blowing rate is increased the streamlines become more curved-an indication of increased circulation. The second part of the flowfield data is shown in Figs. 8a-8c and Figs. 9a-9c to illustrate the effects of changing the angle of attack while holding blowing rates constant. The results are presented for two blowing rates: the mild blowing case C, = 0.047 and the highest blowing rate C, = 0.078. The results show that changes to C, have a significant effect on the jet-separation location and the resulting Cl. In comparison, changes to a have a much smaller effect on the jet-separation location.
STUDY OF CC AIRFOIL USING FLUENT
547
Fig. 7 Circulation control effects on the flowfield at a = 0 deg for various values of C,: a) C, = 0.000; b) C, = 0.047; c) C, = 0.078.
B. Wind-Tunnel Wall Effects In this subsection, the FLUENT results for the GACC airfoil with the effect of wind-tunnel upper and lower walls are presented. Figures 10 to 12 show the influence of the walls on the CFD solution. These figures present the predicted Cl as a function of C, for a = 0, 5 , and - 5 deg, respectively. Figure 10 also includes a
548
G. McGOWAN AND A. GOPALARATHNAM
a)
Fig. 8 Circulation control effects on flow field at C, = 0.047 for various values of a: a) cu = -5 deg; b) cu = 0 deg; c) cu = 5 deg.
comparison with results for experiment and the FUN2D study15 for a = 0 deg, the only angle of attack for which the FUN2D results were presented in Ref. 15. The FUN2D simulations in Ref. 15 did not include any wind-tunnel wall effects. Figures 10-12 indicate that the presence of walls has very little influence on the CFD solution. Because the study was performed on a two-dimensional grid, it can be stated that blockage effects are minimal;
STUDY OF CC AIRFOIL USING FLUENT
549
Fig. 9 Circulation control effects on flowfield at C, = 0.078 for various values of (Y: a) (Y = -5 deg; b) (Y = 0 deg; c) (Y = 5 deg.
however, no conclusion can be drawn for the three-dimensional effects due to side-wall boundary layer effects and the associated trailing vortices. Because of the large lift that these configurations produce, it is believed that threedimensional effects will be extremely important at larger blowing rates. The results for the with-walls simulations consistently show that for low blowing coefficients, the Cl values are predicted to be lower than those for the
G. McGOWAN AND A. GOPALARATHNAM
550
4 Fluent (with walls) Fluent (free−air) FUN2D Jones et al. (free−air) Experiment Jones et al.
3.5
3
2.5 Cl 2
1.5
1
0.5 0
0.02
0.04
0.06
0.08
0.1
Cµ
Fig. 10 FLUENT prediction of wind-tunnel wall effects for varying values of C, at a = 0 deg.
0
0.01
0.02 0.03 0.04 0 i 0.06 0.07 C
c, Fig. 11 FLUENT prediction of wind-tunnel wall effects for varying values of C, at a = 5 deg.
STUDY OF CC AIRFOIL USING FLUENT
551
Fig. 12 FLUENT prediction of wind-tunnel wall effects for varying values of C, at a = -5deg.
free-air simulations. However, at the largest blowing coefficients, the trend reverses and Cl values with walls are predicted to be higher than those without walls. The FLUENT data accrued for the cases with wind-tunnel walls are given in Table 2.
C. Stagnation-PointLocation The motivation for examining the LE stagnation-point behavior is that the stagnation-point location was used successfully in earlier research',* for closed-loop control of a TE flap. It was, therefore, desirable to examine the CFD solutions for the CC airfoils to see if there was any evidence that would suggest that a similar approach could be extended for use with CC airfoils. Table 2 FLUENT results for cases with wind-tunnel walls Lift coefficient, CI Blowing coefficient
cc,,
0.000 0.008 0.024 0.047 0.078
a = -5 deg
a=Odeg
a = 5 deg
0.09 1 0.388 1.063 1.892 2.893
0.702 1.027 1.671 2.475 3.044
1.247 1.491 2.070 2.711 3.178
G. McGOWAN AND A. GOPALARATHNAM
552
1.15
.............. ..............
+c +C
= 0.000 = 0.008 ++ Cw= 0.024 +- C = 0.047 + C = 0.078 i
P //
SIC
I
..........................
1.OE
1
/
............................... .............. .............
1.1
0.5
1
..............
.............
'
Fig. 13 Circulation control effects on LE stagnation-pointlocation.
Stagnation-point location, measured as an arc length from the jet exit around the upper surface of the airfoil, as a function of Cl, is presented in Fig.13. Each line in Fig. 13 represents a different blowing rate and for each blowing coefficient there are three points that correspond to three different angles of attack (- 5, 0, and 5 deg). From Fig. 13 it can be seen that the stagnation point moves in a predictable manner, both with angle of attack and with changing blowing rate. This behavior provides an indication that the stagnation-point location can be used as a means to develop closed-loop control of the jet C, on CC airfoils.
V. Conclusions The results from a two-part CFD study using the FLUENT flow solver have been presented. Results of the first study show that, although the FLUENT predictions do not match the CFD and experimental results of Ref. 15 exactly, the overall trends are followed very closely. Throughout the range of blowing coefficients, with the exception of the no-blowing case (C, = O.O>, FLUENT consistently predicted a slightly lower overall lift coefficient.
STUDY OF CC AIRFOIL USING FLUENT
553
The second study focused on the influence of wind-tunnel walls on the CFD solution. For low blowing coefficients, it was found that the lift is predicted to be lower for the cases with walls. The trends are reversed for the higher blowing coefficients, for which the cases with walls yield a higher predicted lift. Although the solutions are different, the differences are small, and could as well be attributed to differences in the grids rather than the actual presence of walls. The influence of CC on the LE stagnation-point location was examined. It was shown that changes in blowing rate and angle of attack result in systematic changes to the stagnation-point location. This observation indicates that it is possible to use a closed-loop control system that is driven by sensing the stagnationpoint location.
Acknowledgments The authors would like to acknowledge the funding for this research through a grant from the NASA Langley Research Center and the National Institute of Aerospace. The technical monitor, Greg Jones of NASA Langley, is thanked for many valuable discussions and for the geometry of the GACC airfoil and the wind-tunnel test results. In addition, Greg Stuckert from FLUENT Inc. and Hassan Hassan of NCSU are thanked for their advice regarding the CFD simulations. References ‘McAvo~,C. W., and Gopalarathnam, A., “Automated Cruise Flap for Airfoil Drag Reduction over a Large Lift Range,” Journal of Aircraft, Vol. 39, No. 6, 2002, pp. 981 -988. *MCAVOY, C. W., and Gopalarathnam, A., “Automated Trailing-Edge Flap for Airfoil Drag Reduction Over a Large Lift-Coefficient Range,” AIAA Paper 2002-2927, June 2002. 3Pfenninger, W., “Investigation on Reductions of Friction on Wings, in Particular by Means of Boundary Layer Suction,” NACA TM 1181, Aug. 1947. 4Pfenninger, W., “Experiments on a Laminar Suction Airfoil of 17 Per Cent Thickness,” Journal of the Aeronautical Sciences, April 1949, pp. 227-236. ’Davidson, I. M., “The Jet Flap,” Journal of the Royal Aeronautical Society, Vol. 60, No. 1, 1956. %pence, D. A., “The Lift Coefficient of a Thin, Jet-Flapped Wing,” Proceedings of the Royal Society Series A, Vol. 238, No. 121, 1956. ’Spence, D. A., “Some Simple Results for 2-Dimensional Jet-Flap Aerofoils,” The Aeronautical Quarterly, 1958, pp. 395-406. ‘Garland, D. B., “Jet-Flap Thrust Recovery: Its History and Experimental Realization,” Canadian Aeronautics and Space Journal, May 1965, pp. 143-151. ’Lissaman, P. B. S., A Linear Solution for the Jet Flap in Ground Effect, Ph.D. Thesis, California Inst. of Technology, Pasadena, CA, 1965. “Aiken, T. N., and Cook, A. M., “Results of the Full-Scale Wind Tunnel Tests on the H-126 Jet Flap Aircraft,” NASA TN D-7252, April 1973.
554
G. McGOWAN AND A. GOPALARATHNAM
l 1Abramson, J., Rodgers, E., and Taylor, D., “High-speed Characteristics of Circulation Control Airfoils”, AIAA Paper 83-0265, 1983. ‘’Wood, N., and Nielsen, J., “Circulation Control Airfoils Past, Present, Future,” AIAA Paper 1985-0204, 1985. 13Novak,C. J., and Cornelius, K. C., “An LDV Investigation of a Circulation Control Airfoil,” AIAA Paper 86-0503, 1986. 14Novak,C. J., Cornelius, K. C., and Roads, R. K., “Experimental Investigations of the Circular Wall Jet on a Circulation Control Airfoil”, AIAA Paper 87-0155, 1987. 15Jones, G. S., Viken, S. A., Washburn, A. E., Jenkins, L. N., and Cagle, C. M., “An Active Flow Circulation Controlled Flap Concept for General Aviation Aircraft Applications,” AIAA Paper 2002-3 157, 2002. “kagle, C. M., and Jones, G. S., “A Wind Tunnel Model to Explore Unsteady Circulation Control for General Aviation Applications,” AIAA Paper 2002-3240, 2002. ”Jones, G. S . , and Englar, R. J., “Advances in Pneumatic-Controlled High-Lift Systems Through Pulsed Blowing,” AIAA Paper 2002-341 1, 2003. “Anderson, W. K., and Bonhaus, D. L., “An Implicit Upwind Algorithm for Computing Turbulent Flows on Unstructured Grids,” Computers & Fluids, Vol. 23, No. 1, 1994, pp. 1-21.
1II.D. Tools for Predicting Circulation Control Performance: Additional CFD Applications
Chapter 22
Computational Evaluation of Steady and Pulsed Jet Effects on a Circulation Control Airfoil Yi Liu,* Lakshmi N. Sankar,+ Robert J. Englar,' Krishan K. Ahuja,$ and Richard Gaetall Georgia Institute of Technology, Atlanta, Georgia
Nomenclature a = angle of attack
Ajet= area of jet slot, ft2 CL, Cl = lift coefficient C,, Cd = drag coefficient C, =jet momentum coefficient C, = averaged jet momentum coefficient f = pulsed jet frequency, Hz Lref= length reference, in. m =jet mass flow rate, slugs/s s = wing area, ft2 St = Strouhal number TJet, To,,,, = temperature and total temperature of the jet, K Pj,, = pressure of the jet, psia V , = freestream velocity, ft/s V,,, =jet velocity, ft/s pjet,p, = densities, slugs/ft3
*Research Scientist, National Institute of Aerospace. Member AIAA. 'Regents Professor, School of Aerospace Engineering. Associate Fellow AIAA. 'Principle Research Engineer, Georgia Tech Research Institute. Associate Fellow AIAA. %Professor, School of Aerospace Engineering. Fellow AIAA. TResearch Engineer, Georgia Tech Research Institute. Senior Member AIAA. Copyright 0 2005 by the American Institute of Aeronautics and Astronautics, Inc. All rights reserved.
557
558
YI LIU ET AL.
I. Introduction URING the past several decades, there has been a significant increase in air travel and a rapid growth in commercial aviation. At the same time, environmental regulations and restrictions on aircraft operations have become issues that affect and limit the growth of commercial aviation. In particular, the noise pollution from aircraft, especially around airports, has become a major problem that needs to be solved. Reducing aircraft noise has become a priority for airlines, aircraft manufacturers, and NASA researchers. In response to this challenge, NASA has proposed a plan to double aviation system capacity while reducing perceived noise by a factor of two (10 dB) by 201 1, and to triple system capacity while reducing perceived noise by a factor of four (20 dB) by 2025. Large commercial aircraft are dependent on components that generate high levels of lift at low speeds during takeoff or landing so that they can use existing runways. Conventional high-lift systems include flaps and slats, with the associated flap-edges and gaps, are significant noise sources. Since the mid-l980s, many researchers have pointed out that the airframe noise predominantly emanates from high-lift devices and the landing gear of subsonic air~raft.”~ Depending on the type of aircraft, the dominant source varies between flap, slat, and landing gear.4 Furthermore, these high-lift systems also add to the weight of the aircraft, and are costly to build and maintain. An alternative to the conventional high-lift systems is circulation control wing (CCW) technology. This technology and its aerodynamic benefits have been extensively investigated over many ears through experimental studies.596A limited number of numerical a n a l y ~ e $ ~have - ~ also been carried out. Work has also been done on the acoustic characteristics studies of CC wings.’ These studies indicate that very high CL values (as high as 8.5 at a = 0 deg) may be achieved with CCW. Because many mechanical components associated with the high-lift system are no longer needed, the wings can be lighter and less expensive to build. lo Major airframe noise sources such as flap-edges, flapgaps, and trailing/leading edge flow separation can all be eliminated with the use of CCW systems. Earlier designs of CCW configurations used airfoils with a large-radius rounded trailing edge (TE) to maximize lift production. These designs also produced very high drag.” Such high drag levels associated with a blunt, large-radius TE can be prohibitive under cruise conditions when CC is no longer necessary. To overcome this difficulty, an advanced CCW section, called a circulation hinged flap,596has been developed to replace the traditional rounded TE CC airfoil. This concept, originally developed by Englar, is shown in Fig. 1. The upper surface of the CCW flap is a large-radius arc surface, but the lower surface of the flap is flat. The flap could be deflected from 0 to 90 deg. When an aircraft takes off or lands, the flap is deflected as in a conventional high-lift system, and CC is deployed. The large curvature of the upper surface produces a large jet turning angle, leading to high lift. When the aircraft is in cruise, the flap is retracted and a conventional sharp TE shape results, greatly reducing the drag. This kind of flap does have some moving elements that increase the weight and complexity over the earlier CCW design. However,
D
EVALUATION OF STEADY AND PULSED JET EFFECTS Supsrwidcal Contour
559 CCW Flap
Fig. 1 Dual radius CCW airfoil with LE b l ~ w i n g . ~
overall, the hinged flap design still maintains most of the advantages of the CC, while greatly reducing the drag in cruising condition associated with the rounded TE CCW design. To understand and quantify the aeroacoustic characteristics and benefits of the CCW, Munro, Ahuja, and Englar12-15 have recently conducted several acoustic experiments comparing the noise levels of a conventional high-lift system with those of an advanced CC wing at the same lift setting. The present computational fluid dynamics (CFD) study16 is intended to complement this work, and numerically investigates the aerodynamic characteristics and benefits associated with this CC airfoil. Computational fluid dynamic studies such as the one presented here can also help in the design of future generation CCW configurations. The present work is an extension of a previous work where two-dimensional studies of the effects of steady and pulsed jets on the CCW configuration were carried 0 ~ t . The l ~ objective of this study is to isolate and quantify the effects of parameters such as leading edge (LE) blowing, freestream velocity, jet slotheight, and frequency on the performance of two-dimensional steady and pulsed CC jets. The unsteady Navier-Stokes methodology used here has also been applied to study a three-dimensional CC wing, and to model tangential blowing effects.16
11. Mathematical and Numerical Formulation A. Governing Equations In the present work, the Reynolds-averaged Navier-Stokes (RANS) equations were solved using an unsteady three-dimensional viscous flow solver. A semiimplicit finite-difference scheme based on the Alternating Direction Implicit (ADI)18,19method was used. This scheme is second- or fourth-order accurate in space and first-order accurate in time. This solver can model flowfields over isolated wing-alone configurations. Both time-accurate and local time step methods can be used in this solver. For the current study, the time-accurate method is used to predict the unsteady effects. The time step is chosen based on the Courant-Friedrichs-Lewy (CFL) condition. This solver has been validated for clean and iced wings by Kwon and Sankar?' and Bangalore et a1.21 Modifications to this solver have been made to model
YI LIU ET AL.
