applications of circulation control technologies

8/19/2019 Applications of Circulation Control Technologies

http://slidepdf.com/reader/full/applications-of-circulation-control-technologies 1/637

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

olume

2 4

PROGRESS IN

ASTRONAUTICS AN D 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 006 by the American Institute of Aeronautics and Astronautics, Inc. Printed in the United

States of America. All rights reserved. Reproduction or translation o f any part of this work beyo nd that per-

mitted by Sections 107 and 1 08 of the U.S. C opyright Law w ithout the permission

of

the cop yright owner is

unlawful.

The

code following this statement indicates the copy right owner’s consent that copies o f articles in

this volume may be mad e for personal or internaluse, on condition that the copier pay the per-copy fee ( 2.50)

plus the per-page fee ( 0.50) through the C opyright Clearance Center. Inc., 222 Rosew ood Drive, Danvers,

Massachusetts 01923. This consent does n ot 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 Copy right Clearance Center:

1-56347-789-0/06 2.50

S

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 AIA A 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

o

Texas at Arlington

Editorial Board

David A Bearden

The Aerospace Corporation

John D Binder

viaSolutions

Steven A Brandt

U S Air Force Academy

Abdollah Khodadoust

The Boeing Company

Richard C Lind

University

o

Florida

Richard

M

Lloyd

Raytheon Electronics Company

Fred R DeJamette Frank Pai

North Carolina State University

University

of

Missouri Columbia

Gail Klein

Jet Propulsion Laboratory

Ning Qin

University

of

Shefield

George Eitalbery Oleg Yakimenko

German Dutch Wind Tunnels

US

aval Postgraduate School

Sanjay Garg Ben T Zinn

NASA Glenn Research Center

Georgia Institute o Technology

Eswar Josyula Peter H Zipfel

US ir Force Research Laboratory

U.S. Air Force Research Laboratory



Foreword

HIS collection of papers represents a compilation of the state of the art in

irculation 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 chap ter delves into new areas and draw s attention

to exciting new ideas in circulation control. wide range of advanced exper

imental 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

HE G ENESIS of this volume originated during the planning of the

NASA

T NR Circulation Control Workshop, which was held March 2004 in

Hampton, Virginia. Over two full days, 3 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 knowledge-

base for circulation control design, analysis, and experimental testing, and

3

to highlight prediction tools for circulation control. Goals 1 and 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 som e 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 chap ter exposes the reader to

some new insights into applications of circulation-control technologies.

Ronald

D

Josl in

Gregory

S

Jones

December

2 5

xix



Table

of

Contents

Preface xix

I Overview

Chapter 1 Advantages of Combining BLC Suction with Circulation

John L Loth. West Virginia University. Morgantown. West Virginia

Control High-Lift Generation 3

Nomenclature 3

Introduction 4

Designing a CC Technology Demonstrator STOL Aircraft 5

1974 Flight Testing

of

the WV U CC Technology Demonstrator 12

1979 CC Flight Tests with a G rumman Aerospace A-6A 16

Conclusions 18

References

2

Chapter 2 Overview of Circulation Control Pneumatic

Aerodynamics: Blown Force and Moment Augmentation and

Modification as Applied Primarily to Fixed-Wing Aircraft

23

Robert J Englar.

Georgia Institute of Technology. Atlanta. Georgia

Nomenclature 23

Introduction 24

Coanda E ffect 25

Applications of Circulation Control. Past and Present 28

Powered Lift and Engine Thrust Deflection 48

Other Aircraft Applications 53

Nonflying Applications of Circulation Control

57

Conclusions 63

References 64

Chapter 3 Exploratory Investigations of Circulation Control

Robin Imber.

Naval Air Systems Comm and. Patuxent River. Maryland;

Technology: Overview for Period 1987-2003 at NSW CCD

Ernest Rogers and Jane Abramson.

Naval Surface Warfare

Center-Carderock Division. West Bethesda. Maryland

69

Nomenclature 69

Introduction

7

ix

http://77898_pref.pdf/

http://77898_01.pdf/

http://77898_01.pdf/

http://77898_01.pdf/

http://77898_01.pdf/

http://77898_01.pdf/

http://77898_01.pdf/

http://77898_01.pdf/

http://77898_01.pdf/

http://77898_02a.pdf/

http://77898_02a.pdf/

http://77898_02a.pdf/

http://77898_02a.pdf/

http://77898_02a.pdf/

http://77898_02b.pdf/

http://77898_02b.pdf/

http://77898_02b.pdf/

http://77898_02b.pdf/

http://77898_02b.pdf/

http://77898_03.pdf/

http://77898_03.pdf/

http://77898_03.pdf/

http://77898_03.pdf/

http://77898_03.pdf/

http://77898_03.pdf/

http://77898_02b.pdf/

http://77898_02b.pdf/

http://77898_02b.pdf/

http://77898_02b.pdf/

http://77898_02b.pdf/

http://77898_02a.pdf/

http://77898_02a.pdf/

http://77898_02a.pdf/

http://77898_02a.pdf/

http://77898_02a.pdf/

http://77898_01.pdf/

http://77898_01.pdf/

http://77898_01.pdf/

http://77898_01.pdf/

http://77898_01.pdf/

http://77898_01.pdf/

http://77898_01.pdf/

http://77898_01.pdf/

http://77898_pref.pdf/



Dual-Slotted Cambered Airfoil LSB) 70

Self-Driven Rotary Thruster TIPJET) 73

Annular Wing CC-Duct) 79

Circular Wing CC-Disc) 85

Miniature Oscillatory Valve CC-Valve) for Unsteady W ing Load Reduction 91

Dual-Slotted Low Aspect Ratio Wing CC Hydrofoil) 93

Status of Design Capability 99

Conclusions 100

References 101

1I.A. Experiments and Applications: Fundamental Flow Physics

Chapter

4

Measurement and Analysis of Circulation

F. Kevin Owen.

Complere Inc. Paczjic Grove. California;

Andrew K Owen.

Control A irfoils 105

University o Oxford. Oxford. England. United Kingdom

Nomenclature 105

Introduction 106

Experimental Details 107

Sample Results 107

Conclusions 112

References 112

Chapter 5 Some Circulation and Separation Control Experiments 113

Dino Cerchie. Eran Halfon. Andreas Hammerich. Gengxin Han. Lutz Taubert.

Lucie.Trouve. Priyank Varghese. and Israel Wygnanski. University of Arizona.

Tucson Arizona

Nomenclature 113

Introduction 114

Discussion of Results 118

Conclusions 162

Acknowledgments 164

References 164

Chapter 6 Noise Reduction Through Circulation Control 167

Scott E Munro. Krishan K Ahuja. and Robert

J

Englar.


Nomenclature 167

Introduction 168

Background 169

Facilities and Instrumentation 171

Technical Approach 173

Results and Discussion 174

http://77898_03.pdf/

http://77898_03.pdf/

http://77898_03.pdf/

http://77898_03.pdf/

http://77898_03.pdf/

http://77898_03.pdf/

http://77898_03.pdf/

http://77898_03.pdf/

http://77898_03.pdf/

http://77898_04.pdf/

http://77898_04.pdf/

http://77898_04.pdf/

http://77898_04.pdf/

http://77898_04.pdf/

http://77898_04.pdf/

http://77898_04.pdf/

http://77898_05a.pdf/

http://77898_05a.pdf/

http://77898_05a.pdf/

http://77898_05a.pdf/

http://77898_05b.pdf/

http://77898_05b.pdf/

http://77898_05b.pdf/

http://77898_06.pdf/

http://77898_06.pdf/

http://77898_06.pdf/

http://77898_06.pdf/

http://77898_06.pdf/

http://77898_06.pdf/

http://77898_06.pdf/

http://77898_06.pdf/

http://77898_06.pdf/

http://77898_06.pdf/

http://77898_06.pdf/

http://77898_06.pdf/

http://77898_06.pdf/

http://77898_06.pdf/

http://77898_05b.pdf/

http://77898_05b.pdf/

http://77898_05b.pdf/

http://77898_05a.pdf/

http://77898_05a.pdf/

http://77898_05a.pdf/

http://77898_05a.pdf/

http://77898_04.pdf/

http://77898_04.pdf/

http://77898_04.pdf/

http://77898_04.pdf/

http://77898_04.pdf/

http://77898_04.pdf/

http://77898_04.pdf/

http://77898_03.pdf/

http://77898_03.pdf/

http://77898_03.pdf/

http://77898_03.pdf/

http://77898_03.pdf/

http://77898_03.pdf/

http://77898_03.pdf/

http://77898_03.pdf/

http://77898_03.pdf/



xi

Conclusions

184

Acknowledgments 186

References 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

191

192

193

195

202

207

216

236

237

241

Chapter 8 Trailing Edge Circulation Control of an Airfoil at

Michael G Alexander. Scott G Anders. and Stuart

K

Johnson.

NASA Lungley

Transonic Mach Numbers 245

Research Center Hampton Virginia

Nomenclature 245

Introduction

246

Instrumentation 251

Facli y 252

Test Procedures and Conditions

253

Test Conditions 254

Discussion of Results 254

Conclusions 263

Acknowledgments

275

References

275

Model Description 247

Chapter 9 Experimental and Computational Investigation into the

Gerald Angle 11. Brian O ’Hara. Wade Huebsch. and James Smith.

West Virginia

Use of the Coanda Effect on the Bell A821201 Airfoil 277

University Morgantown West Virginia

Nomenclature 277

Introduction 278

Experimental Apparatus and Procedure

279

http://77898_06.pdf/

http://77898_06.pdf/

http://77898_06.pdf/

http://77898_07a.pdf/

http://77898_07a.pdf/

http://77898_07a.pdf/

http://77898_07a.pdf/

http://77898_07a.pdf/

http://77898_07a.pdf/

http://77898_07a.pdf/

http://77898_07a.pdf/

http://77898_07b.pdf/

http://77898_07b.pdf/

http://77898_07b.pdf/

http://77898_08.pdf/

http://77898_08.pdf/

http://77898_08.pdf/

http://77898_08.pdf/

http://77898_08.pdf/

http://77898_08.pdf/

http://77898_08.pdf/

http://77898_08.pdf/

http://77898_08.pdf/

http://77898_08.pdf/

http://77898_08.pdf/

http://77898_08.pdf/

http://77898_09.pdf/

http://77898_09.pdf/

http://77898_09.pdf/

http://77898_09.pdf/

http://77898_09.pdf/

http://77898_09.pdf/

http://77898_09.pdf/

http://77898_09.pdf/

http://77898_08.pdf/

http://77898_08.pdf/

http://77898_08.pdf/

http://77898_08.pdf/

http://77898_08.pdf/

http://77898_08.pdf/

http://77898_08.pdf/

http://77898_08.pdf/

http://77898_08.pdf/

http://77898_08.pdf/

http://77898_08.pdf/

http://77898_08.pdf/

http://77898_07b.pdf/

http://77898_07b.pdf/

http://77898_07b.pdf/

http://77898_07a.pdf/

http://77898_07a.pdf/

http://77898_07a.pdf/

http://77898_07a.pdf/

http://77898_07a.pdf/

http://77898_07a.pdf/

http://77898_07a.pdf/

http://77898_07a.pdf/

http://77898_06.pdf/

http://77898_06.pdf/

http://77898_06.pdf/



xii

Computational Model and Procedure 282

Experimental Results 285

Computational Results 286

Conclusions 290

References 291

Chapter 10 Novel Flow Control Method for Airfoil

Ge-Cheng Zha and Craig D Paxton. University of Miami. Coral Gables. Florida

Performance Enhancement Using Co-Flow Jet 293

Nomenclature 293

Introduction 294


Conclusions 311

Acknowledgments 312

References 312

Chapter 11 Experimental Development and Evaluation of

Robert J Englar.

Georgia Institute of Technology. Atlanta. Georgia;

Pneumatic Powered-Lift Super-STOL Aircraft 315

Bryan A Campbell. NASA

Langley Research Center. Hampton. Virginia

Nomenclature 315

Introduction 316

Experimental Apparatus and Test Techniques 320

Wind-Tunnel Evaluations and Results 321

Comparison of Measurements and Predictions 331

Potential Applications 333

Conclusions

333

Acknowledgments 335

References 335

Chapter 12 Use of Circulation Control for Flight Control 337

Steven l Frith and Norman

J

Wood, University of Manchester. M anchester.

England. United Kingdom

Nomenclature 337

Introduction 338

Half-Span Cropped-Delta Model 339

Full-Span UAV Configuration 345

Conclusions 352

Acknowledgments 353

References 353

http://77898_09.pdf/

http://77898_09.pdf/

http://77898_09.pdf/

http://77898_09.pdf/

http://77898_09.pdf/

http://77898_10.pdf/

http://77898_10.pdf/

http://77898_10.pdf/

http://77898_10.pdf/

http://77898_10.pdf/

http://77898_10.pdf/

http://77898_10.pdf/

http://77898_11.pdf/

http://77898_11.pdf/

http://77898_11.pdf/

http://77898_11.pdf/

http://77898_11.pdf/

http://77898_11.pdf/

http://77898_11.pdf/

http://77898_11.pdf/

http://77898_11.pdf/

http://77898_11.pdf/

http://77898_12.pdf/

http://77898_12.pdf/

http://77898_12.pdf/

http://77898_12.pdf/

http://77898_12.pdf/

http://77898_12.pdf/

http://77898_12.pdf/

http://77898_12.pdf/

http://77898_12.pdf/

http://77898_12.pdf/

http://77898_12.pdf/

http://77898_12.pdf/

http://77898_12.pdf/

http://77898_12.pdf/

http://77898_12.pdf/

http://77898_12.pdf/

http://77898_12.pdf/

http://77898_12.pdf/

http://77898_11.pdf/

http://77898_11.pdf/

http://77898_11.pdf/

http://77898_11.pdf/

http://77898_11.pdf/

http://77898_11.pdf/

http://77898_11.pdf/

http://77898_11.pdf/

http://77898_11.pdf/

http://77898_11.pdf/

http://77898_10.pdf/

http://77898_10.pdf/

http://77898_10.pdf/

http://77898_10.pdf/

http://77898_10.pdf/

http://77898_10.pdf/

http://77898_10.pdf/

http://77898_09.pdf/

http://77898_09.pdf/

http://77898_09.pdf/

http://77898_09.pdf/

http://77898_09.pdf/



xiii

1I.C. Experiments and Applications: Nonaerospace

Chapter 13 Pneumatic Aerodynamic Technology to Improve

Robert J Englar. Georgia Institute

of

Technology. Atlanta. Georgia

Performance and Control of Autom otive Vehicles 357

Nomenclature 357

Introduction 357

Basics of Pneumatic Circulation Control Aerodynamics 358

DOE Pneumatic Heavy Vehicle Model Test Results 360

Pneumatic HV Fuel Economy Testing 367

Updated Wind Tunnel Evaluations 371

Pneum atic Sport Utility Vehicles PSUVs) 374

Conclusions 379

Recommendations 380

Acknowledgments 381

References 381

Chapter 14 Aerodynamic Heat Exchanger: A Novel Approach

Richard J Gaeta. Robert J Englar. and Graham Blaylock.

to Radiator Design Using Circulation Control 383


Nomenclature 383

Introduction 383

Technical Approach 386

Results 389

Conclusions

395

Acknowledgments

397

References 397

1II.A. Tools for Predicting Circulation Control Performance:

NCCR 1510 Airfoil Test Case

Chapter 15 Investigation of Turbulent Coanda Wall Jets Using

D N S a n d R A N S 401

Hermann

F

Fasel. Andreas Gross. and Stefan W e n . University

of

Arizona.