560
CC jets. l6 Both three-dimensional finite wings and two-dimensional airfoils may be simulated with the same solver. The flow around the airfoil is assumed to be fully turbulent, so currently no transition models are used. Two turbulence models have been used: the Baldwin-Lomax22 algebraic model and the Spalart and A11ma1-a~~~ one-equation model. In this work, all the calculations were done using the Baldwin-Lomax model. The effects of the turbulence model are discussed in Ref. 16.
B. Computational Grid Construction of a high-quality grid around the CCW airfoil is made difficult by the presence of the vertical jet slot. In this solver, the jet slot is treated as part of the airfoil surface, as done by S h r e ~ s b u r y , and ~ ~ , Williams ~~ and Franke.26A hyperbolic three-dimensional C-H grid generator is used to generate the grid. The single-block three-dimensional grid is constructed from a series of two-dimensional C-grids with an H-type topology in the spanwise direction. The normal distance of first grid layer to the airfoil surface is set to lop5 chord length to maintain enough points in the boundary layer. The grid outer boundaries are set to 10 chord lengths away to satisfy nonreflective boundary conditions. The grid is also clustered in the vicinity of the jet slot and the TE to accurately capture the jet behavior over the airfoil surface. From our studies, the TE spacing should be less than lop3 chord length in the streamwise direction, and enough points should be placed in the wake region to accurately capture the jet flow behavior. Grid studies have been carried out for different meshes, and results are shown in Ref. 16. The grid generation and the surface boundary condition routines are general enough so that one can easily vary the slot location, slot size, blowing velocity and the direction of blowing. C. Boundary Conditions In CCW studies, the driving parameter is the momentum coefficient C,, defined as follows: mVjet c -- 1/2p,v:s Here, the jet mass flow rate is given by
Conventional airfoil boundary conditions are applied everywhere except at the jet slot exit. Nonreflection boundary conditions are applied at the outer boundaries of the C grid to allow characteristic waves [for example, Riemann invariant 2a/(y - 1) u,] to leave. On the airfoil surface, adiabatic and no-slip boundary conditions are applied, and the normal derivative of the pressure is set to zero.
+
EVALUATION OF STEADY AND PULSED JET EFFECTS
561
At the jet slot exit, the jet is assumed to be subsonic, and the following conditions are specified: total temperature of the jet Tojet, momentum coefficient C, as a function of time, and the flow angle at the exit. In the simulation, the jet was tangential to the airfoil surface at the exit. For example with subsonic jets, one characteristic can propagate upwind into the slot. Thus the pressure at the jet exit is extrapolated from the outside values. Then the static pressure at the jet slot exit can be obtained as Pj,,
= Pi1 = (4PQ - Pi3)/3
(3)
From Eqs. (1) and (2), the momentum coefficient can also be expressed as
c /J-
Pjet 7:tAjet 1/2pwVLS
(4)
From the ideal gas law and the equation of state, the following relations can be obtained:
Substituting Eq. ( 5 ) into Eq. (4), another expression for C, with just one unknown parameter can be obtained:
The only unknown variable is qet,which can be easily solved from Eq. (6). After the qetis calculated, the other jet flow variables, such as yet and pjet, can be obtained from Eq. ( 5 ) . These parameters are also nondimensionalized by corresponding reference values before being used in the solver as the boundary conditions. Formulations for a supersonic jet and for using total jet pressure as a driven parameter instead of C, can be found in Ref. 16.
111. Results and Discussion The CCW configuration and body-fitted grid studied in the present work are shown in Figs. 1 and 2. The flap-setting angle may be varied both in the experiments and the simulations. The studies presented here are all for the 30deg flap setting to take advantages of CC high-lift benefits while greatly reducing drag. In both the experiments5 and the present studies, the freestream velocity was approximately 94.3 fps at a dynamic pressure of 10psf and an ambient pressure of 14.2psia. The freestream density is 0.00225 slugs/ft3. These conditions translate into a freestream Mach number of 0.0836. The airfoil chord was 8 in. and the Reynolds number was 395,000.
562
YI LIU ET AL.
Fig. 2 Body-fitted C-grid near the CC airfoil surface.
A. Validation Studies Prior to its use in studying CCW configurations, the Navier-Stokes solver was validated by modeling the viscous subsonic flow over a small-aspect-ratio wing made of NACA 0012 airfoil section^,'^ and the results were in good agreement with the experimental measurements of Bragg and Spring.27 These validation studies have been previously documented in Refs. 16 and 17, and are not reproduced here. Figure 3 shows the variation of lift coefficient with respect to C, at a fixed angle of attack (a= 0 deg) for the CCW configuration with a 30 de flap. Excellent agreement with measured data from the experiments by EnglarBis evident. It is seen that very high lift can be achieved by CC technology with a relatively low C,. A lift coefficient as high as 4.0 can be obtained at a C, value of 0.33. And the lift augmentation ACl/AC, is greater than 10 for this 30 deg flap configuration. Figure 4 shows the computed Cl variation with the angle of attack, for a number of C, values, along with measured data. It is found that the lift coefficient increases linearly with angle of attack, just as it does for conventional sharp TE airfoils. However, the increase of lift with angle of attack breaks down at high enough angles. This is a result of static stall, and is much like that experienced with a conventional airfoil, but occurs at higher Cl,maw values, thanks to the beneficial effects of CC. The calculations also correctly reproduce the decrease in the stall angle observed in the experiments at higher momentum
EVALUATION OF STEADY AND PULSED JET EFFECTS
563
-
4-
0
Y
CI, Measured
0
-CI, Computed "
I
0
0.05
0.1
0.15
0.2
0.35
0.3
0.25
0.4
CP Fig. 3 Variation of the lift coefficient with the momentum coefficients at a = 0 deg.
/-:
0
0
C,=O.1657
0
0 0
C,=0.0740
0
EXP, C,
0 EXP, C, 0 EXP, C,
0.0 0.074
= 0.15
-CFD 0
0 0
6
-4
-2
0
2
4
6 8 Angle of Attack
10
12
14
16
Fig. 4 Variation of lift coefficient with angle of attack for different momentum coefficients.
564
YI LIU ET AL.
Fig. 5 Streamlines over the CC airfoil at two instantaneous time levels (C, = 0.1657, angle of attack = 6 deg).
coefficients. With the turbulence model used in this study, it is found that the predicted stall angle is less than experimental measurements. However, the lift prediction is in good agreement with experiments before stall. Unlike conventional airfoils, where experience stall because of the progressive growth of TE separation, CCW configuration stall is a result of LE separation. Figure 5 shows typical streamlines around the CC airfoil at an angle of attack of 6 deg, and C, = 0.1657 at a typical instance in time. In this case, a LE separation bubble forms, which spreads over the entire upper surface, resulting in a loss
EVALUATION OF STEADY AND PULSED JET EFFECTS
565
of lift. However, the flow is still attached over the TE because of the strong Coanda effect.
B. Leading Edge Blowing Functioning like a slat, LE blowing is an effective way of alleviating LE stall and achieving the desired performance at high angles of attack. To understand the effects of LE blowing, a dual-slot CC airfoil was designed, and simulations of both LE and TE blowing were carried out. Figure 6 shows lift coefficient variations with angle of attack for three different combinations of LE and TE blowing. In the first case, there is only a TE blowing with C , = 0.08, and it is seen that the stall angle is very small, at approximately 5 deg. If a small amount of LE blowing is used (C, = 0.04), while keeping the TE blowing at C , = 0.08 as before, the stall angle is greatly increased from 5 deg to 12 deg. If even higher levels of LE blowing are used, for example, LE blowing with C, = 0.08 and TE blowing with C , = 0.04, the stall angle is increased to more than 20 deg, but the total lift is decreased at the same angle of attack compared to the previous case, even when the total momentum coefficients (C,,LE C,,TE) of both cases are the same, equal to 0.12 here. In conclusion, LE blowing is seen to increase the stall angle, replacing the slat, whereas the TE blowing is effective in producing high levels of lift. Leading-edge blowing can also reduce the large nosedown pitch moment associated with high lift and the suction pressure peak in the vicinity of the slot. In general, operating at high angles of attack is not necessary for CC airfoils because high lift can be readily achieved with low angles of attack and a moderate amount of blowing.
+
4
---
3.5 LE Blowing, C$ = 0.04
u
2
3
.-
2.5
i 0
2
LE Blowing, Cy I0.08 TE Blowing, Cy I0.04
0
5
1.5
I
I
Os5 0 0
2
4
6
8
10
12
14
16
18
20
22
Angle of Attack (degrees)
Fig. 6 Lift coefficient vs angle of attack for the LE blowing case.
24
YI LIU ET AL.
566
However, in situations where the CCW configuration must operate at high angles of attack, a combination of LE and TE blowing may be necessary to achieve the best performance.
C. Effects of Freestream Velocity on Lift Production As a followup to previous studies,17 numerical simulations have also been carried out where the freestream velocities (and the Reynolds number) were systematically varied. The purpose of theses studies was to determine and isolate how freestream velocities and the Reynolds number affect the beneficial effects of CC at a fixed momentum coefficient. In this case, the jet momentum coefficient C, is fixed at 0.1657, and the jet slot height is also fixed at 0.015 in. The freestream velocities vary from 0.5 to 1.8 times the experimental freestream velocity, equal to 94.3 fps, as stated earlier. The jet velocity also varies with the freestream velocity to maintain a constant C,. As shown in Figs. 7 and 8, for a given momentum coefficient, the lift and drag coefficients are not significantly affected by the variation of the freestream velocity except at very low freestream velocities. At very low freestream velocities, degradation of lift and the generation of high drag are seen. This is because the jet velocity is too low to generate a sufficiently strong Coanda effect to eliminate TE separation and vortex shedding. At sufficiently high freestream velocities, the performance of CC airfoils is independent of the freestream velocity and the Reynolds number under the fixed C, and fixed jet slot height conditions. Thus the momentum coefficient is an appropriate driving parameter for CC blowing if the jet slot height is fixed.
0
0.2
0.4
0.6
0.8 1 (Vm-cfd) I (Vm+xp)
1.2
1.4
1.6
1.8
2
Fig. 7 Lift coefficient vs freestream velocity (Cp = 0.1657, h = 0.015 in., and Vm,exp= 94.3 fps).
EVALUATION OF STEADY AND PULSED JET EFFECTS
567
0.2
- 0.1 - 5-
A
3 **
Q
'U
E
0.1
8
0
01 0
0.2
0.4
0.6
0.8 1 1.2 (V--cfd)/ (V-exp)
1.4
1.6
1.8
2
Fig. 8 Drag coefficient vs freestream velocity (C, = 0.1657, h = 0.015 in., and Vm,exp= 94.3 fps).
D. Effects of Jet Slot Height According to recent acoustic measurement^,'^"^ the jet slot height has a strong effect on the noise produced by the CC airfoil. These studies indicate that a larger jet slot will reduce the noise at the same momentum coefficient compared to a smaller slot. To investigate the effect of jet slot heights on the aerodynamic characteristics of CCW sections, simulations at several slot heights (varying from 0.006 to 0.018 in.) have been carried out, at a fixed low C, (C, = 0.04) and a fixed high C, (C, = 0.1657) value, and at a constant free-stream velocity of 94.3 fps. From Fig. 9, it is seen that a higher lift coefficient can be achieved with a smaller slot height even for the same momentum coefficient, and that the lift coefficient is decreased by 20% as the slot height is increased from 0.006 in. to 0.018 in. A similar behavior is seen for the drag coefficient as shown in Fig. 10. The LID characteristics of the airfoil, which are computed here as Cl/(Cd C,) by adding C, to the drag coefficient in order to consider the rate of change of momentum associated with the jet flow, do not vary much with the change of the jet slot height. As shown in Fig. 11, when the slot height is increased, the efficiency decreases approximately by 7.6% for the C, = 0.1657 case, and increases by about 5.3% for the C, = 0.04 case. However, as shown in Fig. 12, the jet mass flow rate increases by ~ 6 0 % when the slot height is increased from 0.006 in. to 0.018 in., because of the larger jet slot area. As it is always preferable to obtain higher lift with as low a mass flow rate as possible, a thin jet is aerodynamically more beneficial than a thick jet. However,
+
YI LIU ET AL.
568
-
-1
I
+cp=o.o4 Cp = 0.1657
3-
"
-1
I
0.006
0.009
0.012 Jet Slot Height (inch)
0.015
0.018
Fig. 9 Lift coefficient vs jet slot height ( V , = 94.3 fps).
+Cp = 0.04
----
s
-Cp
= 0.1657
~
0.15
0
- .
0.006
0.009
0.012 Jet Slot Height (inch)
0.015
Fig. 10 Drag coefficient vs jet slot height ( V , = 94.3 fps).
0.018
EVALUATION OF STEADY AND PULSED JET EFFECTS
569
20
YE
1
5
1
1 +Cp = 0.04 +Cv
= 0.1657
5m
00.006
0.009
0.012
0.015
0.018
Fig. 11 Variations of the LID characteristicswith the jet slot height ( V , = 94.3 fps).
+cp=o.o4
-Cu
=aL 0
= 0.1657
0.001 -
0
a
i?i 0.0005 !
04 0.006
0.009
0.012
0.015
0.018
Jet Slot Height (inch)
Fig. 12 Mass flow rate requirements of the jet vs. jet slot height ( V , = 94.3 fps).
570
YI LIU ET AL.
the large stagnation pressure losses associated with small orifices or slots means that a higher stagnation pressure is required to generate a jet issuing through a smaller slot than through a larger slot at the same momentum coefficient. The higher power consumption of compressors needed to produce the required high stagnation pressures can negate the beneficial effects of CC for very thin jets. In summary, a smaller jet slot height is preferred from an aerodynamic design perspective. However, as previously mentioned, a larger jet slot height is preferred from an aeroacoustic perspective. Thus, an optimum choice must be made for the jet slot height from aerodynamic, acoustic, and compressor power consumption considerations.
E. Pulsed Jet Effects During the past five years, there has been increased interest in the use of pulsed jets, and “massless” synthetic jets for flow control and performance enhancement. Wygnansky and colleagues28929studied the effects of eriodic excitation on the control of separation and static stall. Lorber et aleo and Wake and Lurie31 have studied the use of directed synthetic jets for dynamic stall alleviation of the rotorcraft blade. Hassan and Janakiram3’ have studied the use of synthetic jets placed on the upper and lower surfaces of an airfoil as a way of achieving desired changes in lift and drag, and offsetting vibratory airloads that otherwise would occur during blade-vortex interactions. Pulsed jets and synthetic jets have also been used to affect mixing enhancement, thrust vectoring, and bluff body flow separation control. In 1972, Oyler and Palmer33 experimentally studied the pulsed blowing of blown flap configurations. More recently, some numerical simulations employing a pulsed jet have also been reported for separation control of high-lift systems,34 and traditional rounded TE CC airfoils with multiport blowing.35 Most of the studies above were focused on the use of low momentum coefficients or zero-mass blowings to control the boundary layer separation or static and dynamic stall. Only a few studies33 considered the use of pulsed jets for lift augmentation, at smaller mass flow rates compared to steady jets. In earlier work,17 it has been shown that the pulsed jet with square-wave form is more efficient than the traditional sinusoidal form, and that the squarewave-form pulsed jet can generate the same lift as the steady jet at a much lower mass flow rate. In this work, we describe the studies done to isolate the effects of freestream velocity, frequency, and chord length on pulsed jet behavior. Figures 13 and 14 show the variation of the time-averaged incremental lift coefficient ACl over and above the baseline unblown configuration at three frequencies, 40, 120, and 400 Hz. Figure 13 shows the variation with the average momentum coefficient; and Fig. 14 the variation with the average mass flow rate. At first glance, Figs. 13 and 14 appear to show opposite trends. Figure 14 appears to favor high frequencies; that is, ACl increases as frequency increases, and the pulsed jet produces a higher ACl than a steady jet. This appears to be consistent with experiment^.^^ However, Fig. 13 appears to show the opposite trend-the steady jet appears to be always more efficient than a pulsed jet,
EVALUATION OF STEADY AND PULSED JET EFFECTS
571
3 -Steady 2.5 -Pulsed
2
Jet Jet, f = 40 Hz
-. Pulsed Jet, f = 120 Hz
3 1.5 1
0.5
0
~
0.02
0.04
0.06
0.08
0.1
0.12
(I
14
Time-Averaged Momentum Coefficient, CpO
Fig. 13 Incremental lift coefficient vs time-averaged momentum coefficient.
Pulsed Jet, f = 40 Hz Pulsed Jet, f
= 120 Hz
Pulsed Jet, f = 400 Hz
0
0.0002
0.0004 0.0006 0.0008 0.001 0.0012 0.0014 Time Averaged Mass Flow Rate (siugkec)
Fig. 14 Incremental lift coefficient vs time-averaged mass flow rate.
0.0016
YI LIU ET AL.
572
and produces a large ACl. To resolve this “apparent” inconsistency between Figs. 13 and 14, four points A, B, C, D are shown in Fig. 13. These points are all at the same mass flow rate of 0.00088 slug/s. It is seen that point A is above point B. That is, a steady jet is indeed able to produce a higher ACl than a low-frequency 40 Hz jet. This is because the flow separates over a period of time before a new cycle of blowing begins, destroying the lift generation. However, ACl at points C and D (120 and 400 Hz jets) are higher than point A. In these cases, bound circulation over the airfoil has not been fully shed into the wake before a new cycle begins. The time-averaged lift at the same specified averaged mass flow rate for a higher frequency pulsed jet is thus higher compared to a steady jet. This is consistent with Fig. 14. It has also been found that high frequencies have the beneficial effect of decreasing the time-averaged mass flow rate of the pulsed jet.” For example, as shown in Fig. 15, when the frequency is equal to 400 Hz, the pulsed jet requires only 73% of the steady jet mass flow rate while it can achieve 95% of the lift achieved with a steady blowing. Examination of the flowfield over an entire cycle indicates that it takes some time after the jet has been turned off before all the beneficial circulation attributable to the Coanda effect is completely lost. If a new blowing cycle could begin before this occurs, the circulation will almost instantaneously reestablish itself. At high enough frequencies, as a consequence, the pulsed jet will have all the benefits of the steady jet at considerably lower mass flow rates.
-1
I
u .-c”5u
1.2-
E
8
+Pulsed
Jet, Ave. C,=0.04
E A
-Steady
Jet, C,=0.04
0 0.8-
0
4 0 I
0
”
’
’
’
8
8
8
8
8
8
8
8
8
8
8
8
8
1
8
20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 320 340 360 380 400 Frequency (Hz) I
I
I
I
I
I
I
1.414
I
I
I
2.828
Strouhal Number ( f * Chord / Vinf)
Fig. 15 Time-averaged lift coefficient vs frequency and Strouhal number.
EVALUATION OF STEADY AND PULSED JET EFFECTS
573
F. Strouhal Number Effects For aerodynamic and acoustic studies, the frequency is usually expressed as a non-dimensional quantity called the Strouhal number. Simulations have been carried out to calculate the average lift generated by the pulsed jet at fixed Strouhal numbers. The Strouhal number is defined as fief St = VC9
(7)
In the present study, for the baseline case, Gefis 8 in., and V , is equal to 94.3 fps. Thus, for a 200 Hz pulsed jet, the Strouhal number is equal to 1.41. From the preceding equation, besides the frequency, there are two other parameters that could affect the Strouhal number: the freestream velocity V , and Gef (chord of the CC airfoil). To isolate these effects, as shown in Tables 1 to 3, three cases have been studied. In the first case (Table 1), the freestream velocity and the chord of the CC airfoil are fixed, and the Strouhal number varies with the frequency. In the second case, as shown in Table 2, the Strouhal number is fixed at 1.41 and the chord of the CC airfoil is also fixed. The frequency varies with the freestream velocity to achieve the same Strouhal number. In the third case, as shown in Table 3, the Strouhal number is fixed at 1.41 and the freestream velocity is also fixed, whereas the frequency varies along with the chord of the CC airfoil. The Mach number and Reynolds number are also functions of the freestream velocity and the airfoil chord, and were changed appropriately. The time-averaged momentum coefficient CF0 is fixed at 0.04 in these studies. Figure 16 shows the lift coefficient variation with the frequency for these three cases. From Tables 2 and 3, it is seen that the computed time-averaged lift coefficient varies less than 2% when the Strouhal number is fixed, and the chord and/or the freestream velocity is varied. Table 2 also indicates that the same Cl can be obtained at a much lower frequency with a smaller freestream velocity as long as the Strouhal number is fixed. Table 3 shows that for a larger configuration with larger chord lengths, the same Cl can be obtained at a lower frequency provided the Strouhal number is fixed. Table 1, on the other hand, shows that varying the frequency and Strouhal number while holding the other variables fixed can lead to a 12% variation in Cl. Thus, it is concluded the Strouhal number has a Table 1 Computed time-averagedlift coefficient for the case where U, and LrePare fixed, and the Strouhal number is varied with the frequency
Frequency, Hz Freestream velocity U,, fps in. Chord of the Airfoil bef, Strouhal number Computed average lift coefficient (CJ
Baseline
Half frequency
Double frequency
200 94.3 8 1.41 1.6804
100 94.3 8 0.705 1.5790
400 94.3 8 2.82 1.8026
YI LIU ET AL.
574
Table 2 Computed time-averaged lift coefficient for the case where Strouhal number and L,f are fixed, and U , and the frequency are varied
Frequency, Hz Freestream velocity U,, fps Chord of the airfoil Gef,in. Strouhal number Computed average lift coefficient, Cl
Baseline
Half velocity
Double velocity
200 94.3 8 1.41 1.6804
100 47.15 8 1.41 1.6601
400 118.6 8 1.41 1.7112
Table 3 Computed time-averaged lift coefficient for the case where Strouhal number and U , fixed, and Lrefand frequency are varied
Frequency, Hz Freestream velocity U,, fps Chord of the airfoil Lef,in. Strouhal number Computed average lift coefficient, Cl
..
A..
1.24 50
Baseline
Double chord
Half chord
200 94.3 8 1.41 1.6804
100 94.3 16 1.41 1.7016
400 94.3 4 1.41 1.6743
..
100
150
200
250 300 Frequency
350
400
450
Fig. 16 Time-averaged lift coefficient vs frequency: Case 1: Strouhal number not fixed, V , and Lref fixed; Case 2: Strouhal number and L,f fixed, V , not fixed; and Case 3: Strouhal number and V , fixed; Lref not fixed.
EVALUATION OF STEADY AND PULSED JET EFFECTS
575
more dominant effect on the average lift coefficient of the pulsed jet than just the frequency.
IV. Conclusions The Navier-Stokes simulations are used to model flow over the CCW configurations because of the complexity of the flowfield and the strong viscous effects. On comparison with experimental measurements, the results indicate that this approach is an efficient and accurate way of modeling CCW flows with steady and pulsed jets. The CC technology is a useful way of achieving very high lift at even zero angle of attack. It can also eliminate vortex shedding in the TE region, a potential noise source. The lift coefficient of the CC airfoil is also increased with angle of attack like the conventional sharp TE airfoil. However, the stall angle of the CC airfoil decreases rapidly with an increase in the blowing momentum coefficient. This stall phenomenon occurs in the LE region, and may be suppressed by LE blowing. In practice, because high Cl values are achievable at low angles of attack, it may seldom be necessary to operate CC wings at high angles of attack. However, because there is always a large nosedown pitch moment for the CC airfoil, LE blowing may be necessary to reduce this pitch moment at high C, values, even at zero angle of attack. At a fixed momentum coefficient, the performance of the CC airfoil does not vary significantly with the freestream velocity and the Reynolds number. However, at a fixed C ,, the lift coefficient is influenced by the jet slot height. A thin jet from a smaller slot is preferred, because it requires much less mass flow, and has the same efficiency in generating the required Cl values as a thick jet. From a practical perspective, a much higher plenum pressure may be needed to generate thin jets for a given C ., This may increase the power requirements of compressors that provide the high-pressure air. A square-wave-shape pulsed jet configuration gives larger increments in lift over the baseline unblown configuration when compared to the steady jet at the same time-averaged mass flow rate. Pulsed jet performance is improved at higher frequencies because of the fact that the airfoil has not fully shed the bound circulation into the wake before a new pulse cycle begins. The Strouhal number has a more dominant effect on the performance of the pulsed jet than just the frequency. Thus, the same performance of a pulsed jet could be obtained at lower frequencies for a larger configuration or at smaller freestream velocities provided the Strouhal number is kept the same. Acknowledgment This work was supported by NASA Langley Research Center under the Breakthrough Innovative Technology Program, Grant-NAG1-2146. References ‘Goldin, D. S., NASA Headquarters, “National Aeronautics and Space Administration Strategic Plan,” NPD 1000.1B, Sept. 2000, pp. 42-43.
576
YI LIU ET AL.
’Crighton, D. G., “Aircraft Noise in Aeronautics of Flight Vehicles: Theory and Practice,” Vol. 1: Noise Sources, NASA PR-1258, 1991, pp. 391-447. 3Sen, R., “A Study of Unsteady Fields Near Leading-edge Slats,” AIAA Paper 97-1696, 1997. 4Davy, R., and Remy, H., “Airframe Noise Characteristics on a 1/11 Scale Airbus Model,” AIAA Paper 98-2335, June 1998. ’Englar, R. J., Smith, M. J., Kelley, S. M., and Rover, R. C. III., “Application of Circulation Control to Advanced Subsonic Transport Aircraft, Part I: Airfoil Development,” Journal OfAircraft, Vol. 31 No. 5, 1994, pp.1160-1168. 6Englar, R. J., Smith, M. J., Kelley, S. M., and Rover, R. C. III., “Application of Circulation Control to Advanced Subsonic Transport Aircraft, Part 11: Transport Application,” Journal of Aircraft, Vol. 31, No. 5, 1994, pp. 1169-1177. ’Shrewsbury, G. D., and Sankar, L. N., “Dynamic Stall of an Oscillating Circulation Control Airfoil,” Proceedings of International Symposium on Nonsteady Fluid Dynamics, June 1990, pp. 15-22. ‘Shrewsbury, G. D., and Sankar, L. N., “Dynamic Stall of Circulation Control Airfoils,” AIAA Paper 90-0573, Jan. 1990. ’Salikuddin, M., Brown, W. H., and Ahuja, K. K., “Noise From a Circulation Control Wing with Upper Surface Blowing,” Journal of Aircraft, Vol. 24, 1987, pp. 55-64. “McLean, J. D., Crouch, J. D., Stoner, R. C., Sakurai, S., Seidel, G. E., Feifel, W. M., and Rush, H. M., “Study of the Application of Separation Control by Unsteady Excitation to Civil Transport,” NASA/CR-1999-209338, June 1999. “Englar, R. J., and Huson, G. G., “Development of Advanced Circulation Control Wing High Lift Airfoils,” AIAA Applied Aerodynamics Conference, AIAA Paper 83-1847, July 1983. ”Munro, S., Ahuja, K., and Englar, R., “Noise Reduction Through Circulation Control Technology,” AIAA Paper 2001-0666, Jan. 2001. ‘3Munro, S., and Ahuja, K. K., “Aeroacoustics of a High Aspect-Ratio Jet,” AIAA Paper 2003-3323, May 2003; presented at the 9th AIAA/CEAS Aeroacoustics Conference and Exhibit, May 2003. ‘‘Munro, S., and Ahuja, K. K., “Fluid Dynamics of a High Aspect-Ratio Jet,” 9th AIAA/ CEAS Aeroacoustics Conference and Exhibit, AlAA Paper 2003-3 129, May 2003. 15Munro, S., and Ahuja, K. K., “Development of a Prediction Scheme for Noise of High-Aspect Ratio Jets,” 9th AIAA/CEAS Aeroacoustics Conference and Exhibit, AlAA Paper 2003-3255, May 2003. 16Liu, Y., “Numerical Simulations of the Aerodynamic Characteristics of Circulation Control Wing Sections,” Ph.D Dissertation, School of Aerospace Engineering, Georgia Inst. of Technology, Atlanta, GA, 2003. ”Liu, Y., Sankar, L. N., Englar, R. J., and Ahuja, K. K., “Numerical Simulations of the Steady and Unsteady Aerodynamic Characteristics of a Circulation Control Wing Airfoil,” AIAA Paper 2001-0704, Jan. 2001. 18Douglas, J., “On the Numerical Integration of ut = u,, uyy by Implicit Methods,” Journal of Society of Industrial and Applied Mathematics, Vol. 3, No. 1, 1955. ”Briley, W., and McDonald, H., “Solution of Multi-Component Navier-Stokes Equations by Generalized Implicit Method,” Journal of Computational Physics, Vol. 24, 1977, p. 372. ”Kwon, J., and Sankar, L. N., “Numerical Study of the Effects of Icing on Finite Wing Aerodynamics,” AIAA Paper 90-0757, Jan. 1990.