Tucson. Arizona

Nomenclature 401

Introduction

4 2

Investigated Configurations 403

Numerical Approach 404

Turbulent Wall Jet on a Circular Cylinder

4 5

Circulation Control Airfoil 415

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_14.pdf/

http://77898_14.pdf/

http://77898_14.pdf/

http://77898_14.pdf/

http://77898_14.pdf/

http://77898_14.pdf/

http://77898_14.pdf/

http://77898_14.pdf/

http://77898_15.pdf/

http://77898_15.pdf/

http://77898_15.pdf/

http://77898_15.pdf/

http://77898_15.pdf/

http://77898_15.pdf/

http://77898_15.pdf/

http://77898_15.pdf/

http://77898_15.pdf/

http://77898_15.pdf/

http://77898_15.pdf/

http://77898_15.pdf/

http://77898_15.pdf/

http://77898_15.pdf/

http://77898_14.pdf/

http://77898_14.pdf/

http://77898_14.pdf/

http://77898_14.pdf/

http://77898_14.pdf/

http://77898_14.pdf/

http://77898_14.pdf/

http://77898_14.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_13.pdf/



xiv

Conclusions 418

Acknowledgments 419

References 419

Chapter 16 RANS and Detached-Eddy Simulation of the

Eric G Paterson and Warren J Baker. Pennsylvania State University.

NCCR Airfoil 421

University Park. Pennsylvania

Nomenclature 421

Introduction 422

Geometry. Conditions. and Data 424

Computational Methods 425

Grid Generation 427

Initial and Boundary Conditions 429

Results 430

Conclusions 441

Acknowledgments 442

References 442

Chapter 17 Full Reynolds-Stress Modeling of Circulation


Peter A Chang 111. Joseph Slom ski. Thomas Marino. Michael P Ebert.

and Jane Abramson. Naval Surface Warfare Center-Carderock

Division. West Bethesda. Maryland

Nomenclature 445

Introduction 446

Mathematical Development 448

Results 453

Conclusions 465

Acknowledgments 465

References 465

1II.B. Tools for Predicting Circulation Control Performance:

NCCR

103RE

Airfoil Test Case

Chapter 18 Aspects of Num erical Simulation of Circulation

R Charles Swanson. Christopher L Rumsey. and Scott G Anders.


NASA Langley Research Center. Hampton. Virginia

Nomenclature 469

Introduction 470

GeometryandGrid 472

http://77898_15.pdf/

http://77898_15.pdf/

http://77898_15.pdf/

http://77898_16.pdf/

http://77898_16.pdf/

http://77898_16.pdf/

http://77898_16.pdf/

http://77898_16.pdf/

http://77898_16.pdf/

http://77898_16.pdf/

http://77898_16.pdf/

http://77898_16.pdf/

http://77898_16.pdf/

http://77898_16.pdf/

http://77898_17.pdf/

http://77898_17.pdf/

http://77898_17.pdf/

http://77898_17.pdf/

http://77898_17.pdf/

http://77898_17.pdf/

http://77898_17.pdf/

http://77898_17.pdf/

http://77898_18.pdf/

http://77898_18.pdf/

http://77898_18.pdf/

http://77898_18.pdf/

http://77898_18.pdf/

http://77898_18.pdf/

http://77898_18.pdf/

http://77898_18.pdf/

http://77898_17.pdf/

http://77898_17.pdf/

http://77898_17.pdf/

http://77898_17.pdf/

http://77898_17.pdf/

http://77898_17.pdf/

http://77898_17.pdf/

http://77898_17.pdf/

http://77898_16.pdf/

http://77898_16.pdf/

http://77898_16.pdf/

http://77898_16.pdf/

http://77898_16.pdf/

http://77898_16.pdf/

http://77898_16.pdf/

http://77898_16.pdf/

http://77898_16.pdf/

http://77898_16.pdf/

http://77898_16.pdf/

http://77898_15.pdf/

http://77898_15.pdf/

http://77898_15.pdf/



xv

Numerical Method 475

Boundary and Initial Conditions 476

Turbulence Modeling 476

Jet Momentum Coefficient 478

Numerical Results 478

Conclusions 495

Acknowledgments 497

Appendix: Coordinates of 103RE Airfoil 497

References 497

Chapter

19

Gregory McGowan. Ashok Gopalarathnam. Xudong Xiao. and Hassan Hassan.

Role of Turbulence Modeling in

Flow

Prediction of

Circulation Control Airfoils 499

North Carolina State University. Raleigh. North Carolina

Nomenclature 499

Introduction

5

Formulation

of

the Problem 501


Conclusions

51

Acknowledgments 510

References

51

1II.C. Tools for Predicting Circulation Control Performance:

GACC Airfoil Test Case

Chapter 20

Warren J Baker and Eric G Paterson. Pennsylvania State University.

Simulation of Steady Circulation Control for the

General Aviation Circulation Control GACC) Wing 513

University Park. Pennsylvania

Nomenclature 513

Introduction

514

Geometry. Conditions. and Data 515

Computational Methods 516

Grid Generation 518

Initial and Boundary Conditions 521

Computational Resources 523

Results 523

Conclusions 536

Acknowledgments 537

References 537

http://77898_18.pdf/

http://77898_18.pdf/

http://77898_18.pdf/

http://77898_18.pdf/

http://77898_18.pdf/

http://77898_18.pdf/

http://77898_18.pdf/

http://77898_18.pdf/

http://77898_18.pdf/

http://77898_19.pdf/

http://77898_19.pdf/

http://77898_19.pdf/

http://77898_19.pdf/

http://77898_19.pdf/

http://77898_19.pdf/

http://77898_19.pdf/

http://77898_19.pdf/

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_19.pdf/

http://77898_19.pdf/

http://77898_19.pdf/

http://77898_19.pdf/

http://77898_19.pdf/

http://77898_19.pdf/

http://77898_19.pdf/

http://77898_19.pdf/

http://77898_18.pdf/

http://77898_18.pdf/

http://77898_18.pdf/

http://77898_18.pdf/

http://77898_18.pdf/

http://77898_18.pdf/

http://77898_18.pdf/

http://77898_18.pdf/

http://77898_18.pdf/



xvi

Chapter

21

Gregory McGowan and Ashok Gopalarathnam.

North Carolina State University.

Com putational Study of a Circulation Control

Airfoil Using FLUENT 539

Raleigh. North Carolina

Nomenclature 539

Introduction 540

Configurations and Experiments 541

Numerical Approach 542

Results 545

Conclusions 552

Acknowledgments 553

References 553

1II.D. Tools for Predicting Circulation Control Performance:

Additional CFD Applications

Chapter

22

Yi Liu. Lakshmi N Sankar. Robert J Englar. Krishan K Ahuja. and

Richard Gaeta. Georgia Institute

of

Technology. Atlanta. Georgia

Computational Evaluation of Steady and Pulsed

Jet Effects on a C irculation Control Airfoil 557

Nomenclature 557

Introduction 558

Mathematical and Numerical Formulation 559


Conclusions 575

Acknowledgment 575

References 575

Chapter 23

Jubaraj Sahu. U S Army Research Laboratory. Aberdeen Proving Ground.

Time-Accurate Simulations of Synthetic

Jet-Based Flow Control for a Spinning Projectile

579

Maryland

Nomenclature 579

Introduction 580

Computational Methodology 581

Projectile Geometry and Computational Grid 584

Results 586

Conclusions 594

References 595

http://77898_21.pdf/

http://77898_21.pdf/

http://77898_21.pdf/

http://77898_21.pdf/

http://77898_21.pdf/

http://77898_21.pdf/

http://77898_21.pdf/

http://77898_21.pdf/

http://77898_21.pdf/

http://77898_22.pdf/

http://77898_22.pdf/

http://77898_22.pdf/

http://77898_22.pdf/

http://77898_22.pdf/

http://77898_22.pdf/

http://77898_22.pdf/

http://77898_22.pdf/

http://77898_23.pdf/

http://77898_23.pdf/

http://77898_23.pdf/

http://77898_23.pdf/

http://77898_23.pdf/

http://77898_23.pdf/

http://77898_23.pdf/

http://77898_23.pdf/

http://77898_23.pdf/

http://77898_23.pdf/

http://77898_23.pdf/

http://77898_23.pdf/

http://77898_23.pdf/

http://77898_23.pdf/

http://77898_22.pdf/

http://77898_22.pdf/

http://77898_22.pdf/

http://77898_22.pdf/

http://77898_22.pdf/

http://77898_22.pdf/

http://77898_22.pdf/

http://77898_22.pdf/

http://77898_21.pdf/

http://77898_21.pdf/

http://77898_21.pdf/

http://77898_21.pdf/

http://77898_21.pdf/

http://77898_21.pdf/

http://77898_21.pdf/

http://77898_21.pdf/

http://77898_21.pdf/



xvii

IV Exploring a Visionary Use of Circulation Control

Chapter 24 Coanda Effect and Circulation Control for

Terence R Day.

Vortex Dynamics Pty Ltd Mount Tamborine

Nonaeronautical Applications 599

Queensland Australia

Introduction 99

Applications 600

Conclusions 612

Acknowledgments 612

References 612

Index 615

Author Index 623

Supporting M aterials 625

http://77898_24.pdf/

http://77898_24.pdf/

http://77898_24.pdf/

http://77898_24.pdf/

http://77898_24.pdf/

http://77898_24.pdf/

http://77898_indx.pdf/

http://77898_aindx.pdf/

http://77898_sm.pdf/

http://77898_sm.pdf/

http://77898_aindx.pdf/

http://77898_indx.pdf/

http://77898_24.pdf/

http://77898_24.pdf/

http://77898_24.pdf/

http://77898_24.pdf/

http://77898_24.pdf/

http://77898_24.pdf/



I. Overview



Chapter 1

Advantages of Combining BLC Suction with

Circulation Control High-Lift Generation

John L. Loth*

W est Virginia University, Morgantown, West Virginia

Nomenclature

CB

=

circulation control blowing efficiency factor

CL

=

lift coefficient

C = blowing coefficient

Di

=

induced drag

D,,

=

parasite drag

CLopt

= optimum lift coefficient where aircraft

L / D

is maximum

mcc

= circulation con trol blowing m ass flow rate

p t

=

total pressure in the compressor bleed air supply duct

q ,

=

dynamic pressure

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

r

=

circulation con trol rounded trailing edge radius

Subscripts

a =

angle of attack

c

= chord

co = free stream conditions

*F’rofessor, Mechanical and Aerospace Engineering. Associate Fellow AIAA.

Copyright 005 by the American Institute of Aeronautics and Astronautics, Inc. All rights

reserved.

3



4

J.

L.

LOTH

I Introduction

HE PU RPOSE of this paper is to present the advantages of combining bound-

T

ry layer control (BLC) by suction with circulation control (CC) by blow ing

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, high-

pressure air to the CC blowing slot involves problems arising from factors

such a s duct size, weight, pressure loss, required insulation and thermal expan-

sion joints, and jet engine take-off thrust loss. It is shown here how adding an

ejector for BL C suction just upstream of the CC blow ing 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 augm entation ratio reaches a maximum.

Flight test results using such a configuration are provided, together with sol-

utions for in-flight transition from the C C 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 D emonstrator Aircraft, at West Virginia University.

The use of blowing air to augm ent airfoil lift had already been proposed',2 in

the

1920s.

D a ~ i d s o n , ~n 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, experi-

mented with C C 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 air foil, computed by S h r e ~ s b u r y , ~re 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

Fig.

1

Com puted stream lines for an e lliptic 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 W illiams,6 then at NSRD C (Naval Ship Research and

Development Center), started to expe riment with CC airfoils developed by Kind.

Williams697 nvestigated 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 trail-

ing 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 com pleted 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 conver-

sion 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



6

J.

L.

LOTH

Fig.

3

Performance comparison between powered h igh-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 two-

dimensional 8 x 10ft 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 F ig.

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 M odel

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 show n are referenced to cruise chord length, which effectively

lowers CLm,, for the Model A in CC mode. The sharp trailing edge wing had a

two-dimensional value of C

= 0.09.

Thus, for curve fitting test data in CC

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 C B=

4.3

pro-

vided 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

0.4 0.6 0.8 1.0

0.2

I

CP

Fig.

6

WVU Model

A CC

wing,

1970

wind tunnel test results.



a

J.

L.

LOTH

5

4

1

0

0 0.2

0.4

0.6 0.8

CU

Fig. 7 WVU Model

A

CC wing em pirical 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 0.012

in.-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 CB eaches 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 D emonstrator.



10

J.

L.

LOTH

Inserting values for slot height gives

3 1

-vs

+-v,

=

1

4 4 or

v,

= 4 - 3 v s

Momentum equation:

tsv,2 t ,v,2 T=

s 1

PVCC

(4)

Inserting values for slot height gives

3 1

4 4

V,2+-V,2-0.5V,2=

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,

=

a

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 C C 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 w eight

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




11

EMPTY WT- 1720 I@

W

BROSS

2400 b

Fig. 10 WVU CC T echnology 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 coo ling purposes. T o 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.

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 adequa te 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

1974 Flight Testing of the WVU CC Technology D emonstrator

Fig. 13 Large position error in cockpit speed indicator against equivalent airspeed

based on boom-mounted pitot tube readings.




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 sea-

level 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:

C =

ACLI

C

=

(4.3

-

2 ) / m

=

6.6

G

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

0

f

U

t

5.6 ’

t.3

2 1

-

20 .

w

3

t

0

1.7

12

L O

.I

OM

-

I I

-

i NOTS

5 3 Sp 473

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

0.002377

76

10/144)

)’”=

78

f t /s

7)

and mass flow rate w ould be

rit

= pAccVcc= 0.002377 x 10/144) x 678 = 0.1 12slug/s

8)

Then the blowing power kinetic energy required is equal to

(9)

V2

2

i z s

= 0.112

x

0.5

x

6782= 25742

ft.lbf/s

= 46.8

hp



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 C C

blowing power required demonstrates the advantage of optimizing the ratio

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 Fow ler flap against the CC pipe w ithout the need for mechanical attach-

ments. 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 alti-

tude loss. Clearly this represents flying on the backside of the power curve, as

V C C / V ~ .

FLAP FOLDING ANGLE /3* w,T:\yE

LAP

ULL

OUT 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 pow er required.




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 com-

putational solutions are being compared. 13- l7 Of particular interest is Englar’s16

“STOL Potential of the Circulation Control Wing for High-Performance Air-

craft.” 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;

) 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.

1.4L - . .

4 o 4 a I Z 1 6 20 2 4 zs

a n degrees

Fig. 19 Wind tunnel test data for wing of Grumman A-6A.



18

4.4

3.6

2.8

>

(CI

w

C

-I

5

2

1.2

0.4

0.4

J. L.

LOTH

0

,, G=,2

C =

a *

I

,

,

,,

f

,

,

,/'. - '

,r

,

,

,

8

,/ ,

'

,

,* /I

,

,/

,

2

,4 8

,

'

/

,

f

,i

IS

,

/

,

I

*'

,

-r

12 20 2E

a

in

degrees

-4 4

Fig. 20 Empirical curve fit to A-6A wind tunnel data shows blowing efficiency factor

Ce

=

6.3.

V. Conclusions

In 1974, the WV U C C Technology Demonstrator STO L was the first aircraft

to demonstrate the high-lift capability of CC . Its wings incorporated an in-flight-

retractable 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 C C 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. A dding 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.




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 coo ling 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 con-

verted 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 trail-

ing 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 m ounted radiators in cars.

4) Lightweight, hot exhaust gas deflectors on helicopter engines in ground

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 developm ents indicate that the future for new CC applications is bright.

effect.

References

‘“Wings with Nozzle Shaped Slots,” NACA Translation TM 521, July 1929 (from

Berichte D er Aerodynamischen Veruchsenstalt in W ien, Vol.

1, No. 1, 1928).

*“The Use of Slots for Increasing the Lift of Airplane Wings,” NACA Translation PW

635, Aug. 1931 Proc eeding s 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,

’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,” NSR DC T N 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.

pp. 170-182.




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 Oper-

ations Meeting, Los Angeles, CA, Aug. 1974.

“Loth, J. L., Fanucci, J. D., and Roberts,

S.

C., “Flight Performance of a Circulation

Control S TO L 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.

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.

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 ST OL Potential for High Performance Aircraft,” AIAA Paper 77-

578,6-8 June 1977.