+
EVALUATION OF STEADY AND PULSED JET EFFECTS
577
’lBangalore, A., Phaengsook, N., and Sankar, L. N., “Application of a Third Order Upwind Scheme to Viscous Flow over Clean and Iced Wings,” AIAA Paper 94-0485, Jan. 1994. ”Baldwin, B. S., and Lomax, H., “Thin Layer Approximation and Algebraic Model for Separated Turbulent Flows,” AIAA Paper 78-257, Jan. 1978. 23Spalart, P. R., and Allmaras, S. R., “A One-Equation Turbulence Model for Aerodynamic Flows,” AIAA Paper 92-0439, Jan. 1992. 24Shrewsbury,G. D., “Numerical Evaluation of Circulation Control Airfoil Performance Using Navier-Stokes Methods,” AIAA Paper 86-0286, Jan. 1986. 25Shrewsbury,G. D., “Numerical Study of a Research Circulation Control Airfoil Using Navier-Stokes Methods,” Journal ofAircraft, Vol. 26, No. 1, 1989, pp. 29-34. 26Williams, S. L., and Franke, M. E., “Navier-Stokes Methods to Predict Circulation Control Airfoil Performance,” Journal of Aircraft, Vol. 29, No. 2, 1992, pp. 243-249. 27Bragg, M. B., and Spring, S. A., “An Experimental Study of the Flow Field about an Airfoil with Glaze Ice,” AIAA 25th Aerospace Science Meeting, AIAA Paper 87-0100, Jan. 1987. 28Seifert, A., Darabi, A., and Wygnanski, I., “Delay of Airfoil Stall by Periodic Excitation,” Journal of Aircraft, Vol. 33, No. 4, 1996. 29Wygnanski, I., “Some New Observations Affecting the Control of Separation by Periodic Excitation,” Fluids 2000 Conference and Exhibit, AIAA Paper 2000-23 14, June 2000. 30Lorber,P. F., McCormick, D., Anderson, T., Wake, B. E., MacMartin, D., Pollack, M., Corke, T., and Breuer, K., “Rotorcraft Retreating Blade-Stall Control,” Fluids 2000 Conference and Exhibit, AIAA Paper 2000-2475, June 2000. 31Wake, B., and Lurie, E. A., “Computational Evaluation of Directed Synthetic Jets for Dynamic Stall Control,” 57th American Helicopter Society Annual Forum, Washington DC, 9-11 May 2001. 32Hassan, A., and Janakiram, R. D., “Effects of Zero-Mass Synthetic Jets on the Aerodynamics of the NACA 0012 Airfoil,” Journal of the American Helicopter Society, Vol. 43, No. 4, 1998. 330yler, T. E., and Palmer, W. E., “Exploratory Investigation of Pulse Blowing for Boundary Layer Control,” North American Rockwell Rept. NR72H-12, Jan. 1972. 34Schatz, M., and Thiele, F., “Numerical Study of High-Lift Flow with Separation Control by Periodic Excitation,” AIAA Paper 2001-0296, Jan. 2001. 35Sun, M., and Hamdani, H., “Separation Control by Alternating Tangential Blowing/ Suction at Multiple Slots,” AIAA Paper 2001-0297, Jan. 2001.
Chapter 23
Time-Accurate Simulations of Synthetic Jet-Based Flow Control for a Spinning Projectile Jubaraj Sahu*
US.Army Research Laboratory, Aberdeen Proving Ground, Maryland
Nomenclature D = drag force, N d = reference diameter, m f = jet frequency, Hz Fy = aerodynamics force in y-direction (lift force) F, = aerodynamics force in z-direction (side force) F = inviscid flux vector G = viscous flux vector H = vector of source terms I = impulse, N-s L = lift force, N M = Mach number p = pressure, N/m2 p s = projectile spin rate, Hz t = time, ms V , = freestream velocity, m/s vj =jet velocity, m/s W = vector of conservative variables x, y, z = axial, normal (vertical), and horizontal axes y+ = normal viscous sublayer spacing a = angle of attack, deg
*Aerospace Engineer. Associate Fellow AIAA. This material is declared a work of the U.S.Government and is not subject to copyright protection in the United States.
579
J. SAHU
580
I. Introduction determination of aerodynamics is critical to the low-cost CCURATE development of new advanced munitions. Competent smart munitions that can more accurately hit a target can greatly increase lethality and enhance survivability. Desert Storm convincingly demonstrated the value of large-scale precision-guided munitions. A similar capability for small-scale munitions would increase the effectiveness of infantry units, reduce collateral damage, and reduce the weight of munitions that must be carried by individual soldiers. The Army is, therefore, seeking a new generation of autonomous, course-correcting, gun-launched projectiles for infantry soldiers. Because of the small projectile diameter (d = 0.02 to 0.04 m), maneuvers by canards and fins seem very unlikely. An alternative and new evolving technology is microadaptive flow control through synthetic jets. These very tiny (of the order of 0.3 mm) synthetic microjet actuators have been shown successfully to modify subsonic flow characteristics and pressure distributions for simple airfoils and cylinders.394The synthetic jets (fluid being pumped in and out of the jet cavity at a high frequency of the order 1000 Hz) are control devices (Fig. 1) with zero net mass flux and are intended to produce the desired control of the flowfield through momentum effects. Many parameters such as jet location, jet velocity, and jet actuator frequency, can affect the flow control phenomenon. Until now, the physics of this phenomenon has not been well understood. In addition, advanced numerical predictive capabilities or high-fidelity computational fluid dynamics (CFD) design tools either did not exist or have not been successfully applied to practical real-world problems involving microadaptive flow control. The present research effort described here is focused on advancing aerodynamic numerical capability to predict accurately and provide a crucial understanding of the complex flow physics associated with the unsteady aerodynamics of this new class of tiny synthetic microjets for control of modem projectile configurations. High-performance CFD techniques are developed and applied for the design and analysis of these microadaptive flow control systems for steering a spinning projectile for infantry operations.
A
',*
Pulsating Synthetic Jet Diaphragm
Fig. 1 Schematic of a synthetic jet.
FLOW CONTROL ON SPINNING PROJECTILE
581
The control of the trajectory of a 40mm spinning projectile is achieved by altering the pressure distribution on the projectile through forced asymmetric flow separation. Unsteady or time-accurate CFD modeling capabilities are developed and used to assist in the design of the projectile shape, the placement of the synthetic actuators, and the prediction of the aerodynamic force and moments for these actuator configurations. Additionally, the advanced CFD capabilities provide a simpler way to explore various firing sequences of the actuator elements. Time-accurate unsteady CFD computations have been performed to predict and characterize the unsteady nature of the synthetic jet interaction flowfield produced on the M203 grenade launched projectile for various yaw and spin rates for fully viscous turbulent flow conditions. Turbulence is usually modeled using a traditional Reynolds-averaged NavierStokes (RANS) approach. RANS models are easy to use and provide very good results for many steady flows, especially at supersonic speeds. Although this approach provides some detailed flow physics, it is not well suited and can be less accurate for the new class of unsteady flows associated with synthetic jets at subsonic speeds. In order to improve the accuracy of the numerical simulation, the predictive capability has been extended to include a higher order hybrid RANS/LES (large eddy simulation) approach.596This new approach computes the large eddies present in the turbulent flow structure (in the vicinity of the microjet) and allows the simulation to capture, with high fidelity, additional flow structures associated with the synthetic jet interactions (in the projectile wake or base flow in the present study) in a time-dependent fashion. Modeling of azimuthally placed synthetic microjets requires adequate grid resolution, highly specialized boundary conditions for jet activation, and the use of an advanced hybrid LES approach permitting local resolution of the unsteady turbulent flow with high fidelity. The addition of yaw (angle of attack) and spin while the projectile is subjected to the pulsating microjets rendered predicting forces and moments a major challenge. Both RANS and hybrid RANS/LES models have been used in the present study. Although the RANS method works well for steady flows, the accuracy of this method for unsteady flows may be less than desired. Because the large-energycontaining eddies are computed using the LES method, this technique is expected to be more capable of handling unsteady shear layers and wakes, and so on. The advanced CFD capability used here solves the full three-dimensional Navier- Stokes equations and incorporates unsteady boundary conditions for simulation of the synthetic jets. The present study investigates the ability of these advanced techniques with time-accurate computations of unsteady synthetic jets for both nonspinning and spinning projectile cases at low subsonic speeds. The following sections describe the numerical procedure, the unsteady jet boundary condition, the hybrid RANS/LES turbulence model, and the computed results obtained. 11. Computational Methodology The complete set of three-dimensional time-dependent Navier- Stokes equations7 is solved in a time-accurate manner for simulations of the unsteady synthetic jet interaction flowfield on the M203 grenade launched projectile
582
J. SAHU
with spin. The three-dimensional time-dependent RANS equations are solved using the finite-volume method’:
where W is the vector of conservative variables, F and G are the inviscid and viscous flux vectors, respectively, H is the vector of source terms, V is the cell volume, and A is the surface area of the cell face. Second-order discretization was used for the flow variables and the turbulent viscosity equations. Two-equation9 and higher-order hybrid RANS/LES6 turbulence models were used for the computation of turbulent flows. The hybrid RANS/LES approach based on limited numerical scales (LNS)6 is well suited to the simulation of unsteady flows and contains no additional empirical constants beyond those appearing in the original RANS and LES subgrid models. With this method, a regular RANS-type grid is used except in isolated flow regions where denser, LES-type mesh is used to resolve critical unsteady flow features. The hybrid model transitions smoothly between an LES calculation and a cubic k--E model, depending on grid fineness. A somewhat finer grid was placed around the body, and near the jet, the rest of the flowfield being occupied by a coarser, RANS-like mesh. Dual time-stepping was used to achieve the desired time accuracy. In addition, special jet boundary conditions were developed and used for numerical modeling of synthetic jets. The grid was actually moved to take into account the spinning motion of the projectile.
Unsteady Jet Boundary Conditions One particular boundary condition (BC) used in the present simulations of the unsteady jets is an “oscillating jet” BC. In its basic form, it is a steady inflow/ outflow BC, inwhich the user supplies the velocity normal to the boundary along with static temperature and any turbulence quantities. When the velocity provided is negative, it is considered to be an inflow, and when it is positive, it is treated as an outflow. In the case of inflow, the static temperature and turbulence quantities are utilized along with the inflow velocity. In the case of outflow, only the velocity is utilized. At inflow, the tangential component of velocity is set to zero, and at outflow, the tangential component is extrapolated from the interior. At outflow, all primitive variables except normal velocity are extrapolated from the interior. At inflow, the static pressure is taken from the interior. This BC also has a set of modifiers. The first modifier available for this BC allows the velocity to oscillate. The base velocity is multiplied by an amplitude that varies as sin(2@), wherefis the frequency of the oscillation. Thus, the oscillating velocity can cycle from being positive to being negative and back within each period (or from being negative to positive and back, based on the sign of the input for the basic BC formulation). A second modifier permits the steady or oscillating inflow/outflow to be on over certain time intervals and off during other intervals. During “on” periods, the basic or the basic multiplied by the oscillating amplitude multiplier (first modifier), is used. The user provides the ranges of time during which the jet is on. The user also provides a repetition A.
FLOW CONTROL ON SPINNING PROJECTILE
583
time period (e.g., the time period corresponding to one spin rotation of the projectile). Within each time period, therefore, there are sets of start and end times that define when the jet is on. During “off” periods, the amplitude is set to zero. In parts of the cycle when the jet is off, the boundary condition thus reverts to the condition of inviscid surface tangency. This allows slip past the boundary, as would exist (in the form of a shear layer) if the jet was emanating from a cavity /hole.
B. Hybrid RANS/LES Turbulence Model Currently, the two most popular forms of turbulence closure, namely ensemble-averaged models (typically based on the RANS equations), and LES with a subgrid-scale model, both face a number of unresolved difficulties. Specifically, existing LES models have met with problems related to the accurate resolution of the near-wall turbulent stresses. In the near-wall region, the foundations of largeeddy simulation are less secure, because the sizes of the (anisotropic) near-wall eddies approach than of the Kolmogorov scale, requiring a mesh resolution approaching that of a direct numerical simulation. On the other hand, existing ensemble-averaged turbulence models are limited by their empirical calibration. Their representation of small-scale flow physics cannot be improved by refining the mesh, and over short time scales they tend to be overly dissipative with respect to perturbations around the mean, often suppressing unsteady motion altoget her. Although LES is an increasingly powerful tool for unsteady turbulent flow prediction, it is still prohibitively expensive. To bring LES closer to becoming a desi n tool, a hybrid RANS/LES approach based on limited numerical scaleskhas been recently developed by Metacomp Technologies.’ This approach combines the best features of RANS and LES in a single modeling framework. The hybrid RANS/LES model is formulated from an algebraic or differential Reynolds-stress model, in which the subgrid stresses are limited by the numerically computed local length-scale and velocity-scale products. It thus behaves like its parent RANS model on RANS-type grids, but reverts to an anisotropic LES subgrid model as the mesh is refined locally, thereby reaching the correct (DNS) fine-grid limit. Locally embedded regions of LES may be achieved automatically through local grid refinement, whereas the superior near-wall stress predictions of the RANS model are preserved, removing the need for ad hoc, topography-parameter-based wall damping. The hybrid RANS/LES formulation is well suited to the simulation of unsteady flows, including mixing flows, and contains no additional empirical constants beyond those appearing in the original RANS and LES subgrid models. With this method a regular RANS-type grid is used except in isolated flow regions where denser, LES-type mesh is used to resolve critical unsteady flow features. The hybrid RANS/LES model transitions smoothly between an LES calculation and a cubic k--E model, depending on grid fineness. A somewhat finer grid was placed around the body, and near the jet, the rest of the flowfield being occupied by a coarser, RANS-like mesh. To date, the hybrid RANS/LES technique has been used successfully on a number of unsteady flows. Examples include flows over cavities, flows around
584
J. SAHU
blunt bodies, flows around airfoils and wings at high angle of attack, separation suppression using synthetic jets, forced and natural convection flows in a room, and mixing flows in nozzles.
111. Projectile Geometry and Computational Grid The projectile used in this study is a 1.%caliber ogive-cylinder configuration (see Fig. 2). Here, the primary interest is in the development and application of CFD techniques for accurate simulation of projectile flowfield in the presence of unsteady jets. The first step here was to obtain a converged solution for the projectile without the jet. The converged jet-off solution was then used as the starting condition for the computation of time-accurate unsteady flowfield for the projectile with synthetic jets. The jet locations on the projectile are shown in Fig. 3. The jet conditions were specified at the exit of the jet for the unsteady (sinusoidal variation in jet velocity) jets. The jet conditions specified include the jet pressure, density, and velocity components. Numerical computations have been made for these jet cases at subsonic Mach numbers, M = 0.11 and 0.24, and at angles of attack a = 0 to 4 deg. The jet width was 0.32 mm, the jet slot halfangle was 18 deg, and the absolute peak jet velocities used were 3 1 and 69 m/s operating at a frequency f = 1000 Hz. A computational grid expanded near the vicinity of the projectile is shown in Fig. 4. Grid points are clustered near the jet as well as the boundary layer regions to capture the high gradient flow regions. The computational grid is a single block; it has 211 points in the streamwise direction, 241 in the circumferential direction, and 80 in the normal direction. The grid is closeted near the body surface with grid spacing that corresponds to a y+ value of approximately 1.0. The same grid was used for both RANS and hybrid RANS/LES calculations. The unsteady simulation took thousands of hours of CPU time on Silicon Graphics Origin and IBM SP3 computers running with 16-24 processors. More details of the CPU time usage and requirement are iven in Section IV. The parallel processing capability in CFD++ code' was designed in the beginning to be able to run on a wide variety of hardware platforms and
Fig. 2 Projectile geometry.
FLOW CONTROL ON SPINNING PROJECTILE
585
Jet
Fig. 3 Aft-end geometry showing the jet location.
communications libraries, including MPI and PVM. MPI was used on various platforms for communications between different processors. The code runs on parallel processors and one can switch the use of an arbitrary number of CPUs at any time. Depending on the number of CPUs being employed, the mesh is domain-decomposed using the METIS tool developed at the University of Minnesota.
Fig. 4 Computational grid near the projectile.
586
J. SAHU
IV. Results Time-accurate unsteady numerical computations using advanced viscous Navier-Stokes methods were performed to predict the flowfield and aerodynamic coefficients on both a nonspinning and a spinning projectile. Limited experimental data (from Ref. 10 and private communication with J. McMichael, GTRI) exist only for the nonspinning case and were used to validate the unsteady CFD results. Three-dimensional numerical computations have been performed for the projectile configuration with jet-interaction using CFD++ code at subsonic Mach numbers, M = 0.1 1 and 0.24, and at angles of attack a = 04 deg. The preconditioned version of the CFD++ code was used to obtain an efficient numerical solution at low speeds. For modeling of the unsteady synthetic jets, both unsteady RANS and a hybrid RANS/LES approach6 were used. For computations of these unsteady jets, full three-dimensional computations are performed and no symmetry was used. A. Nonspinning Projectile Three-dimensional unsteady CFD results were obtained at a subsonic Mach number of 0.11 (V, = 37 m/s) and several angles of attack from 0 to 4 deg using both the unsteady RANS and the hybrid RANS/LES approaches. The synthetic jets are on all the time for these nonspinning cases. These three-dimensional unsteady CFD computations are carried out to provide fundamental understanding of fluid dynamics mechanisms associated with the interaction of the unsteady synthetic jets and the projectile flowfields at subsonic speeds. Many flowfield solutions resulting from the simulation of multiple spin cycles and, hence, a large number of synthetic jet operations, were saved at regular intermittent time intervals to produce movies to gain insight into the physical phenomenon resulting from the synthetic jet interactions. The unsteady jets were discovered to break up the shear layer coming over the step in front of the base of the projectile. It is this insight that was found to substantially alter the flowfield (making it unsteady) both near the jet and in the wake region that in turn produced the required forces and moments even at 0-deg angle of attack (level flight). Time-accurate velocity magnitude (Fig. 5 ) and velocity vectors (Fig. 6 ) confirm the unsteady wake flowfields arising from the interaction of the synthetic jet with the incoming freestream flow at Mach = 0.11. Figure 7 shows the particles emanating from the jet and interacting with the wake flow, making it highly unsteady. More important, the breakup of the shear layer is clearly evidenced by the particles clustered in regions of flow gradients or vorticity (evident in computed pressure contours, Fig. 8). Verification of this conclusion is provided by the excellent agreement (Fig. 9) between the predicted (solid line) and measured" (solid symbols) values of the net lift force due to the jet. In this case, the solid line represents the results obtained with the hybrid RANS/LES turbulence model. Also shown in Fig. 9 is a time-averaged result of the lift force obtained using a RANS turbulence model at 0-deg angle of attack. It is quite clear that the lift force is underpredicted by the RANS model and does not compare as well with the experimental data. This indicates the inability of the RANS model to predict accurately the unsteady wake flowfields resulting from the synthetic jet flow control.
FLOW CONTROL ON SPINNING PROJECTILE
587
Fig. 5 Velocity magnitudes, M = 0.11, (Y = 0 deg.
The net lift force (F,) was determined by time-averaging the actual time histories of the highly unsteady lift force (an example shown in Fig. 10 for various angles of attack) resulting from the jet interaction at zero-degree angle of attack and computed with the new hybrid RANS/LES turbulence approach. Figure 10
Fig. 6 Velocity vectors, M = 0.11, (Y = 0 deg.
5aa
J. SAHU
Fig. 7 Particle traces, M = 0.11, cx = 0 deg.
shows both low- and high-frequency oscillations in the predicted lift force at different angles of attack, a = 0, 2, and 4 deg. The high-frequency oscillations (of the order of 1 ms) are a direct result of the jet actuation that corresponds to the jet frequency of 1000 Hz. The low frequency oscillations observed in the
Fig. 8 Computed pressures, M = 0 . 1 1 , ~=~0 deg.
FLOW CONTROL ON SPINNING PROJECTILE
o.8 h
589
I
u)
c
9 0.6-
1
Q)
w EXPERIMENT +CFD (Hybrid RANSILES) + CFD(RANS)
t Q)
92
0.4-
it: 4
.-E
! L o.2A-
w
w
+ w
Q
5
0
0
time-histories result from the interaction of the jet with wake and the resulting unsteady wake flowfields.
B. Spinning Projectile Of more interest is the spinning projectile case for the real-world applications. Numerical computations have been made in this case for actual flight condition at
Time (ms) Fig. 10 Time-historiesof computed lift force at angles of attack cu = 0,2, and 4 deg, hybrid RANS/LES model, M = 0.11.
J. SAHU
590
Y
t .a .............$ypf Jet-on1
t=3.73 ms
t=O
......a
Fig. 11 Schematic of jet actuation for one spin cycle (view from the nose).
a Mach number, of M = 0.24, an angle of attack, of a = 0 deg, and a spin rate of 67 Hz. The atmospheric flight conditions are used here. The jet width was 0.32 mm, the jet slot half-angle was 18 deg, and the absolute peak jet velocities used were 31 and 69 m/s operating at a frequency of 1000 Hz. In this case, the projectile (40 mm grenade) spins clockwise at a rate of 67 Hz looking from the front (Fig. 11). Unlike the nonspinning cases where the jet was on all the time, here the jet actuation corresponds to one-fourth of the spin cycle from -45 to +45 deg with 0 deg being the positive y-axis. The jet is off during the remaining three-fourths of the spin cycle. The unsteady CFD modeling required about 600 time steps to resolve a full spin cycle. For the part of the spin cycle when the jet is on, the 1000 Hz jet operated for approximately for four cycles. Time-accurate CFD modeling of each jet cycle required over 40 time steps. The actual computing time for one full spin cycle of the projectile was about 50 hours using 16 processors (i.e., 800 processor-hours) on an IBM SP3 system for a mesh size of about four million grid points. Multiple spin cycles and, hence, a large number of synthetic jet operations were required to reach the desired periodic time-accurate unsteady result. Some cases were run for as many as 60 spin cycles, requiring over 48,000 processor hours of computer time. Computed particle traces emanating from the jet into the wake are shown in Fig. 12 at four different instants in time for M = 0.24 and a = 0 deg. As stated earlier, the 1000 Hz synthetic jet operates for about four jet cycles during one spin cycle of the rotating projectile. The four different instants of time selected in Fig. 12 correspond to each of the four jet cycles as the projectile rotates counterclockwise (looking from the back of the projectile). The particle traces emanating from the jet interact with the wake flow making it highly unsteady. It also shows the flow in the base region to be asymmetric because of the interaction of the unsteady jet. The computed surface pressures from the unsteady flowfields were integrated to obtain the aerodynamic forces and moments" from both unsteady RANS as well as the hybrid RANS/LES solutions. The jet-off unsteady RANS calculations were first obtained and the jets were activated beginning at time, t = 28 ms. Computed normal or lift force (F,) and side force (F,) were obtained for two different jet velocities, Vj = 31 and 69 m/s, and are shown in Fig. 13 for the bigger jet as a function of time. These computed results clearly indicate the
FLOW CONTROL ON SPINNING PROJECTILE
591
Fig. 12 Instantaneous computed particle traces at different times jet-on, M = 0.24, a = 0 deg.
Time (ms) Fig. 13 Computed lift and side forces, unsteady RANS, M = 0.24, vj = 69 m/s, a = 0 deg.
592
J. SAHU
unsteady nature of the flowfield. When the jet is on, one can observe a sharp rise in both the lift and the side forces. The peak levels in the forces remain high until the jet is turned off. When the jet is turned off, the levels of these forces drop to the same levels (low-amplitude oscillations) prior to the jet activation corresponding to the jet-off wake flow. The unsteady RANS results clearly show when the jet is on and when it is off during the spin cycle. Figure 14 shows the comparison of the predicted lift force using the unsteady RANS and the hybrid RANS/LES turbulence models for the bigger jet case at zero-degree angle of attack. As indicated earlier, the unsteady RANS results of the lift and the side forces clearly show when the jet is on and when it is off during the spin cycle. The effect due to the jet for the hybrid RANS/LES case is not as easily seen. It is hidden in these oscillations. However, the mean value of the lift force seems to be close to zero when the jet is off during the spin cycle. In general, the levels of the lift force oscillations predicted by the hybrid RANS/LES model are larger than those predicted by the unsteady RANS model. This result can be attributed to the fact that the wake is unsteady and the hybrid RANS/LES model produces large levels of oscillations for the unsteady wake flowfield whether the jet is off or on. As described earlier, the comparisons for the nonspinning cases showed that the level of lift force predicted by the hybrid RANS/LES closely matched the data. Here, the addition of spin as well as the jet actuation for part of the spin cycle further complicates the analysis of the CFD results when the hybrid RANS/LES model is used. The level of oscillations seen is quite large and the effect of the jet cannot be easily seen in the instantaneous time histories of the unsteady forces and moments. In addition, the unsteady wake flowfield is expected to change from one spin cycle to another. To get the net effect of the jet, unsteady computations were run for many spin cycles of the projectile with
Time (rns) Fig. 14 Computed lift forces, unsteady RANS and hybrid RANS/LES, M = 0.24, Vj = 69 m/s, 01 = 0 deg.
FLOW CONTROL ON SPINNING PROJECTILE
593
0.4
0.3 h
t
0.2
?I
0.1
0
0
5
10
15
Time (ms)
Fig. 15 Computed time-averaged lift force over many spin cycles, hybrid RANS/ LES, Vj = 69 m/s, M = 0.24, a = 0 deg, P, = 67 Hz.
the synthetic jets. The CFD results are plotted over only one spin cycle; each subsequent spin cycle was superimposed and a time-averaged result was then obtained over one spin cycle. In all these cases, the jet is on for one-fourth of the spin cycle (time, t = 0-3.73 ms) and is off for the remainder (threefourths) of the spin cycle. Figures 15 through 16 show the time-averaged results over a full spin cycle that corresponds to 15 ms (67 Hz)approximately. Figure 15 shows the computed lift force, again averaged over many spin 0.4
0.3
j 0.2
h
al
t
0.1
0 -0.1 0
5
10
15
Time (ms)
Fig. 16 Computed time-averaged lift force over many spin cycles for different jet velocities, hybrid RANS/LES, M = 0.24, a = 0 deg, P, = 67 Hz.
J. SAHU
594
0
10
20
30
40
50
60
Number of Spin Cycles Fig. 17 Impulse from the lift force vs spin cycles for two jet velocities, hybrid RANS/LES, M = 0.24, a = 0 deg, P, = 67 Hz.
cycles (10,20, 30, and 40) for the peak jet velocity of 69 m/s. The jet effect can clearly be seen when the jet is on ( t = 0-3.73 ms) even after 10 spin cycles. The net lift is about 0.17 N because of the jet actuation and seems to have converged after 20 spin cycles. For the remainder of the spin cycle, the jet is off however, the effect of the jet on the wake still persists and this figure shows that lift force (mean value 0.07 N) is still available. The fact that one can obtain a lift force for this jet-off portion of the spin cycle is a new result solely caused by the spin effect of the projectile. Figure 16 shows the computed time-averaged lift force after 50 and 60 spin cycles for jet velocities 3 1 and 69 m/s, respectively. It clearly shows that the larger jet produces larger lift force than the smaller jet when the jet is activated. The lift force can be integrated over time to obtain the impulse I . Figure 17 shows the impulse obtained from the lift force as a function of the spin cycles for both jets. As seen here, in both cases it takes about 30 to 40 spin cycles before the impulse asymptotes to a fixed value. The computed lift force along with other aerodynamic forces and moments, directly resulting from the pulsating jet, were then used in a trajectory analysis (from private communication with M. Costello, Oregon State University) and the synthetic microjet was found to produce a substantial change in the cross range. These results indicate the viability of the use of synthetic microjets to provide the desired course correction for the projectile to hit its target. V. Conclusions This chapter describes a computational study undertaken to determine the aerodynamic effect of tiny synthetic jets as a means to provide the control authority needed to maneuver a projectile at low subsonic speeds. Computed results have been obtained for a subsonic projectile for both nonspinning and spinning cases using a time-accurate Navier- Stokes computational technique and advanced
FLOW CONTROL ON SPINNING PROJECTILE
595
turbulence models. The unsteady jet in the case of the subsonic projectile is shown substantially to alter the flowfield both near the jet and the base region which in turn affects the forces and moments even at 0-deg angle of attack. The predicted changes in lift force due to the jet match well with the experimental data for various angles of attack from 0 to 4 deg in the hybrid RANS/LES computations. For the spinning projectile cases, the net time-averaged results obtained over the time period corresponding to one spin cycle clearly showed the effect of the synthetic jets on the lift as well as the side forces. The jet interaction effect is clearly seen when the jet is on during the spin cycle. However, these results show that there is an effect on the lift force (although reduced) for the remainder of the spin cycle even when the jet is off. This is a result of the wake effects that persist from one spin cycle to another. The impulse obtained from the predicted forces for both jets seems to asymptote after 30 spin cycles. The results have shown the potential of CFD to provide insight into the jet interaction flowfields and provided guidance as to the locations and sizes of the jets to generate the control authority required to maneuver a spinning munition to its target with precision. This research represents a major increase in capability for determining the unsteady aerodynamics of munitions in a new area of flow control and has shown that microadaptive flow control with tiny synthetic jets can provide an affordable route to lethal precision-guided infantry weapons.