16Englar,R. J., Trobaugh, L. A., and Hemmerly, R. A., “STOL Potential of the Circula-

tion Control Wing for High-Performance Aircraft,”

Journal of Aircraft,

Vol. 15, No. 3,

”Englar, R. J., Hemmerly, R. A., Moore, H., Seredinsky, V., Valckenaere, W. G. nd

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 STO L Aircraft Design Aspects,” NASA Circulation

Control Workshop, 19-21 Feb. 1986, NASA A mes Research Center, NASA Pub. CP-

”Loth, J. L., Funk, M.

S.,

“Thrust Savings Limitations with B lown High Lift W ings,”

AIAAIAHSIASEE Aircraft Design, Systems and Operations Meeting, St. Louis, MO,

14-16 Sept. 1987.

13

14

1978, pp. 175-181.

2432, pp. 569-588.



24

R. J.

ENGLAR

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

= ratio of specific heats

Sf flap deflection angle

j =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

T mployed 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 deflect-

ing 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 differ-

ence, 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 stagna-

tion 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 pneu-

matic 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:

1) Only very sm all flap size or even nonmoving con trol surfaces were

2)

Lift augmentation could be achieved independent of airfoil angle of attack.

required.



OVERVIEW

OF

CC PNEUMATIC AERODYNAMICS 25

TANQENTIAL

BLOWING

OVER

ROUNDED

TRAlLlNQ EDGE

SURFACE

Fig. 1 Basics

of

circulation control aerodynam ics.

3) Jet turning angle was no longer limited by physical je t 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 w here the use of CC aerody-

namics may be heading. It is by no means a com plete 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 con-

centrates 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 acute-

angle “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 aerody-

namics; 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 tan-

gentially blown jet, with force

augmentation/modification

being m ainly a factor

of je t blowing parameters, not the ang les of the sharp flap trailing edge or the je t

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


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 m omentum coefficient

C as

where the last definition only holds for two-dimensional incompressible

flow ( p j= 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

3

is freestream, R =

1716

ft2/(s2

OR),

and

y

=1.4 for air. A jet

X

U

T H I C K N E S S I C H OR D , tlc

Fig.

3

Maximum lift of blown circulation control airfoils.



28

R. J.

ENGLAR

expansion to the actual static pressure just outside the je t 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 je t 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 stop-

pable-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 circular-

cylinder 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 hyperson ic 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 a t 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 S tates in the late 1960s

as a possible means to increase rotorcraft performance while greatly simplifying

the entire rotor system m echanical hardware. Th e US effort was centered a t the



OVERVIEW

OF

CC PNEUMATIC AERODYNAMICS

29

Navy’s David Taylor Naval Ship R&D C enter (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.



30

R. J.

ENGLAR

equivalent drag in the denominator now includes a severe penalty for compressor

power required as well as for intake ram pressure.lo9l1Maximum equivalent L ID

values roughly 6-7% greater than the conventional unblown rotor blade NAC A

0012 a irfoils (varying only

a

re seen for the 20% CC ellipse (at

a

= 0

deg), but

at a lift coefficient 30% higher, at about 1.3. Furtherm ore, 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

Y

d

SECTIONAL

LIFT COEFFICIENT

C/

Fig. 5 Equivalent efficiencies for CC and conventional two-dimensional airfoils.



OVERVIEW

OF

CC PNEUMA TIC AERODYNAMICS 31

In 1979, a CC R otor 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-

l

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 elim-

inate 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 exper-

imentally 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 summar-

ized in much more detail in Refs. 12 and 16.A typical example of CFD-calcu-

lated 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 num ber 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, je t flow turning to a separation point/stagnation

streamline approximately 130-145 deg from the slot was seen (Fig.

8).

Exper-

imental investigations by the present author of a very similar generic airfoil

(Fig. 9), used surface static pressure, static pressure across the je t, 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

0

t

3

MOMEN TUM C OEFFICIENT, C,

NONDIMENSIONALCHORDMSE 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. T he resulting C, nd C,

distributions are seen in Fig. 9. As Fig. 10 shows, jet turning as high as 170-

175 deg was measured for this airfoil. At a constant C greater turning occurred

with a smaller slot height because the resultant je t velocity and entra inment are

higher a s je t area reduces. Figure

9

(left plot) shows that this greater velocity

and je t 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,,,)s 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) o collapse the different slot height results (left) into a single curve

(right).



34

R. J.

ENGLAR

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

C O E f f l C l E N l S

IEXPANDED

TO

LOCAL WND ITIONS AT SLOT)

Fig.

11

Comparison with momentum coefficients based on local jet exit static

pressure (right) and variation with slot height.



OVERVIEW

OF


35

D.

X-Wing

Aircraft

An extraordinary use of unusual CC airfoils is the X-Wing VTOL

~ e h i c l e , ~ l - ~ ~combined rotary/fixed-wing aircraft equipped with a four-

bladed rotor, which was designed to take off and hover w ith the same nonmecha-

nical cyclic and collective benefits as those already described. How ever, 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 segm ent can actually be negative; the rotor might not be trim-

able in roll, and d rag 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 rotor-

borne 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.

D U A L

PLENUM

AIR FOIL SECTION

HELO DIRECTION

Fig. 12 Dual blow ing on a reverse velocity rotor2’ and blow n 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 con trol systems.

E.

Circulation Control Wing (CCW )

The high-lift capability independent of angle of attack, which was demon-

strated by the C C rotor airfoils above , led to the application of C C as a simplified

very-high-lift dev ice for STO L aircraft. The airfoil in Fig. 1 is representative of

this simplified pneumatic concep t, where both the mechanical TE flap and the LE

flap or slat have been replaced with nonmoving pneumatic systems. Primary

development of the concep t took place in conjunction with C C rotor development

efforts at the Navy's DTNSRDC12,16,26-30n 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. Com pared to results from a family of more conservative blown

flaps (Fig. 15, from Ref.

30),

the CCW profiles showed two significant advan-

tages. 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 C CW 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 dev ice on an A-6/CCW STO L

demonstrator a i r ~ r a f t ~ l - ~ ~n 1979. Flow visualizations in Fig. 16 show a full

180 de g of je t 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 exter-

nally and cross-ducted in the fuselage, where they connected to the high-pressure



OVERVIEW

OF


CONVENTIONAL

Fig.

14

Retractable/storable CCW trailing edges.

bleed ports of the standard J-52-P8A turbojet engine . Results using only availab le

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 30-

35% 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 A d / C C W wind tunnel model at DTNSRDC.

there was sufficient extra

CL

generated to increase the liftable payload by 75 if

the conventional takeoff ground roll distance were used. Also show n 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


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).'6,35,36 n

the flight-tested configuration, the CC blown cylinder was mounted at the TE

of a hinged flap that rotated 180 de g 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 ailer-

ons 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 dow nwash on

the tail reduced trimmed

C

increase to a factor of 1.92, but provided three-

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 a ir ~ r a ft .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 mech-

anical 180-deg rotating flap that stowed in the aft wing cavity, and GTR I 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 high-

pressure 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 G rum man took tw o approaches to the d rag 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 t , ~ ~supercritical-type airfoil was

employed as the baseline because it already had a bluff base thickness between

0.005~ nd 0 . 0 1 0 ~ . CW rounded and semiround (96 deg arc instead of 180 deg)



OVERVIEW

OF


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 T E radius of 0.009~ as 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 gen-

erated sim ilar stall angles to the mechanical slat. One further benefit is the cruise

drag polar (Fig. 22), which is slightly higher unblow n than the baseline supercriti-

cal airfoil, but with very slight blowing can reduce

d

to less than the baseline

while also increasing lift, both at constant

a.

For clarity here, measured

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. 1 0 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 favorab le bound-

ry

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 ~ ) ivots 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 CC W airfoil, the large radius produces an arced

CC aft surface with a turning arc much larger than the flap deflection angle.

MULREWEHTMECHANICAL

~ W TIRFOILS

N D W V H G - P A R T CCWlsUPERCRlTlcALAIRFOIL

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 blow ing for three pneumatic ellipse airfoils '



OVERVIEW

OF


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 com er. 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, w here Cl increases of 35% over the CC W/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-D CW SUPERCRITICALIRFOIL, UA LRADIUS LAPS,

DRAG POLARS, THE PENALTYOR LIFT

Fig.

25

Drag polars of CCW dual-rad ius airfoil at various flap ang les.



44

R. J.

ENGLAR

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

s 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, CE = 0,

C

=

0). This plot confirms the ability of CCW airfoils to generate very

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



OVERVIEW

OF


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 C CW 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; w ith a 20 kn headwind (wind over deck), 200-300 ft

Next Page

http://77898_02b.pdf/

http://77898_02b.pdf/



46

R.

J.

ENGLAR

Fig.

28

Comparison of cruise dual-radius CCW Odeg flap) wi t h mechanical flap

airfoil.

737

WlNOlFLAPICONTROL SYSTEM

Fig.

29

Pneumatic airfoils simplify wing complexity.

Previous Page

http://77898_02a.pdf/

http://77898_02a.pdf/



48

R.

J.

ENGLAR

Angle of Attack,a,deg

Fig. 31 Lift augmentation on the GTRI HSCT/CCW semispan model with blown

IV.

Powered L ift 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 mechan ical complex ity ; thus the CCW/USB c o n ~ e p t ~ ' , ~ ~ - ~ *as

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).

Th e possibility then exists fo r high lift and thrust reversing all

in one system jus t by varying the CC W 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 capa-

bility is far more than mere thrust deflection (i.e., ACL=

CT

[sin a

a] .

t

W R I P B L E M U S T O E F W 3 l D + l

UJETOCC.PLENUHPFESWW

U A R m

Fig. 32 USB and CCW /IJSB pow ered-lift concepts.



OVERVIEW

OF


T = ENQINETHRUST LB)

1 * ANQLE OF AT YXK

q = DYNAMIC PRESSURE FT T IhVi

P, = SLOT PRESSURE

mV,

= SWT MOMENTUM

T = RESULTAKTTHRUST

L

0 CCW ALONE)

25.41LB

486T L5

Fig.

33

CCW/USB model static thrust deflection by blow ing only.

results fro m the increased velocity fro m 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 A me s

with the CCW/USB mounted behind one engine of the NA SA Q uiet 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 oper-

ating 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 tur-

bofan-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.




51

GROSS

WEIQHT. L B l l O m

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 air-

craft 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 exper-

imental 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 eng ine thrust.

BLOWING WY) ENT W,

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 G TRI for

NASA

Langley Research

Center. Called the Pneumatic Channel W ing (Fig. 39), it employs blowing at the

TE of a 180 deg channel (similar to the much earlier but unblown C uster Channel

Wing) to entrain the propeller’s thrust, augment the velocity in the channel, and

thus generate high-powered lift. Figure

40

(from Refs. 5 3 and 54) shows typical

GT RI 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 pre-

dicted takeoff ground rolls of less than 100 ft on a hot day at 3000 ft altitude

using wing angle of attack of only 10deg. Further details of this concept’s capa-

bilities and the associated data are found in the paper by E nglar and Cam pbell in

this volume, and al so in NASA C P 2005-213509.

Fig. 39 Conceptual pneumatic channel wing and semispan model in GTRI MTF

tunnel.




53

a,degrees

Fig. 40 Pneum atic 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

F low Control V a l v e

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

devic es will now be mentioned; many o thers 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, a lso produces greatly enhanced longitudinal pitch stability.

In fact, the A-6/CCW flight demonstrator had such large negative values of

dCM/dC l that the center of gravity of the flight-test aircraft was moved af t by

an additional 10- 15 chord to aid in trim, and the aircraft still had greater longi-

tudinal stability than the conventional A-6, flaps A clever application

of CC W for mom ent control is shown in Fig. 42 on a forward swept wing.59 Pre-

viously, increasing blowing on an aft-swept trailing edge pulled the center of

pressure (CP) outboard and aft, but during this tunnel e~aluation,~’he C P was

made to move outboard and thus forward with blowing. T he amoun t of

xcp

move-

ment 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 mo ment increments (Fig. 43), where only one wing of a CCW configur-

ation is blown. A second innovation is also show n 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 conven-

tional 0.20-chord aileron deflected down 30 deg on the outboard

50

span of



OVERVIEW

OF


55

WEIGHT FLOW-0.25 Iblsec

Fig.

42 CCW

applied to forward swept w ing 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 t ip 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 17-

19

are

seen, with greater percentage reductions at lower CL. Lower aspect

ratio aircraft wings using this technique should yield even greater CD eduction.

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 C C 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 sum -

marize 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 d irectional 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 M icroflyer.60 Jet turning on a small-scale, low-

Reynolds-number wing is illustrated in Fig. 46. Pneumatic lift and control sur-

faces 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




57

c

I

a

CY

SIDE FORCE

COEFF.

0.40

0 .30

0.20

a . 10

0.00

0.10O . ’ O

0.201

1

CN

YAWING

MOMENT

COEFF.

I

I

a

c

t

y1

cn

2

0.10

I 5I

0.05

0.00

-0.0s

0.10

- loo -5O 50 l o c

SIDESLIP ANGLE 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 w ater 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 distri-

bution 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 cor-

nering 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 G T Motorsports team). It i s now possible to

have a multipoint aerodynamic race car design that had previously been prohibited

by the

“nonmoving-aerodynamic-components”

ule. More details of these con-

cepts 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 G TRI program originally intended only to reduce aerodynamic drag on pro-

duction 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



60

R.

J.

ENGLAR

Optional

End

P l a t e

Fig. 48 Blown m odel stern plane design, two-slotted.

FAN INFLOW

Fig.

49

Circulation control fan concept.




61

Fig. 50 Pneumatic Formula 1 car model in G TRI 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 high-

performance cars.

These experimental data for automobiles have now been extended to and

adapted in a GTRI program for the Department of Energy68969o 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 GT RI-patented pneumatic aerodynam ic heat exchanger installed on

formula SAE race car by GT M otorsports Team.



62

R.

J.

ENGLAR

Fig.

52

GTRI P neumatic Futu recar 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

13-

14 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 prom-

ising, 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

Pneum atic heavy vehicle configuration with potential for

5

blowing slots.




63

Fig. 54 Pneumatic heavy vehicle test rig undergoing road tests.

model, p lus 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 con-

firmed that blowing only one vertical side slot at the rear of the trailer can e lim-

inate 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 ground-

vehicle program are found in another paper by Englar in this volume as well

as in NASA CP 2005-213509,2005.

VII. Conclusions

A. Capabilities

The high flow-entrainment capability of tangential blowing over curved aero-

dynamic surfaces has been show n in the preceding discussions to yield augmen-

tation and control of virtually all aerodynamic/hydrodynamic forces and

moments by simplified means, which frequently require no moving external com-

ponents. The capabilities of the CC devices demonstrated include the following:

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.



6

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 pneumati-

cally activate these as needed by the pilot or driver-this is particularly appli-

cable in au tomotive usage.

4) Aerodynam ic mom ent increases from blowing o r differential blow ing 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 pneu-

matic-generated pressure difference can cool the vehicle while controlling aero-

dynamic forces and moments.

B. Future of Circulation Control

The preceding capabilities offer the potential

for aerodynamiclhydrodynamic

vehicles simplified by pneumatic multipurpose sui aces synergistically augm ent-

ing l i f , 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 per-

formance, 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 aerodynam ic

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 begin-

ning. 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 Extreme-

STOL 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

’“Wings with Nozzle Shaped Slots,’’ NACA Translation, T M 521, July 1929; Berichte

635, Aug. 1931 (Proceedings L’Aeronautique, June 1931).

Der Aerodynamischen Vereuchsenstalt in Wie n,

Vol. 1, No. 1, 1928.




65

3Metral, A. R., “On the Phenomenon of Fluid Veins and their Application, the Coanda

Effect,”

A F

Translation, F-TS-786-RE, 1939.

4Sproule, R. S., and Robinson, S. T., “Combined Intelligence Objective Sub-

Committee 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. FP-

188),” 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.

Englar, R. J., “Circulation Control Pneumatic Aerodynamics: Blown Force and

Mom ent Augm entation and Modification; Past, Present and Future,” AIAA Paper 2000-

2541, presented at AIAA Fluids 2000 Meeting, June 2000.

“Englar, R. J., and Williams, R. M., “Test Techniques for High Lift Two-D imensional

Airfoils w ith Boundary Layer and Circulation Control for Application to R otary Wing A ir-

craft,” Canadian Aeronautics and Space Journal, Vol. 19, No. 3, 1973 pp. 93-108.

Englar, R. J., “Two-Dimensional Subsonic Wind Tunnel Tests on a Cambered 30-

Percent-Thick Circulation Control Airfoil,” NSRDC, Technical Note AL-201, AD 913-

41 lL , May 1972.

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,” A merican Helicopter Society, Paper AHS 79-5 1, May 1979.

14Mayfield, J., “Aeronautical Engineering-Navy Sponsors Coanda R otor 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.

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 Con trol 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.

9

11

12

19



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., Schw artz, A. W., and Abram son, 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 Circu-

lation 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


74-502, June 1974.

27Englar,R. J., “Subsonic Two-Dim ensional Wind Tunnel Investigations of the High Lift

Capability of C irculation Control Wing Sections,” DTSNRDC, Rept. ASED-274, April 1975.

”Englar, R. J., “Circulation Con trol for High Lift and Drag Generation

on

STOL Air-

craft,”

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 S TO L Potential for High Performance Aircraft,” AIAA Paper 77-

578, 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.,nd

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,

32Englar,R. J., “Developm ent 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.

Englar, R. J., “Low-Speed Aerodynamic Characteristics of a Small Fixed-Trailing-

Edge Circulation Control Wing Configuration Fitted to a Supercritical Airfoil,”

DTNSRDC, Rept. ASED-81/08, March 1981.

38Englar,R. J., and Huson,

G. .

Development of Advanced Circulation Control Using

High-Lift Airfoils,” AIAA Paper 83-1847 July 1983 ; also published in

Journal ofAircraft,

39Carr, J. E.,

“An

Aerodynamic Comparison of Blown and Mechanical High Lift Air-

foils,” AIAA Paper 84-2199, Aug. 1984.

40Englar, R. J., “Two-D imensional Transonic Wind Tunnel Tests of Three 15-Percent-

Thick Circulation Control Airfoils,” NSRDC, Technical Note AL-182, AD 882-075,

Dec. 1970.

1981, pp. 51-58.

37

V O ~ .1, NO. 7, 1984, pp. 476-483.




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.

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 97-

0036, Jan. 1997.

43Mavris, D. N., K irby, M. R., Lee, J. M., Q ui,

S.,

Roth, B., Tai, J., and Englar, R. J.,

“Systems Analyses of Pneumatic Technology for High Speed Civil Transport Aircraft,”

GTR I Final Technical Rept. A-5676, Oct. 1999.

44Ni~ho l s ,. H., Jr., Englar, R. J., Ha m s, M. J., and Huson,

G. .

Experimental Devel-

opment of an Advanced Circulation Control Wing System for Navy STOL Aircraft,”


45Harris, M. H ., Nichols, Jr., J. H., Englar, R. J., and Huson,

G. .

Development of the

Circulation Control WingIUpper Surface Blowing Powered-Lift System for STOL Air-

craft,” 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 CC W /U SB 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. “C irculation

Control Technology Applied to Propulsive High Lift Systems,” Society of Automotive

Engineers, Paper 841497, Oct. 1984.

48Lowndes, J. C., “Aeronautical Engineering: Stud ies Show Lift Coefficient Tripling,”

Aviation Week and Space Technology, Dec. 1, 1980.

49Englar, R. J., Nichols, Jr., J. H., Ham s, M. J., Eppel, J. C., and Shovlin, M. D., “Devel-

opment 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 Air-

craft,” 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,” NA SA, Technical Paper 2684, Jan. 1987.

52Bevilaqua,P. M., and Lee, J. D., “Design of Supersonic Coanda Jet Nozzles,” in Proceed-

ings of the Circulation Control Workshops 1986, NASA, CP 2432, pp. 289-312, Feb. 1986.

53Englar, R. J., and Campbell, B. A., “Developm ent of Pneum atic Channel W ing

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 Developm ent and Evaluation of

Pneumatic Powered-Lift Super-STOL Aircraft,” NASAIONR Circulation Control Work-

shop, March 2004; also published in NA SA C P 2005-213509, 2005.

55Braslow, A. L., “Aerodynam ic Evaluation of Circulation Control Propellers,” Bio-

netics 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.

57Gam er, D., “No Moving Parts, The Circulation Control Airfoil and Fluidic Propeller,”

EAA

Sport Aviation,

Vol.

37, No. 3, 1988, pp. 27-30.

31, NO.

5

1994, pp. 1160-1177.



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.

Englar, R. J., “Pneumatic High-Lift and Control Surfaces Applied to Micro-Aerial

Vehicles,”

Proceedings

of

GTRI International Conference on Emerging Technologies

f o r

Micro

Air

Vehicles,

Feb. 1997.

610 yle r, T. E., and Palm er, W. E., “Exploratory Investigation of Pulsed Blowing fo r

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.

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 W illiams, R. M ., “Design of a Circulation Control Stem Plane for

Submarine Applications,” NSRDC, Technical N ote AL-200, M arch 197 1.

6 5 F ~ ry , . J., and Whitehead, R. E., “Static Evaluation of a Circulation Control Centrifu-

gal Fan,” DTNSRDC, Rept. 77-0051, AD A041-463, June 1977.

aeta, R. J., and Englar, R. J., “Pneumatically Augm ented Aerodynamic Heat Exchan-

ger,” Paper presented at NASA/ONR Circulation Control Workshop, March 2004; also

published in NASA C P 2005-213509, 2005.

67Englar, R. J., Smith, M. J., N iebur, C. S., and Gregory, S. D., “Development of Pneu-

matic 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 S AE SP-1145, “Vehicle A erodynamics,” Feb. 1996.

68EnglarR. J., “Drag Reduction, Safety Enhancement and Performance Improvement for

Heavy Vehicles and SUV s Using Advanced Pneum atic Aerodynamic Technology,” 2003

SA E International Truck and Bus Meeting and Exhibition, Society of Automotive Engin-

eers, Paper 2003-01-3378, Nov. 2003.

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.

60

TR-32, July 1972.

69



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

A R =

aspect ratio

C , = momentum coefficient of slot flow ( r i zv j /qS)

C L

=

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 ( p A V )

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.

overnment and is not subject to copyright protection

in the United States.

69



7

R.

IMBER, E. ROGERS, AND

J.

ABRAMSON

V = velocity

a =

angle of attack

p

=

density

u

=

rotor solidity, cavitation index

ACL/AC, = lift augmentation ratio (slope of

CL

vs C, curve)

Subscripts

=

induced

= j e t

I. Introduction

EGINNING in 1967, when the Naval Surface Warfare Center, Carderock

B

Division (NSWCCD) was known as the David Taylor Model Basin

(DTMB), researchers there were involved with many circulation control (CC)

exploratory projects, including C C-airfoils, CC-centrifugal fans, dual-directional

CC-airfoils, CC-fixed wings, including the A6 aircraft modification, CC-rotor-

craft, 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 Adminis-

tration, Ames Research Center, in 1986.' Papers, presentations, and reports of the

research performed from 1967 to 1985 at what is now NSW CCD (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 pre-

sented at the 2004 Circulation Control Workshop held in Hampton, V irginia.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, CC-

Hydrofoil. For further details regarding these investigations, the reader is encour-

aged to examine the publications that are referenced for each of the projects.

Used throughout this summary is a frequently used measure of CC perform-

ance, the lift augmentation ratio. This ratio is defined as the ratio of the gain in

lift

A C L )

o 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.

11.

Dual-Slotted Cam bered Airfoil (LSB)

The Dual-Slotted Cam bered 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 su rface slot along with the usual upper surface slot provides

the ability to produce lift in both positive and negative d irections. 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 sam e 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 m odel is shown in Fig. 1 and a cross-section draw ing

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 0 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 sam e cross-section as the parent m odel. Lift perform-

ance results from the single and dual slot models are shown in Fig. 3. The com-

parison 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 d irections was available. F igure 4 displays a plot of lift coefficient

Air Supply Ducts

Fig.

2

Cross-section of LSB airfoil

Upper Surface

Blowing Slot

Coanda

Surface

Lower Surface

Blowing (LSB)

Slot



72

R. IMBER, E. ROGERS, AND J. ABRAMSON

ACr

due to

blowing

Momentum Coefficient, Cp

Fig. 3 Comparison of lift performance for dual-slottedLSB 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 comp ared with 60). This empirically unexpected performance enhance-

ment 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 ~ a p a b il it y .~



INVESTIGATIONS OF CC TECHNOLOGY AT NSWCCD

73

where the low er slot flow momentum level was 25 that of the upper slot, acti-

vation 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 A irfoil inves-

tigation, it was found that 1) incorpora tion of a lower slot did not affect perform-

ance 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 NSW CCD engineers, the Self-

Driven Rotary Thruster was the first integrated lift/reaction-drive rotor system

combining Cbanda CC aerodynamics with cold cycle reaction drive technol-

ogies.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 com pletely pneum atic rotor. Circulation control

slots are located along most of the rotor blade span on both the leading and trail-

ing 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

Fig. 5

TIPJET

vertical takeoff and landing unm anned air veh icle with circulation

controlled stoppable rotor.6



74


No drive shaft, unarticulated, flat-pitch blades

NOZZLE

AIR

SUPPLY

DRIVE

FORCE

Fig.

6 Sketch of TIPJET ompletely 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 p lace 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 80-

in. 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 experimen-

tal performance. The results of this comparison, show n in Fig. 10, indicate excel-

lent correlation for both the rotor thrust developed and drive power required.

The ultimate goal of the experimental roto r 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




75

SLOT HEIGHT ADJUSTMENT

SIDE-B Y-SIDE OPPOSING

SCREWS

Fig. 7 Draw ings with d etails of

TIPJET

pneum atic rotor?



76

R.


J.

ABRAMSON

Table

1

TIPJET rotor specifications’

Blade

Fixed pitch angle

OJ,

deg

Rotor diameter, ft

Number of blades

Chord, in.

25 span

Solidity ratio

Geometric twist, deg

93 span

Airfoils

0

6.67

2

7.95

5.40

0.110

0

25 span 93 span

Thickness ratio

( t / c )

0.213 0.170

Trailing edge radius (rt, /c)

0.05 0.03

Slot height

( h / c )

variable variable

Area/nozzle, in.* 0.764

Camber ratio d / c ) 0.053 0.01 1

Tipjet nozzles (rectangular)

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 on e of the ob jectives

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 m echanically simple, easily con-

trolled 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-drive-

nozzle 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


D

c

Fig. 1 TIPJET performance numerical analysis correlation with experimental

results.'

cc

slot Slot

flow

area

t

I

Internal air

r

otor

Drive

rotor

drive

torque

Fig. 11 Conceptual schematic of TIPJET rotational rate equilibrium.




79

a,

a,

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.


J.

ABRAMSON

Fig.

14

Annular wing m odel 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 C C-Duct is shown in Fig. 15 and its geometry

is presented in Tab le

2.

The model is 18411. in d iam eter 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.




81

Dimensions n inches

Fig.

15

Annular wing m odel 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 trail-

ing ed ge 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 photo-

graph on the right of Fig. 16was taken fo r a com plimentary halves configuration,

where the ou ter lower slot is active and the upper inner slot is active. The ya m

strand is now at an angle to the free stream, indicating the w ake 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 m odel specifications

Model geometry Dimensions, in.

Outside diameter 18.2

Inside diameter 14.2

Chord

10

Slot gap

0.009

h l c

0.0009

Slot position 0.970c

d/c (16.2110) 1.62

AR

(effective) 2.1



82

R.


J.

ABRAMSON

Table

3

CC annular wing modes of operation as a m arine propulsor duct

Aft view of Slot ejection

ducta configuration Effect Operation benefit

Inner slot only Increased duct Higher prop efficiency

flow-through: accelerating

nozzle

Outer slot only Decreased duct flow: Reduced cavitation

diffusion

Complementary Side-force: yaw Steerage

quadrants

Complementary Side-force: pitch Depth keeping

quadrants

Alternating Vortex generation:

very

high Braking: crash-back

drag,

no

side force

Both slots Drag reduction, auxiliary Cru ise efficiency, dock

thruster side positioning

aDashed line is trailing edge; solid lines represent active slot.

for the force coefficients is duc t 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

T 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

C L

that should be

compared to that of an airfoil is half the

CL

shown in Fig. 17, and thus




83

No Blowing 180 I 180 deg blowing

Fig. 16 Lateral force capability; wake deflection with asymm etric 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

or the geom etry of the duct. (See the later discus-

sion and performance of the CC-Hydrofoil flat wing, which has about the same

aspect ratio.)

Slot Flow Momentum Level

CF =

mV,/(

pVm2 dc)

Fig. 17 CC annular wing lift and drag performance' (the drag force is in the

direction of a negative propulsive force).



84

R.


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 counter-

rotating vortices. Figure 1 8 shows the measured performance demonstrating

this capability. The drag is about the same as it was when lift was being devel-

oped. 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?




85

Wake Filaments looking upstream

Fig.

19

Braking configuration com putational 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 devel-

opment 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 CC-

enhanced 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 mom ent 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



8

R.


J.

ABRAMSON

Fig. 2 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) con-

stant 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.

Resu lts from the first configuration group are show n in Fig. 23 for the model at

zero pitch angle, gradually increasing the flow ejection circumference region.

Lift is show n 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

iameter (chord)

2

ft

Thickness ( t l c ) 19

Reference area

(S) 3.14

ft2

Aspect ratio AR) 1.27

Camber

2.4

Coanda radius

r s / c = 0.050

r t e / c= 0.040

Slot position

3.2

from edge

Slot lip thickness

0.026

in.

Slot height ( h ) 0.032 in.

0.0013

0.027

h l c

hlrs

Pressure tap diam eter 0.040 in.




87

Surface Pressure Taps

Centerline Cross-section

Fig. 21 Circular wing model with full perim eter 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 configur-

ation at several

AOA

and C, levels. The m aximum C, was more than twice that

available fro m a disc without circulation control. Both w ith 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 Cen tered Abou t the Trailina edae:

Fig. 22 Circular wing test configurations: planform view (sections with blowing

shown in black).



aa R.


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 w ings, there are no basic effects unique to lift developm ent by m eans

of the Coanda form of CC.

Angle-of-Attack (deg)

Fig. 24 Influence of angle of attack?




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 m ap resulting from two aerodynamic centers?



90

R.


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.

Fo r 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. Th e area of fluid ejection

is easily seen as the darker shaded regions around the perimeter, which have sig-

nificant 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 character-

istic 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, fo r a circular planform, can provide an om nidirectional capa-

bility 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 je t turning);

3)

roll con trol was demonstrated using asymmetric

blowing; 4 ift 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.




91

High

pressure

Low

pressure

Full perimeter blowing

Fig. 28 Upper surface pressure resu lts 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 w as 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 dem-

onstration is shown in Fig. 29. The flow ejection for this application was normal

to the surface, as in a je t 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

electromagnet ic

actuator

Fig. 3 Trailing edge of dual-slotted hydrofoil: rotor pivots to open /close upper

lower slots. Constant fluid pressure results in high-response-rate, efficient system.

4

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 pro-

peller 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

Actuator Rotor

Magnet

Coil

Valve

Assembly

Assembly Assembly

Fig. 31 Actuator mechanism constr~ction.'~




93

24-inch

water tunnel

Fig. 32 Operational schematic of actuator test.

During operation, fluid pressure to the trailing edge of the foil is

ot

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 attrib-

uted to a Gurney flap. (The term “low-rate’’ ejection is used to m ake a distinction

from the high momentum level of a true jet flap.)

As an example of the experimental results, two plots of force against fre-

quency are show n in Fig.

33.

The top plot shows the periodic force spectrum pro-

duced 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 fol-

lowing: 1) functional model-scale actuators can follow steady or time-varying

input signals up to 500Hz; hydrofoil forces were successfully varied up to

110Hz; nd 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 ques-

tion as to the effect of cavitation on the performance of a CC-foil had been pon-

dered for many years and was finally answered during this research.