References ‘Sahu,J., Heavey, K. R., and Ferry, E. N., “Computational Fluid Dynamics for Multiple Projectile Configurations”, Proceedings of the 3rd Overset Composite Grid and Solution Technology Symposium, Oct. 1996. ’Sahu, J., Heavey, K. R., and Nietubicz, C. J., “Time-Dependent Navier-Stokes Computations for Submunitions in Relative Motion,” 6th International Symposium on Computational Fluid Dynamics, Sept. 1995. 3Smith, B. L., and Glezer, A., “The Formation and Evolution of Synthetic Jets,” Journal of Physics of Fluids, Vol. 10, No. 9, 1998. 4Amitay, M., Kibens, V., Parekh, D., and Glezer, A., “The Dynamics of Flow Reattachment over a Thick Airfoil Controlled by Synthetic Jet Actuators,” AIAA Paper 99-1001, Jan. 1999. ’Arunajatesan, S., and Sinha, N., “Towards Hybrid LES-RANS Computations of Cavity Flowfields,” AIAA Paper 2000-0401, Jan. 2000. 6Batten, P., Goldberg, U., and Chakravarthy, S., “Sub-grid Turbulence Modeling for Unsteady Flow with Acoustic Resonance,” 38th AIAA Aerospace Sciences Meeting, AIAA Paper 00-0473, Jan. 2000. ’Pulliam, T. H., and Steger, J. L., “On Implicit Finite-Difference Simulations of ThreeDimensional Flow,” AIM Journal, Vol. 18, No. 2, 1982, pp. 159-167. ‘Peroomian, O., Chakravarthy, S., Palaniswamy, S., and Goldberg, U., “Convergence Acceleration for Unified-Grid Formulation Using Preconditioned Implicit Relaxation,” AIAA Paper 98-01 16, June 1998.
596
J. SAHU
’Goldberg, U., Peroomian, O., and Chakravarthy, S., “A Wall-Distance-Free k-e Model With Enhanced Near-Wall Treatment,” ASME Journal of Fluids Engineering, V O ~120, . 1998, pp. 457-462. ‘‘finehart, C., McMichael, J. M., and Glezer, A., “Synthetic Jet-Based Lift Generation and Circulation Control on hisymmetric Bodies,” AIAA Paper 2002-3 168, June 2002. “Sahu, J., “Unsteady Numerical Simulations of Subsonic Flow over a Projectile with Jet Interaction,” AIAA Paper 2003-1352, Jan. 2003.
IV. Exploring a Visionary Use of Circulation Control
Chapter 24
Coanda Effect and Circulation Control for Nonaeronautical Applications Terence R.Day* Vortex Dynamics Pty Ltd, Mount Tamborine, Queensland, Australia
I. Introduction T THE “Coanda Effect/CC Workshop in Hampton, Virginia (March 16- 17, 2004)”’ the question was posed, “What are the roadblocks to further development?” Those roadblocks may be a result of a failure to address certain deficiencies or an inability to find solutions. Examples of operational deficiencies are insufficient quantity of CC air, heavy, complicated air pumps, heavy, energywasting plumbing, and so on. To address some of these issues the author describes here a number of practical nonaeronautical devices employing the Coanda effect or Coanda/Circulation Control (CC), a novel high-volume pump and a novel fan to supply CC air. These projects are proposed commercial outcomes for the Coanda effect and CC. The purpose is to describe these novel applications and propose that some creativity may be beneficial in promotion of the Coanda effect and CC to gain credibility in a wider arena than only within the Coanda effect/CC scientific community. The overview papers in this book and other available contain adequate history and applications of the Coanda effect as it relates to CC and the present author will start from this platform of knowledge and show its applications to novel nonaeronautical situations.
A
*Consultant. Copyright 0 2005 by Terence R. Day. Published by the American Institute of Aeronautics and Astronautics, Inc., with permission.
599
T. R. DAY
600
11. Applications A. Oscillating Channel Flow Including Self Oscillating Channel Flow (Coanda Effect) Although this phenomenon has been understood for quite some time,4 it apparently has been a curiosity with little vision for many useful applications. The geometry of a rectangular channel that enables jet self-oscillating flow must be relatively precise to work at all. Gas jets in a channel will oscillate by imposition of a pressure change alternating either side of the jet. With precise geometry, a round jet will self-oscillate (Fig. 1). It is not difficult to produce either type of oscillating flow if the air supply is sourced conveniently from the lab compressor. For some applications including airborne odor treatment, certain chemicals are coated onto surfaces in order to interact with a turbulent airflow. If the airflow is laminar, the odor molecules contained in the airflow cannot contact the chemical coated surfaces. Oscillating channel flow gives the desired turbulence. A second reason for employing oscillating channel flow is that as the jet skips from wall to wall, a particularly formed passageway is able to accept each branch of the flow. The significant breakthrough here is being able to convert a highly turbulent fan flow into a flow structure that can self-oscillate in a channel. The author is not aware of any previous work describing this. The result is a practical device employing the Coanda effect (oscillating or self-oscillating jet flow), which is efficient, easy to manufacture and has higher efficiency distribution of air throughout a room.
B. Ring Vortex Projection The vortices shown in Fig. 2 are generated from air slugs such as would be produced by a piston stroke or the stroke of an acoustic driver, but are far less expensive to produce as they are fan-flow derived. The geometry required is proprietary, but it can be said that the slug of air is then tripped through an orifice plate and turned into a ring vortex.
Fig. 1 Wool tuft enables visualization of self-oscillating wall jet.
NONAERONAUTICAL APPLICATIONS
601
Fig. 2 Ring vortices containing smoke generated from a proprietary vortex generator
.
A ring vortex is able to travel many times the distance of a nozzle discharge because the ring stores kinetic energy like a flywheel for a short time. Ambient fluid is entrained from in front of the ring and transported to the rear and so the result is propulsion with minimal drag. The strength of the ring vortex is purpose tuned and the atomized chemical is transported over a large distance bound within the vortex. Proprietary techniques enable the self-oscillating wall jet to remain attached to one wall longer than on the opposite wall. A useful feature is that as the jet oscillates, one side may be routed through a labyrinthine pathway with walls coated with a chemical that may possess a large surface area for longer interaction time and then returned to the inflow to the fan. Makeup air is venturied into the recirculating main flow within the system. Only the smaller part is ejected as a ring vortex. These ring vortices may contain fragrances or insecticides. They may transport chemicals to foliage in orchards and the turbulence of the ring enables full wetting of each side of the leaves. The chemical may be vaporized by pressure reduction, heating, ultrasound, or any other suitable means. The self-propelled ring vortex promotes whole room circulation because it displaces air at a great distance, which must flow back around the room towards the source. The amount of air in a ring vortex is less than nozzle flow, but with the same system power it is more effective because the nozzle air is unable to travel the required distance and can recirculate back through the fan and so the objective is not achieved.
C. Coanda Vacuum Cleaner One of several versions of this vacuum cleaner is presented here. Figure 3 shows an underside view of the vacuum while operating over glass with flour representing the dirt. Viewing the picture from centrally, a ring of small nozzles is seen. A novel high pressure fan (a Jetfan) drives air through these nozzles, which stirs the carpet pile. Viewed further out is an annular slot blowing air over a Coanda surface. The Jetfan must generate a significant pressure differential on both sides to induce a vortex and a jet simultaneously. This jet entrains dirt and then enters an annular suction slot. The air ascends,
602
T. R. DAY
Fig. 3 Vacuum underside.
but the reduced pressure causes the air to spin while it simultaneously travels medially . This makes it very difficult for particles to ascend as they have to travel inwardly while spiraling. The vortex deposits the dirt into a flexible bag (Fig. 4), which does not collapse onto the low-pressure vortex because an even lower pressure is generated between the bag and the bowl. The vortex flows inwardly to form a central vortex, which then returns through the fan to recirculate. In this way most of the air is recirculated, minimizing the quantity of dirt needing removal by a filter. Some nondomestic versions need no filter. Some other features are proprietary. The Coanda effect and the simple, low cost Jetfan, are the main features of this vacuum cleaner.
D. Coanda Chicken Shed The Beaudesert Shire Council, a local government authority in Queensland, Australia, gave approval for a housing estate near to chicken meat production
Fig. 4 Flexible bag in bowl.
NONAERONAUTICAL APPLICATIONS
603
sheds. Large fans discharge foul air and dust towards the houses, which caused the residents to threaten legal action. The company refused to close down, so the Shire Council explored various ways to solve the problem. Their consultants suggested ducting the discharge horizontally and then vertically to dilute with prevailing winds. That is impractical because of the losses through ducting, especially at the right angle, and the ducting is expensive. The system resistance causes the fan motors to overheat, which may bum out in hot weather or draw excessive current, thereby increasing running costs. The author proposed a solution, as depicted in Fig. 5 , which shows a wool tuft turning 90 deg around a Coanda surface. The difficulty was how to capture the turbulent fan flow onto the Coanda surface, especially when the air speed is relatively low. Once captured, the flow entrains ambient air from the direction of the housing estate instead of blowing towards it. The Shire Council agrees that this technique could be a large part of the solution. The author is negotiating with private enterprise to build these low-cost Coanda surfaces at the end of chicken sheds where there is a need.
E. Coanda Ceiling Fan Figure 6 illustrates a smoke-filled air pathway from top side to underside of a toroidal body. An annular jet exits the top at a certain angle over a step with particular geometry. The jet trips over the step and three counter-rotating ring vortices circle the top side (standing ring vortices). These entrain ambient air and a turbulent flow travels outwardly and circulates to the underside. The jet is the working fluid and that same amount of air reenters the underside peripheral suction slot. The surplus ambient air entrained into the jet on top is shed underneath. By altering underside geometry, shed air can be diffused or alternatively shed as a concentrated plume. The body can be translucent with a circular fluorescent tube inside. Excellent Coanda mixing enhances airconditioned air distribution throughout the room.
F. The Jetfan The Jetfan (Fig. 7) may be a low-cost solution to many applications of the Coanda effect that use fan-generated flow instead of compressor air. The
Fig. 5 Operating proof of concept prototype.
604
T. R. DAY
Fig. 6 Smoke visualizes flow.
following results are for the Jetfan water pump performance and demonstrate the unique characteristics of the impeller compared to other pump impellers; they are highly indicative of similar characteristics for the fan version, that is, no stall without stators or a d i f f ~ s e r . ~ “The visual inspection of the onset of cavitation indicated that over the range of flow rates tested, cavitation first appears at a rotational speed of 3300 rpm. Above this speed, cavitation bubbles were observed to form on the concave su8ace of each blade near the leading edge (LE) and to be reabsorbed a short distance inside the blade passage. The point of reabsorption corresponds to a
Fig. 7 Injection-moldedJetfan.
NONAERONAUTICAL APPLICATIONS
605
line drawn at right angles from the LE of the convex su$ace of the adjacent blade. This reabsorption indicates that the pressure is rising as the water enters the blade passage. In comparing the pe$ormance of the Jetfan water pump with other pump designs, it must be noted that the pe$ormance detailed in this report has been achieved without the use of a complex volute or stator blades, which are commonly used to direct the $ow of water from the rotating impeller into the outlet pipe in many pump designs.” These fans and water pumps are useful in producing fan or pump flows of sufficient power for some CC applications. Potential applications are the NOTAR (Fig. 8) and some other high-flow but lower-velocity CC applications and water applications where high-speed water jets may boil. The Jetfan gives a 60% mechanical efficiency in a 5-in.-diam version with a high static efficiency and enjoys no stators or diffuser. The Jetfan performance is similar to an efficient mixed-flow fan employing a stator row and diffuser. It has a “no stall” characteristic. It has an axial inflow and discharge. Significant static pressure is generated within the blade passageways and by employing no stators and no volute with its tongue or cutwater, wake collisions are eliminated and noise reduced. This may have applications for stealth and even such mundane applications as water pumps for kitchen sinks, and so on, in shipping, including submarines. The Jetfan, including the water-pump version (Fig. 9), is of complex geometry with overlapping blades. These fans and water pumps are, however, able to be made at low cost because a manufacturing method (Fig. 10) has been invented to enable them to autorotate from the tooling and can be made for approximately the same price as any low-cost, injection-moulded impeller. (Note, the Jetfan technology and patents are the property of DBG Investments Ry Ltd.) The same manufacturing method enables axial flow fans with overlapping blades (Fig. 11) to be manufactured at low cost, and metal centrifugal impellers
Fig. 8 NOTAR.
606
T. R. DAY
Fig. 9 Jetfan water pumps.
to be made with high-performance geometry and with the ability to be rotationally extracted from the mould instead of employing investment casting and subsequent milling for precision.
G. Wind Turbines and Orbital Pump Full-span and tip blowing6 is proposed. Wind-tunnel testing has indicated that turbine efficiency increases of 30-40% are likely after all parasitic losses are subtracted. There are two main points here. First, wind turbine power generation is a potential application for CC, which could be revolutionized by a significant increase in efficiency. The author believes this should be explored fully as soon as possible before the world trend toward alternative energy sources, including wind-power, progresses further, thus making it difficult later to retrofit this innovation. Secondly, it is likely the practitioners of CC have discovered that there are few CC applications where adequate air supply can be obtained for control air. The NOTAR is a successful exception. The V22 tilt-rotor exhaust deflection is another good example, but CC there is not, strictly speaking, critical to the aircraft performance. Many proposed applications, including some successfully achieved, are risky, because other aircraft systems may be compromised generally or occasionally. If the only reason that CC development has stagnated
Fig. 10 Manufacturing tooling.