94


Frequency (Hz)

Fig. 33 Load cancellation effectiveness. Two plots show three frequencies of

unsteady lift simultaneously reduced to the broadband fl0 0r. l~

There are several compelling reasons to incorporate circulation control fluid

dynamics on underwater vehicles. T he buoyancy of waterborne vehicles means

that they can and d o 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 m uch greater force than otherwise possible, it becomes an attractive con-

sideration for low-speed m aneuvering enhancement. Furthermore,

CC

augmenta-

tion has its best pumping-power efficiency at low speed, in terms of the control

force advantage over a conven tional 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 three-

dimensional (finite wing) performance. The slot-height-to-chord ratio of approxi-

mately 0.0018 was maintained ove r 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 per-

formed 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 square-

root-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

95

Fig. 34 Dual-slotted low aspect ratio circulation control hydrofoil.'6

Fig. 35 Dual-slotted hydrofoil installed in the L arge Cavitation Channel.



96

R.


J.

ABRAMSON

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 je t was causing the loss in lift. The lower slot was then em ployed to

produce a very sm all 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 perform ance 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 two-

dimensional airfoil data. This is the same ratio of three- to two-dimensional per-

formance a s found for the C, vs AOA on conventional wings of the sam e 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 con-

ventional 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 corre-

sponding to the vaporization of water, about 0.5 psia depending on temperature.

The cavitation index, sigma

a),

s the term for the absolute value of the pressure

coefficient that will result in vaporization and is a function of the test section

static and dynamic pressures.

0

0.5

1

1.5

2

2.5

3

3.5

0 0.1 0.2 0.3 0.4 0.5 0.6

with 5% lower slot Cµ

no lower slot flow

AOA = 10°

R52

w i t h 2 n d s l o

t f l o w

single slot

CL

(max conventional)

Cµ total (upper + lower)




97

Slot Momentum Coefficient total), 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 con-

tinued to increase in response to increasing duct pressure. Eventually the lift

began to roll over, but not abruptly. At no time did the Coanda je t detach

prematurely from the trailing edge due to cavitation. For this particular model

CP

Fig. 38 Lift response to Coanda surface cavitation developrnent.I6

0 0.05 0.1 0.15 0.2 0.25 0.3

CDCD

CL

CD

0

0.5

1

1.5

2

2.5

3

3.5

airfoil data

wing data

CD

drag pred.

0= AO A o

tols-dn2/w

prediction

prediction

CD

Slot Momentum Coefficient (total), Cµ

G3

0

5.0

1

5.1

2

0 50.0 1.0 51.0

σ 5.31=

2.01

6.6

nonoitativacf otesno

r of ecaf r usadnaoC

σ 6.6=

°0= AO A

tolselgnis

Cµ

CL

= pC nim ezir opavdluowr etawhcihwta

5.31=amgis

2.01=amgis

6.6=amgis



98

R.


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 CC-

Hydrofoil, compliments of the cavitation that caused the white bubbles of vapor-

ized water. Cavitation is not likely to occur operationally, bu t if it does, it is not

catastrophic to the fundamental C C effect.

Another advantage of the dual slots is the ability to vector the je t 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 an d 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 me rger of the two slot flows. The bottom set of images represent

a con tinued increase in the upp er slot pressure until it equa ls the lower pressure




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 tw o 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 m ode may be viable as a je t 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 his-

torical 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.

W hereas 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 fo r viscous-based com-

putational codes to guide the subtle design details of the Coanda trailing edge

region, as well as the upstream contour. Until a CF D code is available that has

been properly and thoroughly validated for sensitivity to CC contour changes

R



100

R.


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 perform-

ance 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 vali-

dation efforts.

IX.

Conclusions

The re have been many diverse experimental CC investigations at NSW CCD

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 compu-

tational codes for the evaluation of various proposed app lications.

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 w ere 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 com-

pared with a single-slot configuration, or, as in the case of the miniature oscillat-

ing valve, unsteady load reduction. Tw o 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 perform-

ance. 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 pre-

sently contemplated hydrodynamic applications.

Four of the investigations were low aspect ratio geom etries and these helped to

extend the understanding of the viability of CC on short-span surfaces. The suc-

cessful omnidirectional demonstrations of the CC-Disc suggest future application

to very maneuverable low

R

vehicles or to appendages. It is helpful to know

that analysis of the low

R

nvestigations 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 aug-

mentation. The findings from the TIPJET hover test were significant in furthering

rotary-wing CC knowledge and in demonstrating a novel application of CC .




101

In conclusion , the six diverse exploratory applications presented all met basic

performance expectations, suggesting a certain degree of maturity of this technol-

ogy. These experimental results, plus those reported by other organizations, now

make for a rather extensive body of know ledge on the subject, ready to encourage

additional creative applications and to support a variety of full-scale applications.

Com plementing this broad range of empirical knowledge, the anticipated emer-

gence of thoroughly validated viscous-based two- and three-dimensional compu-

tational codes for CC w ill contribute to the achievem ent 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.

Englar, R. J., and Applegate, C. A., “Circulation Control-A Bibliography of

DTNSRDC R esearch and Selected Outside References (Jan. 1969-Dec. 1983),”


31mber, R. I., “Exploratory Investigations of Circulation Control Technology: Over-

view for Period 1987-2003 at NSWCCD,” NASAICP-2005-213509, Proceedings of

the

2 4 NASAIONR

Circulation Control Workshop, compiled by

G.

S. Jones and

R. D. Joslin, March 2005.

4Abram son, J.

S.,

and Rogers, E.

O.,

“Design of a Circulation Control Airfoil Having

Both Upper and Lower Surface Trailing Edge Slots (Model LSB17),” DTNSRDC/TM-

16-86/03, Sept. 1986.

’Abramson, J., “Characteristics of a Cambered Circulation Control Airfoil Having Both

Upper and Lower Surface Trailing Edge Slots,’’NSW CCD-50-TR-2004/030, April 2004.

6Reader,

K. R., Abramson, J. S., Schwartz, A. W., and Biggers, J. C., “Tipjet VTOL

UAV Summ ary: Volum e 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 U AV: 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 Con-

trolled Shrouded Propeller,” West Virginia Univ., Morgantown, WV, TR-39, Feb. 1974.

“Hoerner, S. F., and Borst, H. V. (ed.), Fluid-Dynamic

Lif t ,

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 W ing with Tangential

Fluid Ejection,” 34th Aerospace and Sciences Meeting and Exhibit, AIAA 96-0558, Jan.

1996.

13Fry,D. J., and McG uigan, S., “Hydrofoil Circulation Control Via a M iniature Valve for

Alternating Flows Between Tw o Exit Slots,’’ CDNSW CISH D-1401-02, Dec. 1993.

2



102 R.


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. 7 5 , 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.

14



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

C L infinite aspect ratio lift coefficient

C

blowing momentum coefficient

U

mean axial velocity

U

edg e velocity

u rms velocity fluctuations

x

streamwise position

y

crosswise position

airfoil angle of attack

i induced flow angularity

eff airfoil effective angle of attack

*Consultant.

'Research Assistant. Department of Engineering Science.

Copyright 005 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

IRC UL AT ION control (CC) airfoil concepts have been studied extensively

C

or 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.

The se rounded trailing edges allow the rear stagnation point to move. This move-

ment 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, r emains attached to the airfoil.

At low blowing rates, this Coanda effect entra ins 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 je t controls the location of the airfoil stagnation points, and therefore

the circulation and lift. However, eventually there c om es a point where there is no

longer a balance between the static pressure and centrifugal force and je t 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 dev ices, because they only have to overcom e

the viscous losses in the flow. By compensating for the viscous losses, the flow

field more closely resembles the ideal inviscid case. Accordingly, lift augmen-

tation 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.

Ow ing to the strong adverse pressure gradients o n the airfoil upper surface, strong

secondary flows can be generated in the sidewall boundary layers. Th e 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



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 devel-

opment 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 symm etric 6-in. chord airfoil, 20 thick, 3 cam ber ellipse with a nom-

inally 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 je t exit flow. The je t exit velocity was calcu lated from isentropic relation-

ships referenced to tunnel static conditions.

There w ere a total of 9 1 pressure taps on the model, 5 9 of which were posi-

tioned 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 sampl-

i n g ~ a p a b i l i t y . ~he 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, turbu-

lent 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. 1and 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



MEASUREMENT AND ANALYSIS OF CC AIRFOILS

109

01

0 0.005 0.01 0.015 0.02 0.025

0 03

0.035

Blowing Mom entum 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 angular-

ity in the freestream flow ahead of the model, thus ch anging 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 0.03 0.035

Blowing Mo mentum Coefficient, C

Fig. 4 Infinite AR lift coefficients.



110

F. K. OWEN AND A. K. OWEN

0 0.005 0.01 0.015 0.02 0.025 0.03 0.035

Blowing Momentum Coefficient, C,

Fig. 5 Induced flow angularity.

1.2 I

0 1

0

0.005 0.01 0.015 0.02 0.025 0.03 0.035

Blowing Mom entum Coefficient, C,

Fig. 6 Calculated finite

AR

lift coefficients.

0 1

0

0 005 0.01 0.015 0.02 0.025 0.03 0.035

Blowing Momentum Coefficient, C,

Fig.

7

Measured lift coefficients.



MEASUREMENT AND ANALYSIS OF CC AIRFOILS 111

0 1

-

40

50

nit I I

2

$

0.05

Q

- 0

~1 0.05

Ll

i -0.1-

- . 18

WakeTurbulence Level, u '/Ue

Fig. 8

Wake turbulence profile C, =

0 x / c =

0.17).

0 10

20

3

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 plane-

mixing 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, o nce je t 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

1 t

, I

WakeTurbulence Level, u '/Ue

Fig.

9

Wake turbulence profile C,

=

0.009

x/c

=

0.17).



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. Th e 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 pre-

dictions. In turn, the analysis suggests that two-dimensional CFD computations

could well be meaningless unless the airfoil effective a ngle 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 5CL

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 ang les 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 exper iments 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 Sim ulation

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

-

, )

dy/q

c )

2 1/2

C

=

integrated force coefficient

(Ct +

CO,)

C

= lift coefficient

L / q

c )

C

=

pressure coefficient

( p -

, ) / q

CQ

=

steady volume flow coefficient (Q/SU, )

C = steady momentum coefficient

[(2

h/c) (Uslot /

U,)’]

CMac=mom ent coefficient about the aerodynamic center

(c,)

=

oscillatory mom entum coefficient [(h/C)(USlotMax/ u,)~I

d

=

reference length, diameter

= frequency of excitation

h = slot height

J

=jet momentum

q

=

dynamic pressure ( J p U k )

F+

=

nondimensional frequency (f d l

U,)

*Research A ssociate.

’Research Assistant. Currently at Tel-Aviv University, Ram at-Aviv, Israel.

*Research Assistant.

gPostdoctoral Fellow.

TResearch Assistant. Currently at L’Ecole Nationale Supdrieure de M dcanique et d’Adrotechni-

**Professor.


que, Poitiers, France.

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 =

ang le of attack or slot location on a circu lar cylind er

Sf

=

flap deflection

8

=

angular distance from the leading ed ge of a cylinder

I. Introduction

TH IN je t being em itted tangentially from a slot milled in a circular cylinder

A r other convex, highly curved surface, alters its direction and wraps itself

around the surface.

A

circular cylinder can turn a je t around and alter its direction

by more than 180deg. Th e centrifugal force ac ting on the deflected je t 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 on e side of a circu lar cylinder

and attached the flow on the side of the suction slot and generated lift. This idea

m

B

Fig. 1 Stream lines representing a wall jet flowing around a cylinder.



CIRCULATION AND SEPARATION CONTROL EXPERIMENTS

115

was applied by Schrenk to thick airfoils that were otherw ise plagued by ea rly sep-

aration.2 Steady suction has been characterized historically by a volume flow

coefficient CQ, because i ts 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 con-

formal 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. Th e 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^.^

Th e 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

NA 39 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 determ-

ined 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 attemp ted to calculate the lift by assum-

ing that the “jet-flap” was equivalent to a physical flap.5 More realistic assump-

tions were made by Helmbold,6 S p e n ~ e , ~egendre,* and Woods,’ who replaced

the jet by a vortex sheet originating at the trailing edge. Woods used the hodo-

graph method, whereas Spence7 and Malavard” linearized the problem, assum-

ing small incidence and small jet deflection.

In all the theoretical models, the mixing of the je t with the ambient flow is neg-

lected. In reality, the je t entrains fluid from i ts 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 pro-

portional 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 je t m omentum

was recovered as thrust as long as the value of C was quite large. At la rger flap

deflections, the C required to overcome separation and o ther “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 ' ~n con-

junction 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 Stratf~rd,'~ho 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 ~ ~ ' ~ecause 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 thickness-

to-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 o r 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

(

C J c r i t ?

3) Is separation effectively controlled by suction?

4)

Is

C, displaced by CQwhen 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 em ployed?

The present chapter focuses on som e of these questions, in an attempt to categor-

ize 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 < 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 mom entum

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 exci-

tation 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 generat-

ing

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 fre-

quency 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




119

1.6-

1.4-

1.2-

1

o-

0.8-

R e =

300K

-#-Baseline

AFC <C

>

= 3 , F+= 0.3

I

+AFC<CP>=3 ,F =1.1

- -

Blowing

C

=

3

P

0.6 I I I

20

40

60

Flap

deflection Sf(')

Fig. 4 Effect of blowing and AFC atC, =

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

1.4 -

1 2 -

1

o

-

0.8 -

Re = 300K

AFC <C > = 1.2 , F+= 1.1

FC <CP>= 1.2 , F = 2.5

P

+

--C Blowing

C = 1.2

. .

I I I

20 40 60

Flap

deflection

4

)

Fig. 5 Effect

of

blowing and AFC a t C, =

1.2 ,

as a function of flap deflection on a

NACA 0015 at a =

0

deg.



120

D. CERCHIE ET AL.

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 tech-

niques 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 imm ediately apparent in this figure. All of the con-

figurations share the same d C L /d a 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 ,

here 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 exper-

imen t, although the flow separates partially from the flap a x - deg. The other

two curves, plotted in Fig. 7, are not expected to have a discontinuity in dC L/d a.

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 exc ited flow at Sf

=

20 deg

is attached over the flap until a = astall.he excitation level for Sf= 20 deg

F+

= 0.9,

(c,)

=

2.2%) was specially selected in order to overlap the lift

0.5 1.0

1

0

-1

-2

-3C P

X/

C

Re = 300K α = 0°, δ f = 35°

Baseline

AFC <Cµ> = 3%, F

+ = 0.3

AFC <Cµ> = 3%, F

+ = 1.1

Blowing Cµ = 3%

0.0




121

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, s approximately the same for both cases.

When the same AFC is applied while

S = 40

deg, the maximum lift coeffi-

cient,

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,

ecause 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 exam ine 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 geo-

metric 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

S f

20 deg

for the baseline airfoil, the incidence must increase to a = + 2 deg in order to

-10 0 10 200

1

2

C L

α (°)

Re = 200K F+

= 0.9

δ f = 20° Baseline

δ f = 20° <C

µ> = 2.2%

δ f = 40° Baseline

δ f = 40° <C

µ> = 2.2%



122

D. CERCHIE ET AL.

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

=

-

deg, resulting in a more uniform

pressure distribution over the upper loft and reducing the suction peak near the

LE. Because

a

s the same for the two flap deflections as the total circulation,

the flow near the LE is identical over the upper surface, a s is the pressure distri-

bution 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

- .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 coef-

ficient at the trailing edge being positive ( C px 0.25), suggesting that the flow

downstream of the trailing edge continues with its downward mom entum, gener-

ating 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

the pressure distributions on both the upper and lower surfaces over the upstream

half of the airfoil are almost identical. The case w ith AFC show s the flow over the

f -

1

0

-1

-2C p

X/C

Re = 200K F+ = 0.9

δ f = 20° α = 2° Baseline

δ

f

= 20° α = -2° <C

µ

> = 2.2%

δ f = 40° α = -2° Baseline

δ f = 40° α = -8° <C µ> = 2.2%

0.0 0.5 1.0




123

CD

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 sam e

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 sam e dimension. It implies that the flow over the flap was com -

pletely separated at both angles of incidence and that the additional lift was gen-

erated 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

C p

-

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 elemen t

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 = , 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 circula tion, 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 i s 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

2

1

0

-1

-2

-3

-4C p

X/C

Re = 200K F+ = 0.9

δ

f

= 20° α = 8° Baseline

δ f = 20° α = 2° <C

µ> = 2.2%

δ f = 40° α = 2° Baseline

δ f = 40° α = -4° < C

µ> = 2.2%

0.5 1.00.0



124 D. CERCHIE ET AL.

deflection, the suc tion peak is further attenuated when sufficient lift is generated ,

even at zero A OA. Although the flow over the flap may not be fully attached, the

upstream effect of the A FC is strong enough to load the main elemen t 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 perform ance: first by p reventing flow separation on

a deflected flap (this mechanism w as investigated by Nishri and Wygnanski” and

Darabi and Wygnanski2’); secondly, by enhancing circulation through the invis-

cid je t 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.