NONAERONAUTICAL APPLICATIONS
607
Fig. 11 Axial fan rotational extraction from mould, enabling blade overlap.
is that of insufficient control air, then CC application to wind turbines would not be likely to be any more successful! It is likely that if CC is to be applied to wind turbines that the problem of inadequate supply of control air be addressed simultaneously. That potential solution may also apply to other uses of the Coanda effect or CC. Therefore, this subject has two elements: 1) considering circulation control for wind turbines and 2) examining the air pump needed to provide the CC air. The basic idea of the Orbitalpump is shown in Fig. 12. It shows how the pins (in some versions) that support the pistons are activated to allow the pistons to change over, one replacing the other. The main features of the Orbitalpump are that it is highvolume, relatively low-speed, low-noise, low-wear, fills and exhausts simultaneously, and can function as either a compressor or high-volume air pump, or both. The Orbitalpump is intended to be the hub of a CC wind turbine (Fig. 13). For most applications, the Orbitalpump shell, being a hollow toroidal body, remains stationary while the shaft is turned. In the case of wind turbines the shaft may be held stationary while the pump body rotates with the blades. The advantage here is that the pressurized air can be fed almost directly into the hollow blades, thus eliminating significant amounts of plumbing and the accompanying losses. It also simplifies air delivery to the blowing slots. The Orbitalpump may be attached to the hub of fans, including CC centrifugal fans. Furey and Whitehead show the results of applying CC to a centrifugal fan. “The better performing combination of these variations was the low solidity (0= 0.65) impeller mated with a reduced internal volume volute. This fan demonstrated a flow rate increase of 100% over that achieved at the design point, through increasing the flow of control air, while maintaining a constant head rise. The peak efficiency of this combination was 83% percent.” Notice the fan achieved a 100% increase in flow over the design point while maintaining head pressure with 83% efficiency.
Fig. 12 Orbitalpump piston changeover.
608
T. R. DAY
Fig. 13 Nylon Orbitalpump.
It is likely that shifting the rear stagnation point and attenuating or eliminating tip vortices in wind turbines and fan blades is as valid as it is for aircraft wings. Applying CC to wind turbines may have other benefits. A smaller diameter wind turbine may achieve the same efficiency as a larger one. This would reduce manufacturing costs, reduce maintenance, and reduce stress on components. It may also enable higher efficiency in areas of lower wind speed. A wind turbine and Orbitalpump combination is now being developed. The Orbitalpump appears to be the highest volume positive displacement pump possible. This high capacity is increased by multistaging on one shaft. Other applications of the Orbitalpump may include a compressor, a pump, a supercharger, a refrigeration compressor, and low-speed, high-volume water pumps. A manufacturing license has been granted to apply small versions for sleep apnea (respiratory support). For CC aircraft applications it can be placed close to the preslot plenum with minimum plumbing.
H. Hovercraft/WIG This model hovercraft/wing in ground effect craft (WIG) is aimed at the hobby market and the entertainment industry. Figures 14 and 15 show existing WIG craft' and Fig. 16 shows the proposed X Hovercraft/WIG. It employs two methods of blowing generally called the Coanda effect. One method is upper surface blowing (USB), where a large mass flow scrubs the upper surface. It also employs CC, which is achieved by a thin wall jet circulating over the rim. In existing USB applications for wings, USB gas may be supplied
Fig. 14 EkranoplanlWIG.'
NONAERONAUTICAL APPLICATIONS
609
Fig. 15 ArnphistarlWIG.’
from the engine nozzle, the jet spreading out to scrub the top of the wing. This USB flow may be induced to coflow with the CC jet around the trailing edge (TE). Similarly, this circular planform employs two annular blowing slots. The more central slot produces “USB” and the more peripheral CC slot entrains the USB flow over to underneath. Small models of 2 ft in diameter cannot carry a compressor and so the peripheral blowing slot is replaced by several suction slots. These suction slots serve to reduce the pressure over the rim and return air to the internal fan (a Jetfan having proved the most efficient). One of several models is shown in Fig. 17 hovering above a table in a still taken from a video. The two wires seen underneath are restraints in case of instability. That particular version employs a SuperTigre 90 model aircraft engine, a tuned pipe, and a Jetfan. The model lifts onto an air cushion by the following mechanisms. The fan (shown in Fig. 18) pumps a large amount of air to scrub the top surface (USB). The suction generated is by Bernoulli’s principle. Ideally, a peripheral CC slot would also blow. In the case of this model, as stated, suction slots are employed instead. This lowers the pressure over the rim and the USB flow circulates to underneath and pressurizes the underside by jet stagnation, which lifts the craft onto an air cushion. Suction slots have been employed before for other applications and otherwise have been suggested by many. Jacques Cousteau’s yacht the “Halcyon,” employed suction slots each side of a metal sail (Figs. 19 and 20) with a reported dramatic increase in thrust (available at http://www.cousteau.org/en/cousteau-world/o~-s~ps/alcyone.php?sPlug= 1). It is claimed that the Turbosail has efficiency 3.5 to 4 times that of a cloth sail. The disadvantage of using suction slots in this manner is that inflow to the fan
Fig. 16 X Hovercraft/WIG.
610
T. R. DAY
Fig. 17 Model hovering.
throat is impeded and so efficiency of these CC sails and of the hovercraft/WIG suffers somewhat. All the proprietq information regarding roll, pitch, and yaw control of the Hovercraft/WIG cannot be presented here. It should be noted that with this particular model although roll control was achieved, pitch control was impaired by asymmetric inflow because of the tuned pipe positioned in the inlet duct, which distorted the underside plate. This caused the model to dip on that side, so a small stay was placed under the edge. As this video was aimed at the movie industry to demonstrate other skills, that stay and a thin wire preventing countertorque were digitally removed. Pitch control, countertorque, and yaw control are achieved, but are not depicted as they are proprietary. The main point here is that a curious result emerged. When weights were placed on the model to test lift, it supported a 100% payload. A paper by
Fig. 18 Top removed, showing fan.
NONAERONAUTICAL APPLICATIONS
61 1
Fig. 19 Coanda sails.
Imber and Rogers’ discussed testing performed on a similar configuration. Imber and Rogers’ work was aimed at other applications such as air and underwater control surfaces, radome scanning sensors, rotor hub fairings on helicopters, marine propellers and aircraft wings that have parabolic tips, and towed underwater arrays. Imber and Rogers showed that by varying positions of azimuthal blowing, they could achieve roll and pitch moments. This was achieved entirely pneumatically. They did not address the issue of counter-torque; however, the author has addressed that with satisfactory results, also achieved pneumatically without any projectin surfaces. Imber and Rogers paper reveals achievement of a) roll control, b) pitch control, c) omnidirectional capability, and d) lift augmentation. In addition, the author shows a) upper surface blowing of high mass flow (USB), b) rim blowing slot (CC) or suction slots or both, c) coflow of USB/CC wall jets, and d) self-contained powerplant and fan. The author has also established propulsion means. These small models have achieved VTOL through a type of surface effect or air cushion. It is well understood that to translate from this hovering/loitering mode into a WIG mode of ground effect travel will require further work and experimentation on larger models. Indeed, if a manned craft is attempted, like any other CC applications, a suitable high-flow pump will need to be found to provide adequate CC air. Perhaps the Orbitalpump will fill that need.
8
Fig. 20 View of TE slot.
612
T. R. DAY
111. Conclusions There are many other important applications for the Coanda effect and CC in addition to aeronautical ones. The Coanda effect has proven to be very effective when applied to the underside of a vacuum cleaner pickup head. This may be one of the first commercial applications. The performance of other smaller domestic appliances may be improved by employing the Coanda effect as it can simplify design and reduce production costs. For example, self-oscillating channel flow eliminates the need for complex and more expensive mechanical and electrical actuators. This in turn allows for ring vortex propagation, which can transport a substance much further than any nozzle discharge employed in small appliances at present, and gives better whole room circulation than present nozzles. Coanda ceiling fans may be far safer than conventional fans. Results suggest that CC may make wind turbines more efficient. The Coanda effect and CC may yield improvements in many industries and applications. The author believes future research should concentrate on developing reliable, lightweight, and low-cost portable sources of blowing air instead of laboratory compressor air. Wind turbine CC blades appear to benefit from the bluff TE as cruise is not needed. CC aircraft wing TE geometry or mechanical factors will need to be improved because of the need for cruise capability. The applications given should stimulate increased interest in solving the very few but important impediments to being able to incorporate the Coanda effect and CC into aeronautical, entertainment, industrial, and domestic applications. Acknowledgments The author is a member of the International Society of Automotive Engineers and is consultant to 1) the entertainment industry producing special effects (including on-stage tornados 22ft high) and 2) industry in fluid movement including Coanda effect applications and ring vortex technology for air-care, insect control, and odor elimination. References ‘Jones, G. S., and J o s h R. D., (eds.), 2004 NASA/ONR Circulation Control Workshop, NASA CP 2005-213509, Mar. 2005. 2 Englar, R. J., “Development Of The A-G/Circulation Control Wing Flight Demonstration Configuration,” David W. Taylor Naval Ship Research And Development Center Bethesda, MD, Jan. 1979. 3Rogers, E. O., Schwartz,. A. W., and Abramson, J. S., “Applied Aerodynamics of Circulation Control Airfoils and Rotors,” 1lth European Rotorcraft Forum, Sept. 1985. 4Murai, K., Kawashima, Y., Nakanishi, S., and Taga, M., “Self Oscillation Phenomena of Turbulent Jets in a Channel,” JSME International Journal, Vol. 30, No. 266, May 1987, pp. 1243-1247. 5Dekkers, W., “Performance Tests on a 93 mm JETFAN water pump,” School of Mechanical, Medical and Manufacturing Engineering, Univ. of Technology, Queensland, Australia, Rept. No. C 2967 (C), Oct. 1998. 6Taylor, R. M., “Aerodynamic Surface Tip Vortex Attenuation System,” US Patent No. 5,158,251, Oct. 27, 1992.
NONAERONAUTICAL APPLICATIONS
613
’Furey, R. J., and Whitehead, R. E., “Static Evaluation of a Circulation Control Centrifugal Fan,” David W. Taylor Naval Ship Research and Development Center, Bethesda, MD, June 1987. ‘Ekranoplans & Very Fast Craft by The University of New South Wales, The Institute of Marine Engineers (Sydney Branch), Univ. of New South Wales (Dept. of Naval Architecture), Australian Maritime Safety Authority, Australian Maritime Engineering CRC Ltd., Russian Australian Advanced Technology Group, Dec. 1996, p. 152 (Amphistar), 154 (Ekranoplan). ’Imber, R. D., and Rogers, E. O., “Investigation of a Circular Planform Wing with Tangential Fluid Ejection,” 34th Aerospace Sciences Meeting & Exhibit, Jan. 1996.
AUTHOR INDEX
Index Terms
Links
A Abramson, J.
69
445
Ahuja, K. K.
167
557
Alexander, M. G.
245
Anders, S. G.
245
Angle II, G.
277
469
B Baker, W. J.
421
Blaylock, G.
383
C Campbell, B. A.
315
Cerchie, D.
113
Chang III, P. A.
445
D Day, T. R.
599
E Ebert, M. P.
445
513
Index Terms Englar, R. J.
Links 23
167
383
557
F Fasel, H. F.
401
Frith, S. P.
337
G Gaeta, R. J.
383
557
Gopalarathnam, A.
499
539
Gross, A.
113
H Halfon, E.
113
Hammerich, A.
113
Han, G.
113
O’Hara, B.
277
Hassan, H.
499
Huebsch, W.
277
I Imber, R.
69
J Johnson, S. K.
245
Jones, G. S.
191
315
357
Index Terms
Links
L Liu, Y. Loth, J. L. LutzTaubert
557 3 113
M Marino, T.
445
McGowan, G.
499
Munro, S. E.
167
539
O Owen, F. K.
105
Owen, A. K.
105
P Paterson, E. G.
421
Paxton, C. D.
293
R Rogers, E. Rumsey, C. L.
69 469
S Sahu, J.
579
Sankar, L. N.
567
513
Index Terms
Links
Slomski, J.
445
Smith, J.
277
Swanson, R. C.
469
Lucie-Trouve
113
V Varghese, P.
113
W Wernz, S.
401
Wood, N. J.
337
Wygnanski, I.
113
X Xiao, X.
499
Z Zha, G.-C.