Th e demarcation between (BLC) and (C C) was quite well defined when steady

blowing or suction w ere used to control the flow, because C C implied “an artifi-

cial increase in circulation ove r 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 m easured and calculated w as 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 exper-

imental results agree quite well w ith X foil’s viscous prediction, with the excep-

tion of the base pressure observed over the separated flap. Th e experimental data

for the forced flow suggest that the flow ove r 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, sugges ting 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



126

D. CERCHIE ET AL.

CP

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

NAC A 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,

0.1, in order to use the upper limit of

C,

=

0.1 for

comparison with data acquired by Hynes.13 For C, = lo%, CLmaxs 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 CDp nd 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

-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

Cp

Baseline, slat 2

F = 120 Hz, Uj/Uinf = 2.8

Rec = 0.75e6

Flap = 40 deg.

alpha = 13 deg.

ADVINT/ATT 5% Model in NDF at IIT

Nagib & Kiedaisch; 2002

Slot Location for AFC




127

C, s a,Baseline &Steady Blowing

Fig. 12

CL - 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



128

D. CERCHIE ET AL.

C, vs. C,,,

Baseli ne, 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

CD 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 ~ ” ’ ~ ’ ’ ~




129

CL vs. C D ,Basel ine, 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 < C , < l 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 CDx

-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.



130

D. CERCHIE ET AL.

a)

a Pi k h i n g M o m e n t a C14 vs. Cp

9

8

7

6

5

a 4

3

2

0

1

Variation

o f

a & Pitching Moment fo r f ixed

C L = l

F ap=3Oo

..............

..............

-2 J . . . 4.5

0

b)

4

c [ ,.I

8

10

CDp & CrJ vs.

C,

@

C p l

for

various

C,,

hap=300

0.15 ...... ...... ................. ......

...... ......

. . . . . . . . . . ........

......

Dp ..;

............ ....

:.

.....I..

..

-

+- CD

_ _ . I

c

0 1

0

1 2

4 5

6 7

8 9 10

c [ I

Fig. 15

Dependence of a)

(Y

and CM,, and

b) CD

and

CDp

on

C,

at

CL= 1.

C. Controlling the GLAS I1 Airfoil

The Glauert Laminar Airfoil Section I1 (GLAS 11)has a maximum thickness-

to-chord ratio of

31.4

and was designed to operate with massive suction

through a slot located at 69 of the chord. The des igner intended to have favor-

able pressure gradients over most of the upper and lower lofts, and maintain




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,

0.6.

The L I D increased to approximately 30 at

C,Sweep, Steady Blo win g for different

Flap

Angles

Fig. 17 Dependence of CL on C, at

(Y = 3

deg.



132

D. CERCHIE ET AL.

CL vs. a or 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,

-

Compar i son be tween S teady Suc t i on

&

B l o w i n g

fi x ed C,-I ,C,=lO ,~,,,=3O0

Fig.

19

Representative pressure distributions on the truncated 0015 airfoil using

suction at

C , = 10 .




133

Re 3 x lo6. It appeared that the flow was intermittently reattaching to the ram p

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.

Th e dependence of

CL

on

a

s plotted in F ig.

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 3 grees 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

-

Basel ine

Fig. 20 Dependence of C on (Y and on

Re

for the baseline airfoil.



134

D. CERCHIE ET AL.

Baseline

C,

pro f i les at

q 4 . 5

or di f ferent Re

Fig.

21

Pressure distribution measured on the baseline airfoil at constantC 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 s minimum depends strongly on Re as it changes from x /c =

0.07

at the

lowest Re used presently to x /c = 0.33at 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 separ-

ated 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

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 litera-

ture. The Australian researchers faced similar difficulties and they, too, moved

the suction location. The chosen location of the slot corresponds to the separa tion

line predicted by C FD and it will be discussed separately. Con trary to the obser-

vations of Glauert et a1.,22 blowing was more effective than suction, approxi-

mately 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

14 deg. At higher inci-

dence, the difference in

C,

between the steady blowing and the periodic exci-

tation became noticeable, yielding

CLm,, =

3.5 for the periodic excitation.

Because the

C

obtained for the basic a irfoil was only 1.7, suction and peri-

odic excitation represent 100% increase in

CLm,,,

whereas the much stronger

steady blowing represents almost 200% increase in this coefficient.




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, 0.11. 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 fric-

tion drag is approximately 0.024 for the negative lift coefficients.

The application of suction at a

<

6 deg generated a small dC L/da and it pos-

sessed relatively large C,. For C,

>

0.4, suction reduced the drag to C, 0.02.

This represents a substantial drag reduction relative to the baseline airfoil that

attains C, 0.4 around a 20 deg. The form drag in this case is negative, indi-

cating 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 . Accord-

ing 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-Quinton4 his drag

is equal to CDsink 2cQ= 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,,,,, = - cQ 0.034.

The maximum thrust recovered may be CT= C, +2cQ 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 exci-

tation 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 respectiveC

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,

gen-

erated by steady blowing dropped from CLmax to 3.1, whereas suction

attained a

CLmaX

f 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

CDp

associated 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 con-

trast to the results accumulated on the NACA 0015 and its truncated derivative,

CDp s usually smaller than C, except near the stall angle; it is possible that the

concave curvature affects the result as well.

The pressure distributions over 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 mechan-

isms 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




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, ), 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

i t

also maintains a good pressure recovery downstream,

culminating w ith 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 down-

stream. The bubble is generated by the oblique injection angle of the jet,



P

n

n

Fig. 24 Pressure distributions measured at CL= 1 and 2.1

for

two values

of Re

and various control mechanisms.

Next Page

http://77898_05b.pdf/

http://77898_05b.pdf/




139

because the slot is inclined at 30 deg to the downstream surface. After reattach-

ment 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 L E edge radius of curvature of the GLAS I1 airfoil is very small, pointing to

a potential problem with this design. There i s 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 com bination 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 corre-

spond 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

Previous Page

http://77898_05a.pdf/

http://77898_05a.pdf/



140

D.

CERCHIE ET

AL.

Effect of F’ and C,

on

C, at

GO*,

U i , p l 4 m l ~

Effect of F’ and C on C, at

a= ,

i,,=14mls

Fig. 25 Increase of

C,

with

C,

for a variety of

F’

at two values of a.

which CL ncreases (Fig. 26). A further increase in

C

results in a “spike” in the

observed C p us t upstream of the slot and in a higher pressure recovery over the

ram p (Fig. 27).

These results seem to contradict the previous concept that required a threshold

value of CPcrit o 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 d rag at the lowest amplitudes imposed. The response

of CDp o 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 CDp esulted 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 l m l s

Fig. 27 Effects of C, on the pressure distribution for two values of F + and a.

study of Nishri and Wygnansk ilg 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 co lleagu es 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




143

a=6 ,

P a n d C,

effect

on

CD,

CD,,

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,.



144

C L

D.

CERCHIE ET

AL.

a=6',

C , sweep at U,,,=l4m/s

Fig.

30

Effects of suction, blowing, and AFC on lift at small values of m omentum

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.




145

0.41

/>

-@-

Suction C@ 1 go/,)

+AFC f = 130Hz,<C,> = 1.9p)

w . w - I

I

I

100 110 120 1 o 140

TE

slot exhaust angle

( )

Fig. 31

ellipse.

Comparison among three different types of

flow

control on a 30 thick

The pressure distributions at C

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

Fig.

32

Comparison among different flow control techniques at constant C, 30

thick ellipse.

0.0 0.2 0.4 0.6 0.8 1.0

1.0

0.5

0.0

-0.5

-1.0

-1.5

-2.0

120° Slot Location

110° Slot Location

C P

x / c

AFC C L = 0.448 <C µ> = 1.9% TE = 110°/0.030"

Blowing C L = 0.436 C µ = 1.9% TE = 110°/0.030"

Suction C L = 0.441 C µ = 1.9% TE = 120°/0.030"



146

D.

CERCHIE ET

AL.

control and steady suction produced the same increase in C p at the LE w hen 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 low-

pressure 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 w ere 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

C p

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 C p 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).




147

130

Slot Location f f 0

+ - - - - -

\

cp

1

.o

-0.6

I \

-0

-0

0

o

02

AFC C;= .619 -dP>2.38%

-m- Baseline

FL3D 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 s 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 10deg, but a (i.e., a corre-

sponding 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 per-

formance 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

suction coefficient produced larger

C

However, the

a

at which

CLm,,

was

realized was not directly proportional to the change in the suction slot location



148 D.

CERCHIE ET

AL.

CL

1.2

0.8

0.4

0.0

1.6

F* =

0.35, E = 130 /0.030 , 1Oms-l

+Baseline

Cp>=l 2%

C#>=l.8%

0

5

10 15 20

( )

Fig. 35 Angle

of

attack sweep using AFC.

or its width. Fig ure

38

show s the effect of steady suction o n the

C

at

a =

0

deg

where the basic ellipse provides n o lift. T he low er surface velocity is decreased

only slightly by the suction and almost at the same increment along the entire

lower surface. T he strongest effect occurs on the upper surface whe re the increase

in velocity is constant from the L E to 90 of the chord. The change in the TE

CL -W-

Baseline

20

-

10

-

00

.2

=

0 ,

=

0.030 ,

p

1.9%

0.8

0.4

0.0

.d/

0 5

10 15

Angle

of

Attack

( )

Fig. 36 Suction on

30

ellipse:

C,

=

1.9

and h =

0.030

in., U = 10 d s .




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 = 4 deg and =

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 .



150

ACL

D.

CERCHIE ET

AL.

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 co llapse onto a sing le curve, indicating that som e other parameter or par-

ameters 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 depen-

dence 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 separ-

ation location or imposed Kutta condition. Unlike the previous four geometries

discussed, the LE and the T E flows are closely coupled because of their immedi-

ate proximity. The flow is also sensitive to transition location, which is, in turn,

sensitive to a variety of inflow parameters.




151

Fig.

40

Collapse of data based on em pirically 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 some-

times 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 je t mom entum. This force multiplier makes som e 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

je t flow. The pressure distribution is not seen to be very sensitive to the jet-related

Reynolds number.



152 D.

CERCHIE ET

AL.

Fig. 41 Pressure distributions on a cylinder for various Reynolds num bers.

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 con tinuous 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 seg-

mented 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 func-

tion 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 sam e 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 C p generated is -2 2 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




153

Comparisono f Pressure

Distribution

o f C o d a Flow

ReM=llOOO

Fig.

42

Pressure distributions on a cylinder for various slot geom etries.

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 dist r ibut ion for d i f ferent s lot locat ions

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 distri-

bution 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 CFwas measured and plotted in Fig. 45a as a

function of slot angle relative to the L E 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 cy linder 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 je t mom entum

J

Again, the trend is a proportional

relationship between slot velocity and the freestream, but neither

q

nor

J

normal-

ize 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




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 se lf-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




157

the following form for discussion purposes:

In fact, when a je t of kinematic mom entum

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 is calibrated in the absence of an

external stream, which is normally the case. The last equation is applied to

some early da ta 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 jus t C Similar reprocessing was carried ou t to the w in

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 par-

ameter, 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 geome-

tries and control strategies to a unique line without the introduction of terms that

account fo r 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




159

CL

vs.

Cp or (0.5*CpCQ)

(from

John

Attinello,

1961)

Fig. 49 Blowing over a wing section using data from R ef. 2 (Attinello).

location, under steady blowing. This phenomenon is similar to an airfoil when the

angle of incidence is altered around

astall.

herefore, the initial flow condition is

also important to the performance of BLC and CC, as was discussed by Nishri

and W y g n a n ~ k i . ' ~revious tests on a Wortmann airfoil27 have shown that this

hysteresis can be reduced or eliminated by using AFC. The pressure distribution

around the cylinder w ith a slot located at 120 deg from the L E indicates the flow

can be either attached, yielding a maximum lift, or completely detached, provid-

ing 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 m ade 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 rela-

tive to the single slot by improving the downstream flow conditions. It is interest-

ing 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 Com parison,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 w ithout introducing lift (Fig. 52) .The inviscid

value of the pressure coefficient of is realized at 50 of the chord position.

The cylinder’s drag drops from approximately

C D >

1.0 for the natural case (not

blown) case to a CD

x

0.4 for the symmetrically blown case. Proper accoun ting

should be m ade in order not to waste momentum, because too much flow control

may be applied for given conditions. Figure 53 shows that for symm etric 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,




161

Comparison of Pressure Distribu tion slot located at u=9O0

Cylinder # 3, Re=39000, Cp=1.31

Fig.

51

Pressure d istributions 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 slo tted Cylinder 1 o o , 0 sweep bac k,

Re135000, Steady Suction, C, per slot

Fig. 52 Pressu re distributions on a du al-slotted cylinder for different suction C,.



162

D.

CERCHIE ET

AL.

C

0.00

0.46 1.02

C D ~

.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 dom inant S trouhal number associ-

ated with vortex shedding is constant, requiring a frequ ency increase resulting

from a de crease in the mom entum thickness.

111.

Conclusions

Flow control tests ove r five two-dim ensional, aerodynamic, and bluff bod ies

provided s om e new insight into the parameters gov erning active control of sep-

aration 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.

Th is chapter questions some of the acce pted concepts associated with separ-

ation and circulation control, although it is not able to provide definitive

answers to these questions. For exam ple, the enhan cemen t of circulation can be

achieved without a ttach ing the flow; how ever, by forcing separa ted flow to reat-

tach to the surface, circulation is generally enhanced. Flow reattachmen t and the

control of circulation d o not have to occur sequentially, as it is widely believed,

leading to so me critical value of an input param eter that separates the two.




163

Th e use of steady blow ing is ineffective a t 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 blow ing might beco me the technique of choice whenever the integration

of propulsion and ae rody nam ics are considered. Suction and periodic excitation

are much more effective in reattaching the flow at low C , values and enhanc ing

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 m ake it the most attractive technique for lift enhancem ent.

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 low-

intensity, 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 C D p

s being the m ajor contri-

butor to drag is wrong. In most cases, as C , increases,

so

does

C D p

while 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

Th e con-

trolled 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 o r by cha nging the thickness ratio of the ellipse. Th e TE perform-

ance 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.

Th e curvature of the surface downstream of a slot is an important variable, a s

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 em ana ting 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 separ-

ation. 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 Appli-

cation,

Pergamon Press, New Y ork, 1961.

3Goldschm ied, 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’Emission d’un Jet au bord de Fuite d’un Prof1 sur 1’Ecou-

lement 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 A erofoils,” 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 Two-

Dimensional 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 Mech-

anics, 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 o f Flow Separation by Periodic

Excitation,” Progress in Aerospace Sciences, Vol. 36, 2001, pp. 487-545.




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 M echanics, Vol. 510,

”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.,

nd 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 Streamw ise Structures in a

Wall Jet Flowing over a Circular Cylinder,”

Physics of Fluids,

Vol. 16, No. 6, 2004,

26Englar, R. J., “Test Techniques for High Lift Two-Dimensional Airfoils with Bound-

ary 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,” M ay 1972, pp. 1-50.

”Neuburger, D., and Wygnanski, I., “The Use of a Vibrating Ribbon to Delay Separ-

ation on Two Dimensional Airfoils,”

Proceedings

of

Air Force Academy Workshop on

Separated Flow,

F.J. Seiler Research Labs. Rept. TR-88-0004, 1987.

2004, pp. 131 144.

pp. 2158-2169.



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 d istance from je t exit to m easurement location

r radius of CCW surface

T

temperature

V velocity

a angle of attack

Af

frequency bandwidth for narrowband acoustic spectra

polar angle (with respect to the flow axis)

p

density

Re Reynolds number

Subscripts

s associated with slot

T associated with tunnel freestream

*Graduate Student, School of Aerospace Engineering; currently at Naval Air Warfare Center,

'Regents Researcher and Professor, Georgia Tech Research Institute, and School of Aerospace

*Principal Research Engineer, Georgia Tech Research Institute, ATAS Lab. Associate Fellow


Wea pons Division, China Lake, California. Student Member A I M .

Engineering. Fellow, AIAA.

AIAA.


167



168

S. E. MUNRO, K. K. AHUJA, AND

R.

J. ENGLAR

associated with je t

o ambient condition

I. Introduction

NE of the major environmental dilemmas facing today's aircraft industry is

oise 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 conven-

tional flaps.

Circulation control wings (CCW ) have been researched and developed exten-

sively, primarily for the purpose of increasing performance and reducing o r repla-

cing the conventional flap system of an aircraft.' Ov er the years, C CW 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 replace-

ment for conventional flap systems in the aerodynamic realm,' they are immedi-

ately 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



NOISE REDUCTION THROUGH CC

69

impact of several parameters must be investigated, such as the blowing slot

height, slot velocity, and CCW geometric configuration (i.e., flap type and deflec-

tion 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 ~ t . ~t 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 Schem atic circulation control wing concept.



170


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 availab le reference

length. In general, a given C, will provide a g iven 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 comprom ise 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 associ-

ated 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 d rag 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 con-

ventional 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 advan tage over a conventional system.

There are many potential uses for circulation control. How ever, the two appli-

cations that have received the m ost research attention have been C C rotors (C CR )

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 B oeing 737

type of aircraft. A summary of the effort is docum ented in Ref.

6.

The only known

acoustic work on CC W configurations was performed by Salikuddin e t a1.2 There

are other otential uses for CC , including autom otive and in heli-

copter~,’,’~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. Facil it ies and Instrumentation

Th e anechoic flight simulation facility (AFSF) was used in the experiments. It

is located at Georgia Tech Research Institute (G TR I) located at its Cobb County

Research Facility in Smyma, Georgia. The AFSF operates in an open-jet wind-

tunnel configuration. It is an anechoic facility that allows acoustic measurements

to be made in the presence of a freestream (Fig. 2). Th e tunnel inlet has a square

inlet that converges down to a 28-in. round duct. The duct terminates in an anec-

hoic room as an open jet. Protruding out from the downstream wall is the collec-

tor, which is 4 ft wide by 5 ft high. Th e 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. T his locates the wing across the je t opening immediately d own-

stream 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 indi-

vidual 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 micro-

phone from angles of 3 0 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


R.

J. ENGLAR

Fig.

3

Photo

of

a conventiona l wing mounted in AFSF.

microphone w as 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 CC W 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 aerodynam ic studies, the

best overall aerodynamic characteristics were obtained with the sm all CC W flap

configurations. The sm all 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 w ith the flap deflected 90 deg. This was used a s the starting configur-

ation 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 configur-

ation. Both flaps spanned the entire wing, but one flap had a cutout in at the




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 CC W 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.

Th e 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

com pared directly. Lift da ta from previous studies were used for this comparison.

V.

Results and Discussion

A. Acoustic Optimization of Existing CCW State-of-the-Art

Configurations

The C CW concept has been around fo r 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 configur-

ations would be unreasonable, because many of the changes were made

to improve the system. There is little reason to test acoustically a system that




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 aerodynam ic 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 perform-

ance of several configurations tested in previous stud ies. The flap was eventually

adjusted to 30-deg deflection to prevent flap-edge vortex shedd ing 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 w ere 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 con-

figuration was com pared with a conventional flap configuration. The conven tional

model had the same generic airfoil shape as the CCW, except near the trailing

edge to accom modate 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 geom etric angles of attack of 0,7, and 14 deg.

The m ajority of the data presented in this section was acquired at a geom etric

angle of attack of 0 deg and at the highest freestream velocity of about 240 ft/ s

unless otherwise noted. Figure

7

show s acoustic spectra for several slot velocities

with no freestream flow for the CCW with the 90-deg flap configura tion. It show s

a similar trend to the basic jet velocity scaling property develo ed for round jets.

For the measured velocities, V 8 scaling of jet noise theory' predicts about a

19 dB increase between the two m ost extreme cases, which is similar to that

Frequency,

kHz

(Af

=

32 Hz

Fig. 7 CCW blowing system noise spectra with no freestream flow; VT=

0

ft/s,

h

=

0.006

in.




177

experience/theory on round jets16 this will provide for the jet mixing noise inten-

sity proportional to slot exit area. This translates into a 3 dB increase in noise

after shifting the spectrum for 0.006 in. to the left ove r the spectrum for

h

0.012

in. by a factor of one octave to allow for the shift in the noise frequen-

cies 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 contami-

nated by any internal noise.

W e 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

V8

et no ise 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<

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 trail-

ing edge of the deflected flap Stshedding 0.2 would produce a shedding fre-

quency 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 compu-

tational study also produced the sam e 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.



78

S. E. MUNRO, K. K. AHUJA, AND R. J. ENGLAR

2

'fo

X

II

p

n

n

v

Frequency,

kHz (Af

=

32

Hz)

2

'p

z

X

II

b)

Frequency,kHz

(Af = 32 Hz)

Fig. 9 CCW with 90-deg flap and freestream velocity,

=

90 deg, V ,

=

220 ft/s,

a ) f =

0 60 kHz;b ) f =

0-5

kHz.




179

Frequency,

kHz

(Af

=

32

Hz

Fig. 10

CCW

with 30-deg flap and freestream velocity,

=

90 deg,

VT= 220

ft/s,

f = -5

kHz.

Data for test conditions similar to those for the 90 deg deflection are shown in

Fig.

11.

Again, w ith no blowing the tone is present. However, with small amounts

of blowing, the separation is eliminated, and hence the tone is completely elimi-

nated. Because this configuration showed more promise, the remaining par-

ameters were optimized using the 30-deg flap configuration. Both slot height

and slot je t 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 depen-

dent 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 som e variation of lift with

h

for high

C,,

but in the

C,

range of interest here, 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



180


R.

J. ENGLAR

2

X

Frequency, kHz

Af=

32 Hz

Fig.

11

CCW with 30-deg flap and freestream velocity,

=

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 , p

=

0.04, = 90 deg,

VT

=

220 f t / s , f =

0-60

kHz.




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 / R T , but it varies pro-

portionally 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 a t 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 p

=

constant.



82


R.

J. ENGLAR

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. 1 3 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 con-

figurations 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 f rom the surface). Thus, once the lift for

the unblown CC W was found, this cou ld be com pare d to the c1 for the conven-

tional airfoil and the needed

Acl

could be calculated by subtracting the

two values. This Acl was then used to de term ine the

C

needed to match lift pro-

vided 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 m inimum CC W noise condition, they are pre-

sented in the figure. In the range between 1 and 10kHz, 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. Th is difference

contributes a fair share of noise to a conventional wing system, because flap edg e

noise has been identified as a major contributor to airframe noise. Thu s, this wing




183

h

n

v

Frequency, kHz

A f

= 32 Hz)

Fig.

14

CCW

and conventional wing

2-D

flap at similar lift condition,

=

90

deg,

V T

= 220 ft/s.

was missing a noise source that would most likely be greatly reduced in a CC W

system. Be cause the C CW flap is much smaller, there is no need for a gap in the

flap to avoid en gine exhaust. Its small size would a lso in many case s redu ce the

need for gaps due to structural concerns. Th us the CCW system with a full span

flap is not unreasonable.

Acoustic tests were performed on the new configuration, similar to the pre-

vious tests. Figure

15

shows the comparison of the wing with the cutout flap

with the CC W . As expected, the cutout in the flap increased the noise on the con-

ventional system significantly and shows a significant advantage to using a CC W

system in the region below 10kHz 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 d ata from a micropho ne at 90deg 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 line-

of-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


R.

J. ENGLAR

Frequency,

kHz

(Af = 32

Hz

Fig.

15 CCW

and conventional wing with cutout at similar lift condition, =

90

deg,

VT

= 220 ft/s.

wing at about 56 deg. Even with freestream velocity, the je t 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 30 deg and

60

deg. At 30 deg the CCW system produces no real advantage over a conven-

tional 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.

T o resolve som e 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 experim ents

presented here. In addition, these data will be used to augment the present results

by demonstrating the even greater benefits possible for a CC W high-lift configur-

ation 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



a


Frequency, kHz

(Af = 32

Hz

185

Fig. 16

CCW

and conventional wing with cutout at similar lift condition,

VT

=

220

ft/s: a) =

30

deg,

b) = 60

deg.



186


R.

J. ENGLAR

acoustic advantages. Previous studies have show n that the CC wing is an aerody-

namically viable alternative for conventional mechanical flaps. This study show s

that there is also a substantial advantage in the acoustic realm. The results pre-

sented showed a lower noise spectrum for a CC W system compared to a conven-

tional 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 advan-

tage. This is because a CC W 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 C CW 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 2000-

2541, 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 Sim ulations 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 C ircular

Cylinder,”

Journal

of

Fluid M echanics,

Vol. 33, No. 3, 1968, pp. 495-514.

5Englar, R. J., and Applegate, C. A., “Circulation control-A Bibliography of

DTNSR DC 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 Cir-

culation 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 83-

1847, July 1983.

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/ASED-

81/08, March 1981.

8



NOISE REDUCTION THROUGH CC 187

’Nichols, J. H., Jr., Englar, R. J., Harris, M. J., and Huson, G. G., “Experimental Devel-

opment of an Advanced Circulation Control Wing System

for

Navy STOL Aircraft,”


“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 Trans-

mission, 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,

15Ahuja,

K. K.,

Teste r, 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,

”Tarn, C.

K.

W., and Zaman,

K.

B. M. Q., “Subsonic Jet Noise from N on-Axisymmetric

18Tam, C.

K.

W., and Auriault, L., “Jet Mixing Noise from Fine Scale Turbulence,”

‘’Tarn, C.

K.

W., Golebiowski, M., and Seiner, J. M., “On the Two Components

of

”Tam, C.

K.

W., “Influence of Nozzle Geometry on the Noise of High Speed Jets,”

’lAhuja,

K. K.,

“Correlation and Prediction of Jet Noise,” Journal of Sound and

pp. 51-85.

pp. 317-341.

and Tabbed Nozzles,” AIAA Paper 99-0077, 1999.

AIAA Paper 98-2354, 1998.

Turbulent Mixing Noise from Supersonic Jets,” AIAA Paper 96-1716, 1996.

AIAA Paper 98-2255, 1998.

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.

d=

section profile-drag coefficient

Cl

= section lift coefficient

C,

cos

a C

sin

a)

C = moment coefficient

C = normal force coefficient

C = fluidic pow er 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.

L

= lift, lbf

M = mach number

riz = mass flow, lbm /s

Z,J,K = pressure tare coefficients for balance

*Research Scientist, Flow Physics and Control Branch. Senior Member AIM

Copyright

005

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 a re reserved by the copyright owner.

191



192 GREGORY S. ONES

NPR = nozzle pressure ratio =

P,/P,)

f= fluid power, ft.lb/s

P = pressure, lbf/in.2 or lb f/ft2

p’

=

fluctuating pressure, lb f/in.2 or lbf/ft2

q

= dynamic pressure, lbf/ft2

=

pU 2

r

= trailing edge radius, in.

s

= airfoil reference area, ft2

T = static temperatu re, 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 m )

Ojet = Coanda jet separation angle, deg

E = blockage interference ratio, u / U

p

= density, lbm/ft3

r= circulation

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

= freestream conditions

I. Introduction

IRCULATION control (CC) technologies have been around since the early

C

930s, 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 em phasis on p roviding better vehicle performance and predic-

tion 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 Con-

ventional high-lift systems that use flaps and leading edge (LE) slats can be

associated with significant weight and volume penalties of a typical wing assem-

bly. These assemblies are also complex (up to 3 and

4

subelements) and very



PNEUMATIC FLAP PERFORMANCE OF CC AIRFOIL

193

sensitive to location relative to the main element of the wing. The need to sim-

plify 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 m odem aircraft have been limited o r non-

existent, and so the ability to buy its way onto an aircraft is generally unproven.

Nevertheless, several blocks to real aircraft applications reappear in every discus-

sion 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 mom ents associated with aircraft

trim. Although this is not a comprehensive list, these issues will be used as a

guide in developing a C C wing for general aviation applications.

A primary objective of this effort is to evalua te the benefits of pulsed C C 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. Second-

ary objectives of this study were to evaluate the dual blown pneum atic 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 eng ine, 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 ES TOL 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.



194 GREGORY S. ONES

airport infrastructure. However, as these airport requirements become more strin-

gent, 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

a

C

= 4.0,

yielding an

L/D,,,

of

20.

In the near term, reduced approach

speeds enable a 1000 t field length a nd ca n improve safety in addition to redu-

cing community noise signatures. If equivalent control margins and gust sensi-

tivity 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 C C 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). Th is is a potential sourc e for air augmenta-

tion to a CC system. Because engine out conditions are an issue for CC appli-

cations, 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 require-

ment. 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)

anding 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 capa-

bility 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 FlowControl

(High

Lf l )

Fig.

2

Notional concept of NASA ESTOL 100-passenger vehicle showing potential

CC

vehicle.




195

of either of these vehicle concepts. The focus of this chap ter 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 ~olution.~ractically, 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 necess-

ary to extend the effective TE beyond the physical TE location with a virtual or

pneumatic flap, as simulated in Fig. 3.

To understand the lim its of airfoil performance, i t in necessary to be aw are 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 4

+

:)

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 rc round

the body

lo:

L

=

pure

( 2 )

Fig.

3

CFD simulation

of

pneumatic flap and streamline tuning using a Coanda jet.



196 GREGORY S. ONES

The magnitude of the circulation r, s a function of geometry alone and will be

referred to as induced lift and can be related to the modified pressure on the inte-

grated boundary of the body:

257

L

= S p r s i n 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 stream-

lines are significantly influenced by the m agnitude of the circulation r,. n prac-

tice, the inviscid limit is never reached because of flow separation. However, for

an airfoil employing a BLC or a CC device, the m aximum 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) :

L

= ~ u ( r cq e t )

(4)

where

r,,,

=

EE a+

8jet)

PUW

and can be related to lift and drag as

CYLINDER MAPPED

INTOAIRFOlL

LE STAGNATION=

TE STAGNATION

LE

STAQNATDN

TE

STAGNATION

Fig. 4 Classic lift resulting from circulation for a circular cylinder and mapped into

airfoil profile.




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. F or pneumatic systems this

reactionary force should not be confused with the thrust vectoring that an articu-

lating nozzle generates on an eng ine nacelle. The reactionary force that is charac-

teristic 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), com-

pared to typical C C 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. Nomin-

ally, je t flap airfoils depend largely on the reaction force of the je t m omentum.

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 cap-

tured 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 press-

ures 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

angle. The integrated wind-tunnel wall signature and wake

rake must also account fo r 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 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 C C 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 C C subsystem reduces the overall system penalty

associated with CC. The recovery of this thrust will be dependent on the effi-

ciency 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. ONES

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 tw o-dimensional study

(where nozzle AR is meaningless), it is appropriate to neglect the nozzle effi-

ciency 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.



where

and


199

m

=

p et U.,

thw

9)

Th e 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

system s can be formulated by estimating the power requirements of CC.