293
INDEX
Index Terms
Links
A Acoustic optimization, noise reduction and Active flow control (AFC) Advanced CCW airfoils
174 403 40
dual-radius
41
supercritical
41
Aerodynamic heat exchanger (AHE) circulation control and
383
concept of
384
future use of
395
test results
389
aerodynamics
391
heat transfer
392
testing of
386
AFC. See active flow control. AFSF. See anechoic flight simulation facility. AHE. See Aerodynamic heat exchanger. Airfoil development CFD techniques
31
circulation control and
31
Index Terms
Links
Airfoil development (Cont.) cruise configuration
228
high-lift mode
216
Airfoils Bell A821201
279
blowing momentum
110
circulation control concepts and
106
experiments on
107
measurement and analysis
105
numerical simulation and
469
sample results
107
co-flow jet method
294
conventional flap
118
elliptical
144
GACC design
202
NACA 0015 flapped
125
wake turbulence
111
wake velocities
108
Anechoic flight simulation facility (AFSF)
171
Annular wing (CC-duct)
79
model specifications
81
Automobiles, pneumatic aerodynamic technology and
357
Index Terms
Links
B BART, basic aerodynamic research tunnel
207
Basic aerodynamic research tunnel. See BART. Bell A821201 airfoil, Coanda effect on
279
computational model and procedure
282
computational results
286
experiment results
285
experimental apparatus and procedure
279
BLC. See boundary layer control. Blowing coefficient, circulation control stimulation test results and Blowing momentum
525 110
Blowing, boundary layer control, circulation control Blown airfoils, two-dimensional drag
115 200
Blown airfoils, pneumatic flap performance and
200
Boundary conditions, circulation control airfoils and
476
Boundary conditions, FLUENT flow solver and
543
Boundary conditions, steady and pulsed jet effects
560
Boundary layer control
3
Index Terms
Links
Boundary layer control, suction, circulation control (CC) high lift generation history of
3 4
C Cavitation
440
CC propeller
53
CC. See circulation control. CC/jet deflection
51
CC-disc
85
CC-valve
91
CCW airfoils, advanced
40
CCW. See circulation control wing
36
CCW/supercritical airfoils
41
CCW/upper surface blowing (USB) concept
318
CCW/USB, powered lift and engine thrust deflection and CFD techniques
48 31
CFD. See computational fluid dynamics. CFJ. See co-flow jet. Channel wings, STOL aircraft wind-tunnel evaluations and
326
Circular Coanda surface, dual blowing
228
cylinder
405
Index Terms
Links
Circular (Cont.) stopped-rotor aircraft, circulation control and
28
controlled flow and
150
DNS
405
RANS
409
TE
217
wing (CC-disc) specifications
85 86
Circulation control aerodynamic heat exchanger (AHE)
383
airfoil computational fluid dynamics (CFD) concepts development, CFD techniques
106 106 31
flow prediction, turbulence modeling
499
FLUENT flow solver
539
full Reynolds-stress modeling and
445
geometry and grid
472
measurement and analysis of
105
experiments on
107
sample results
107
numerical simulation
469
appendix
497
boundary and initial conditions
476
jet momentum coefficient
478
Index Terms
Links
Circulation control (Cont.) numerical method
475
results
478
turbulence modeling
476
pneumatic flap performance
193
appendix
237
results
216
steady and pulsed jet effects
557
transonic mach numbers test
245
configuration tested
247
facilities used
252
instrumentation used
251
procedures and conditions
253
results of
254
turbulent Coanda wall jet and
415
wake turbulence profile
111
wake velocities
108
blowing
20
blowing momentum
110
circular cylinder, controlled flow
150
co-flow jet (CFJ) airfoil method
294
demonstration of
12
elliptical airfoils
4
experiments
113
elliptical airfoil flow
144
flow control
118
GLAS II airfoil
130
NACA 0015 flapped airfoil
125
Index Terms
Links
Circulation control (Cont.) flight control
337
full-span UAV model
345
half-span model
339
flight testing of
12
Grumman Aerospace A-6A
16
larger aircraft
16
high-lift generation
3
noise reduction
167
nonaeronautical applications
599
hovercraft
608
orbital pump
606
wind turbines
606
pneumatic aerodynamics advanced CCW airfoils
40
airfoil development
31
applications of
28
boundary layer control (BLC)
24
CC propeller system
53
circular cylinder stopped-rotor aircraft
28
circulation control wing (CCW)
36
Coanda effect
25
Coanda, device
26
elliptic-airfoil CC rotor
28
fixed-wing aircraft applications
23
induced drag reduction
54
introduction to
24
Index Terms
Links
Circulation control (Cont.) microflyer and pulsed blowing
56
moment control
54
nonflying applications
57
other aircraft applications
53
powered lift and engine thrust deflection
49
stability augmentation
54
X-wing aircraft
35
rounded trailing edge
4
short take-off and landing (STOL)
4
simulation, GACC wing and
515
boundary conditions
521
computational methods
516
computational resources
523
grid generation
518
initial conditions
521
test results
523
blowing coefficient
525
grid study
530
plenum vs. no plenum
524
technology design capability status
99
workshops annular wing (CC-duct)
79
circular wing (CC-disc)
85
dual-slotted cambered airfoil (LSB)
70
Index Terms
Links
Circulation control (Cont.) dual-slotted low aspect ratio wing (CC hydrofoil)
93
exploratory investigations, NSWCCD
69
miniature oscillatory valve (CC-valve)
91
self-driven rotary thruster (TIPJET) wings, (CCW)
73 36
conventional wings, noise reduction comparison demonstrator design noise reduction, experiments Coanda, ceiling fan
182 5 168 603
Coanda, device
26
Coanda effect
25
Bell A821201 airfoil and
279
computational model and procedure
282
computational results
286
experiment results
285
experimental apparatus and procedure
279
nonaeronautical applications
599
ceiling fan
603
jetfan
606
oscillating channel flow
600
ring vortex projection
600
278
Index Terms
Links
Coanda effect (Cont.) vacuum cleaner
601
slot, setup errors
212
Co-flow jet (CFJ) method
294
advantages of
296
test results
296
energy expenditure
307
F-5E aircraft
308
performance
298
Computational fluid dynamics (CFD)
106
Conventional flap airfoil
118
Conventional wings vs. circulation control wings, noise reduction comparison
182
Cruise configuration airfoil performance and
228
circular Coanda surface, dual blowing
228
pulsed blowing
232
Custer channel wing aircraft
316
D DES. See detached-eddy simulation. Detached-eddy simulation (DES) computational methods, unsteady RANS
421 425
NCCR airfoil computational methods
424
grid generation
427
Index Terms
Links
Detached-eddy simulation (DES) (Cont.) initial and boundary conditions
429
test conditions
424
test results
430
cavitation
440
RANS simulation
430
Direct numerical simulations. See DNS. DNS circular cylinder and
405
direct numerical simulations
403
test calculations, turbulent Coanda wall jet and turbulent Coanda wall jet and
404 402
Drag, pneumatic heavy vehicles and
363
Dual blowing, cruise configuration and
228
Dual-radius CCW
41
Dual-slotted cambered airfoil (LSB)
70
Dual-slotted low aspect ratio wing (CC hydrofoil)
93
E Elliptic-airfoil CC rotor, circulation control and Elliptical airfoil flow
28 4
Equal lift condition
182
Equivalent drag
201
144
Index Terms
Links
Exploratory investigations, circulation control technology workshops, NSWCCD
69
F-5E aircraft, co-flow jet method and
308
F
Flight control circulation control full-span UAV model
345
half-span model
339
wing and
337
Flight testing circulation control and
12
Grumman Aerospace A-6A
16
Flow attachment, STOL aircraft wind-tunnel evaluations and
327
Flow control, conventional flap airfoil, circulation control experiments and
118
Flow prediction, turbulence modeling
499
FLUENT flow solver
539
experiments
541
numerical approach
542
boundary conditions
543
grid details
542
solver settings
543
test results
545
Index Terms
Links
FLUENT flow solver (Cont.) free-air conditions
545
wind-tunnel wall effects
547
Freestream velocity, steady and pulsed jet effects and
566
Fuel economy, pneumatic heavy vehicles and
367
Full Reynolds-stress modeling, best turbulence models
460
circulation control airfoils
445
mathematical development
448
Full-span UAV model circulation control flight control and
345
experiments results
345
G GACC airfoil design
202
BART
207
juncture flow regions
207
solid blockage
206
wake blockage
206
balance limits
208
general aviation circulation control
202
wing, steady circulation control simulation
513
boundary conditions
521
Index Terms
Links
GACC (Cont.) computational methods
516
computational resources
523
grid generation
518
initial conditions
521
test conditions
515
test results
523
General aviation circulation control. See GACC. GLAS II airfoil
130
Grid computational, projectile geometry and
585
creation, steady and pulsed jet effects and
560
details, FLUENT flow solver and
542
generation
427
generation, circulation control stimulation and generation, GACC wing and
518 518
study, circulation control stimulation test results and
530
Grumman Aerospace A-6A
16
H Half-span CCW model, circulation control flight control
339
Index Terms
Links
Heat transfer, aerodynamic heat exchanger and
392
Heavy vehicles (HV), pneumatic aerodynamic technology pneumatic test results
357 360
blown
362
drag increase
363
drag reduction
363
stability and control
365
unblown
361
wind tunnel evaluations
371
High-lift mode baseline performance
216
circular TE
217
TE performance comparisons
226
Hovercraft
608
Hybrid RANS/LES turbulence model, jet-based flow control computer simulation and
583
I Induced drag reduction
54
Initial conditions, circulation control airfoils and
476
Index Terms
Links
J Jet momentum coefficient
478
Jet-based flow control
579
computer simulations
581
hybrid RANS/LES turbulence model unsteady jet boundary conditions
583 582
projectile geometry
584
simulation results
586
nonspinning projectile
586
spinning projectile
589
Jet slot height effects, steady and pulsed jet effects and
567
K Kutta condition
114
L Large eddy simulation (LES)
403
Leading edge blowing, steady and pulsed jet effects and
565
LES. See large eddy simulation. Lower surface blowing. See LSB. LSB, lower surface blowing
70
Index Terms
Links
M Mass flow, pneumatic flap performance and
202
Microflyer and pulsed blowing
56
Miniature oscillatory valve (CC-valve)
91
Moment control
54
N NACA 0015 flapped airfoil
125
NASA, circulation control wings, requirements for NCCR airfoil, detached-eddy simulation (DES) computational methods unsteady RANS
193 421 424 425
grid generation
427
initial and boundary conditions
429
test conditions
424
test results
430
cavitation
440
RANS simulation
430
Noise reduction, acoustic optimization
174
circulation control and
167
circulation control wings vs. conventional wings experiments
182 168
Index Terms
Links
Noise reduction, (Cont.) equal lift condition
182
experiments background information
169
facilities and instrumentation
171
results and discussion
174
technical approach
173
facilities and instrumentation, anechoic flight simulation facility (AFSF)
171
Nonaeronautical applications circulation control and
599
hovercraft
608
orbital pump
606
wind turbines
606
Coanda effect and
599
Coanda ceiling fan
603
Coanda vacuum cleaner
601
jetfan
606
oscillating channel flow
600
ring vortex projection
600
Nonflying applications, circulation control and
57
Nonspinning projectile, simulation results
586
NSWCCD, circulation control technology and exploratory investigations
69
Index Terms
Links
NSWCCD, Naval Surface Warfare Center, Carderock Division
70
Numerical method, circulation control airfoils and Numerical simulation
475 469
boundary and initial conditions
476
circulation control airfoils and, results
478
jet momentum coefficient
478
turbulence modeling
476
O Orbital pump
606
Oscillating channel flow
600
Outboard wing ON, STOL aircraft wind-tunnel evaluations and
322
P PCW. See pneumatic channel wing. PHV. See pneumatic heavy vehicles. Plenum vs no plenum, circulation control stimulation test results and
524
Pneumatic aerodynamic technology automobiles and
357
heavy vehicles (HV) and
357
sport utility vehicles and
374
aerodynamics
Index Terms
Links
Pneumatic (Cont.) basics of
358
boundary layer control (BLC)
24
CC propeller
53
circulation control advanced CCW airfoils
40
airfoil development
31
applications of
28
circular cylinder stopped-rotor aircraft
28
circulation control wing (CCW)
36
elliptic-airfoil CC rotor
28
other aircraft applications
53
powered lift and engine thrust deflection X-wing aircraft Coanda device channel wing (PCW) flap performance airfoil performance
49 35 26 52 193 216
blown airfoils, two-dimensional drag
200
equivalent drag
201
experiments
207
Coanda slot setup errors
212
GACC airfoil design
202
balance limits
208
319
Index Terms
Links
Pneumatic (Cont.) mass flow requirements
202
NASA requirements
193
theoretical considerations
195
heavy vehicles blown test results
362
drag reduction test results
363
fuel economy testing
367
stability and control test results
365
test conclusions
379
test recommendations
380
test results
360
unblown test results
361
wind tunnel evaluations
371
powered-lift super STOL aircraft
315
sport utility vehicles (PSUV)
374
tests on Powered lift and engine thrust deflection
376 49
CC/jet deflection
51
CCW/USB
48
pneumatic channel wing
52
Projectile geometry
584
PSUV. See pneumatic sport utility vehicles. Pulsed blowing, cruise configuration and
232
Pulsed jet effects, test results
570
Index Terms
Links
R RANS circular cylinder and
409
detached-eddy simulation (DES)
421
Reynolds-averaged Navier-Stokes
403
simulation, NCCR airfoil and
430
test calculations, turbulent Coanda wall jet and
405
turbulent Coanda wall jet and
402
unsteady
425
RANS/LES turbulence model, hybrid, jet-based flow control computer simulation and
583
Reynolds-averaged Navier-Stokes. See RANS. Ring vortex projection
600
S Self-driven rotary thruster (TIPJET) Separation control experiments
73 113
Sharp trailing edge, circulation control rounded trailing edge
4
Short take-off and landing. See STOL. Solid blockage
206
Solver settings, FLUENT flow solver and
543
Index Terms
Links
Spinning projectile jet-based flow control
579
computer simulations
581
projectile geometry
584
simulations, results
586
Sport utility vehicles (SUV), pneumatic aerodynamic technology Stability augmentation
374 54
Steady and pulsed jet effects boundary conditions
560
circulation control airfoil and
557
grid creation
560
mathematical equations
559
test results
561
freestream velocity
566
jet slot height effects
567
leading edge blowing
565
pulsed jet effects
570
Strouhal number effects
573
validation of
562
STOL aircraft CCW/upper surface blowing (USB) concept Custer channel wing aircraft
318 316
experiments on evaluation and techniques
320
predictions vs actual
331
Index Terms
Links
STOL (Cont.) wind-tunnel evaluations
321
future configurations
333
pneumatic channel wing (PCW)
319
wind-tunnel evaluations channel wings
326
flow attachment
327
outboard wing ON
322
circulation control demonstrator design STOL, short takeoff and landing
4 6 316
Strouhal number effects, steady and pulsed jet effects and
573
Supercirculation
114
Super-STOL aircraft
315
SUV. See sport utility vehicles.
T TE performance
226
TE. See trailing edge. TIPJET
73
TIPJET rotor specifications
76
TKE. See turbulent kinetic energy. Transonic mach numbers
245
Tests using configuration used
247
facilities used
252
Index Terms
Links
Transonic mach numbers (Cont.) instrumentation used
251
procedures and conditions
253
results of
254
Turbulence modeling
460
Turbulent Coanda wall jet circular cylinder
405
circulation control airfoil
415
DNS
402
RANS
402
test configurations
403
test numerical approach
404
DNS calculations
404
RANS calculations
405
Turbulent kinetic energy (TKE)
479
Two-dimensional drag
200
U Unsteady jet boundary conditions, jet-based flow control computer simulation and
582
Unsteady RANS
425
Upper surface blowing (USB) concept
318
Upper surface blowing. See USB. USB, upper surface blowing
48
476
499
Index Terms
Links
V V/STOL, vertical short takeoff and landing
316
Vertical short takeoff and landing. See V/STOL.
W Wake blockage
206
Wake turbulence
111
Wake velocities
108
WIG. See wing in ground effect craft. Wind turbines
606
Wind-tunnel evaluations pneumatic heavy vehicles and
371
STOL aircraft and
321
wall effects, FLUENT flow solver and
547
Wing demonstrator design, circulation control Wing in ground effect craft. See hovercraft. Wing load reduction, unsteady
91
X X-wing aircraft
35