For a crude estimate of fluid pow er

(Pf),

t is assumed that the je t is taken fro m

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. his ideal power can

be described as1*

where

1 m

Pjet 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”uc a

[

+

2( )2] (qcoucos )

and nondimensionally

(14)

If the je t 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

S.

ONES

00

ratio

h l c :

Figure

8

shows the nondimensional ideal power for a typical CC jet orifice.

A.

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 de tail by Betz13 and Jones.14 The profile drag can be

determined by integrating the wake measured one to three chords

downstream of the TE

Two-Dimensional Drag with Blown Systems

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 -

1.0

-

0.8

w

0.6

0.4

-

0.2

-

Power

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.




201

exceed that indicated by conventional wake rake calculations by the quantity

riz U,. Considering a frictionless hypothetical case w here 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 C C

airfoils, je t flaps, blown flaps, engine augmented pow ered 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 avo ids 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

=

Dprofi le -k Dpower + D r a m + Dinduced

where

Dpr o f i l e

is the profile drag,

Dpower

s fluid power, Dram s 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

mq?

m

Dequiv drag

+

pU

2uc.a P

and

19)

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.



202 GREGORY S. ONES

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 geo-

metry. 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 C oanda surface shape (e.g., circu-

lar, elliptical, and biconvex).

4) To provide pulsed pneumatic control to minimize the mass flow require-

ments 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 aero-

dynamic community (not restricte2 due to proprietary issues) and that has the

2bO

200

I 1 5 0

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.



204

GREGORY

S.

JONES

GACC AIRFOIL

PROFILE

BLUNT

LEA DING EDGE

RADIUS-1.93%

Fig.

10

Seventeen-percent thick

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 TE s 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 GAC C 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.25926 ominally, 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 diam eter. 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/C CW airfoil2’ was a 6% thick supercritical wing section that incorpor-

ated 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 0 001 0.002

0 003 0.004 0.005

h / C

Fig. 11 Effective Coanda performance for d ifferent radius and jet slot heights.



PNEUMATIC FLAP PERFORMANCE OF CC AIRFOIL 2 5

system2* only. Any operational use of this design would require a mechanical

retraction of the C C system in to the w ing to avoid a large cruise drag penalty.

To minimize the GA CC airfoil drag performance without the use of a mech-

anical system a dual-blowing pneumatic concept with a small radius TE was

designed. A baseline circular

r / c

of 2 was chosen for the G ACC . Three differ-

ent TE shapes were designed to be interchangeable and integrate with the G ACC

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 (0.174n.). The 2:l elliptic shape reduced the Y/C to

1 and the biconvex shape had an

Y/C

of 0.

To com pare 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 inde-

pendently 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 ac tuator 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, a s shown in Fig. 14.Twenty actuators were dis-

tributed in the upper plenum along the span to optim ize the pulsed authority to the

VARIABLE

UPPER SLOT

2 : l

BI CONVEX

Fig.

12

Sketch of three interchangeable

TE

shapes for the GACC airfoil.



206

UPPER STEADY

MANIFOLD

GREGORY S. ONES

ACTUATOR UPPER

DIFFUSER SLOT

LOWER STEADY

MANIFOLD

Fig. 13 Sketch of internal

flow

path of the GACC airfoil.

upper Coanda je t for the high-lift mode. A ir for all three sources was fed from

one end of the model and was expanded into large plenums then channeled to

the trailing edge je t exit. Both the upper and low er slots were adjustable

(0.005 < h < 0.025) and were fed from a smooth contraction that had a

minimu m 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 rela-

tive 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 wind-

tunnel interference for CC systems, several experimental design considerations

were considered: Solid Blockage (physical chord and span related to wind-

tunnel cross-section), Wake Blockage (how much streamline turning will be

20ACTUATORS

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.




207

achieved with blown system), and

Juncture

flow regions (aspect ratio of

model).

The GACC model was sized and built for the NA SA Largley Research Center

(LaRC) Basic Aerodynamic Research Tunnel (BART) and had a chord-to-test-

section 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 2 8 in. These values are conservative

for the unblown c ~ n f i g ur a t io n ,~ ~owever, 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 ai r results

of this preliminary C FD evaluation indicated streamline turning and wake deflec-

tion would not impact the tunnel walls for the BA RT test conditions but would be

influenced by the presence of the solid tunnel walls. The study of wall interfer-

ence is ongoing for this experiment.

V. Experimental Setup

Experimental results have been obtained for a GAC C airfoil in the open return

Langley BAR T, 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 measure-

ments (steady and unsteady) w ere used to highlight boundary layer transition and

separation.

A block diagram of the BA RT data acquisition is shown in Fig. 16. T o capture

the transients and time-dependent characteristics of the pulsed flowfield two

approaches were developed: arrayed thin films and miniature pressure trans-

ducers. 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 trans-

ducers. 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.



208 GREGORY S. ONES

UNSTEADY

P+p’

Fig. 16 Block diagram of BAR T data acq uisition for G ACC 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

psid, and

2.5 psid.

The five-component strain gauge balance was also custom designed and fab-

ricated for the GACC model. Normal, axial, pitching moment (ref. 50% chord),

rolling moment, and yawing moment lim its 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 GAC C balance limits

Normal, Axial, Pitching Rolling Yaw ing

lbf lbf mom ent, in. lbf mom ent, in. lbf mom ent, in. lbf

100

1 1600

400

40




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 com pensate 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. U npressurized 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.30931ach

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 de livery system having three independent air supply lines.



210 GREGORY S. ONES

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. How ever, upon careful inspection of the issues related to junc ture 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 fab-

ricated for the BAR T. The streamw ise 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.




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 interfer-

ence 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 d

=

f0.0005. For the momentum

sweep at AOA = 0, the wake moved approximately one chord below the center-

line. 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.



21 2 GREGORY S. ONES

0.0125

0.0075

0.0025

CP

-0.0025

-0.0075

NO BL OWING

-0.0125 I

I

-2 -1 0 1 2

U C

(WAKE POSITION)

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

CP

AOA

- - --10.0

- -

-6.0

0.0

-0.10

-0.15

0.20

*C,o,,

2 -1 0 1 2

U C

(WAKE POSITION)

Fig. 21

C

= 0.075, x/c = 4.64.

Wake profile of GACC with circular trailing edge, -10

c

AOA c 10,




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 approxi-

mately 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 microtele-

scope 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 vari-

ations 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 je t vel-

ocities 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 separ-

ation 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 vari-

ations an d/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 M ach numbers

biconvex TE configuration,h = 0 020 in.).



PNEUMATIC FLAP PERFORMANCE OF CC AIRFOIL 215

2.000

1.500

0.000

0 0.2 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.



216 GREGORY S. ONES

VI. Airfoil Performance

Airfoil performance will be discussed for two modes of the GACC airfoil: the

high lift mode w ith upper slot blowing and the cruise mode with upper and lower

slot (dual) blowing. The efficiency of pulsed blow ing will be discussed as part of

the high-lift mode.

A. High Lift Mode

1. Baseline (No Blowing)

Lift, drag, and pitching m oment will be used to establish the two-dimensional

baseline performance of the GACC airfoil with different TEs. The original

GA CC airfoil was designed around the circular TE having an

r / c

of 2%. There-

fore, 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 F ig. 27.

Com parisons of the drag performance for the three TE are show n 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 = . The airfoil efficiency shown in Fig. 29 indicates that the circular TE

is more efficient than the elliptic or biconvex TEs with no blow ing. The peak effi-

ciency 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 20

AOA

Fig. 26 Baseline lift coefficient with no blowing balance data).

Next Page

http://77898_07b.pdf/

http://77898_07b.pdf/




21

7

-5.0

-4.0

3 . 0

-2.0

-1

.o

CP

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 b lowing, 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).

Previous Page

http://77898_07a.pdf/

http://77898_07a.pdf/




219

4.0

3.0

2.0

CL

1 o

0.0

-1

.o

-20 -1

0 0

10 20

AOA

Fig. 31

represent lower b lowing).

Airfoil lift performance with circular

TE

and h l c

=

0.0022 (open symbols

of upper and lower blowing with variations in AOA enables the designer to cus-

tomize 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 com pensates for the T E 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

2.5

2.0

1.5

l

1 o

0.5

0.0

0.00

0.02 0.04

0.06 0.08

0.10

CP

Fig. 32 Lift performance

of

circular

TE,

AOA =

0.

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

0.00 0.02 0.04 0.06 0.08 0.10

C

Cl

0.0014

0.0022

h/C

∆CL

∆C

= 60.3

∆CL

∆C

= 45.3

BOUNDARY LAYER

CONTROL

SUPERCIRCULATION

CONTROL



220

GREGORY

S.

ONES

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 effi-

ciency 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 super-

circulation 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 GA CC. 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.

T o gain a greater understanding of drag characteristics for this airfoil, the total

drag measured in the wake can be decomposed into a two-dimensional circula-

tion induced force represented by the pressure distribution on the airfoil

(shown in Fig. 34) and the reactionary force created by the Coanda je t 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 two-

dimensional 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, h l c =

0.0022,

wake rake data (open

symbols represent lower blowing).




WC

22

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 approxi-

mately 0.03.

Th e 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



222

GREGORY

S.

ONES

0.00 0.02 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. Com parison of the two slot con-

figurations 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 L I D 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.




223

U D

EQUlV

0.00

0.02

0.04

0.06

0.08

0.10

CP

Fig. 37 Efficiency of circular TE, A O A =

0.

pow er to the equivalent drag. Any values that deviate abov e or below this line can

be related to the two-dimensional c irculation induced effects described above and

highlight the magnitude

of

the dominating contribution of the fluidic pow er to the

equivalent drag.

Evaluating the measured drag per fluidic power reveals that the most efficient

use of the fluidic pow er occurs in the boundary control region. This is shown in

Fig.

40,

where

ACd /Cp ,

s a minimum. Th e magnitude

of

the incremental thrust

LID

EQUIV

0.00

0.1

0 0.20 0.30

CPf (FLUIDIC POWER)

Fig.

38

Pumping power required to achieve equivalentG A C C irfoil efficiency for

circular

TE,

AOA =

0.



224

‘d

EQUlV

GREGORY

S.

ONES

0.00

0.10 0.20

0.30

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 synthetic ets 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

0.1

5

0.20

0.25

Cpf (FLUIDIC POWER)

Fig. 40 Drag efficiency per fluidic power for GACC airfoil with circular TE,

AOA

=

0.




d

225

Fig. 41 Drag per power ratio for GAC C 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 com-

parison 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 2 3

ACI

Fig. 42 Lift per power ratio for GACC airfoil with circular

TE,

AOA

= 0.



226

GREGORY

S.

ONES

Table 2 Comparison of GACC lift increment-per-power

to similar powered systemsI2

Item

ACL/Cp, (ACi

0.5

ACLICp, (AC1

=1.0)

GACC h/c 0.0014)

Elliptic CC43

TE blown flap3'

Flap knee44 (BLC mode)

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 w ith 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 supercir-

culation 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 G ACC airfoil performance is compar-

able 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 Com parisons

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, ellip-

tic, 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,

h l c =

0.0022.




227

0.00 0.02 0.04 0.06 0.08 0.10

CP

Fig. 44 Comparison of lift performance for the GACC airfoil for different TE

shapes,

h l c

= 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 bound-

ry

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 GAC C having three different

TE shapes.



228

GREGORY

S.

ONES

the reactionary thrust, implying a net two-dimensional circulation induced thrust.

The d rag 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.

Com parisons of pitching m oments 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 com e at the price of mom entum 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

T o address the issue of a blunt TE for typical C C configurations at cruise, the

GA CC w as designed w ith 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.

It should be recognized that the cruise condition for this airfoil would be oper-

ated at a substantially higher Mach num ber 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 configur-

ations may provide. For cruise conditions, the CC performance characteristics are

Dual Blowing for Circular Coanda Sur ace

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.




229

0.00 0.50 1.00

1.50

2.00 2.50 3.00

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.



230

GREGORY

S.

ONES

0.001

0.010 0.100 1.000

CP(UPPER

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 per-

formance. 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 d rag 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

0.5 1 o

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).




23

0.20

0.15

0.10

0.05

c p 0.00

0.05

0.10

0.15

0.20

2 1

0 1

2

WAKE R AKE

POSITION

(UC)

Fig.

51

Wake profiles for the dual blowing cruise configuration of the GACC irfoil,

circular TE, reference Cpupper0.003,

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 jet s 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 0.100

CI+UPPER

LOWER)

Fig. 52 Equivalent drag for the GACC dual blown circular TE.



232

GREGORY

S.

ONES

U D

EQUlV

0.000

0.025

0.050

0.075

0.100

CP(UPPER 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 con-

sistent 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 d rag 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 co m arable steady

blowing performance?6937 The G ACC pulsed b lowing systemZBis based on a

high-speed valve that delivers a high volumetric flow to the upper jet ex it. 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, fre-

quency, and duty cycle. An exam ple of the pulse train i s shown in Fig. 54. The

quality of the rise time and decay of the pulse train i s related to the overall actua-

tor 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 con-

dition was identified to occur in the range 75 < < 90 deg, whereas steady

blowing was in because

60 < <

75 deg. This corresponds to the lift perfor-

mance 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 je t increases just dow n-

stream of the jet exit, then increases along the Coanda surface to peak at




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 P E G )

Fig.

55

Comparison

of

steady and pulsed pressure distribution for the circular TE,

h / c =

0.00106.



234

GREGORY

S.

ONES

2.0

DC=80

C=6O

DC=40 \

DC=20

c, 1.0

55

0.5 I

STEADY

0.0

I I I I

0

0.005 0.01 0.015 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.




235

0.01 0.10 1.00 10.00 100.00

8 10 12 14 16 18

b)

FU U (LREF:TE DIAMETER)

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.

1.5

c,

1.0

0.5

DC=40

DC=S?

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.

ONES

The performance benefit of the pulsed elliptic T E 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 ellip tic T E 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 sm all 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 C C 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.

U D

(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

efficiencycI/cdequiva em h




237

Th e 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 cam ber effects of the GACC

airfoil are demonstrated in the generation of higher lift for comparable momen-

tum coefficients. Comparing the GACC efficiency to a typical blown flap38

reveals the lift benefit of attaching the je t through C oan da turning. It is specu-

lated 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 L E blowing w as required to extend the lift coef-

ficient beyond

Cl RZ

for the dual radius flap.39 Th e poor efficiency of the jet

flap is generally related to the large blowing requirements associated with the

reactionary force:' and the min imal effect on the two-dimensiona l induced

pressure field.

The efficiency of the GACC's dual blown configuration highlights the low-

speed cruise conditions. Nominally, the lift requirements for cruise are

Cl

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 perform-

ance of the GACC airfoil is good, but has not been tested to its limits. It is rec-

ommended 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 GA CC airfoil section for use on an ESTOL o r 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 C C 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 . ~ ~or 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:



238

GREGORY

S.

ONES

where

and

and

2

ACm -*(L)92

PH

CL

4cotT

[

1 + (2 M2)&]quncom

Examples of the wall interference corrections described by Eqs. (A.l-A .4) are

small, as seen in Figs. A.l-A.4.

. a . + *

. . . .

.0000

0.0025

-0.0050

20 -1

0 0 10

AOA

0

Fig. A . l Angle of attack correction from wall interference (circular

TE).




239

0.025

M

-0.0501 I

20

-10 0 10

AOA

Fig.

A.2

Lift corrections from wall interference (circular TE).

0.010

0.008

0.006

ACm 0.004

0.002

A0 177

0 134

0 093

o . o o o ~

m e * * *

0.002

20

-10 0 10

20

AOA

Fig. A.3 Moment corrections from wall interference (circular TE).

A0 177

0 134

+

0 093

r e

20

-1

0 0 10

20

AOA

Fig. A.4 Dynamic pressure corrections from wall interference (circular

TE).



240

GREGORY

S.

ONES

B. Balance Corrections

are given by

Data reduction equations and tare corrections fo r pressure lines across balance

NF ~ F ( N F S C ) interactions)

AF 6AF(AFSC)-

AFinteractions

+ Pressurecorrection)

PM &M(PMsc) PMinteractionsPressureCorrection)

YM &M(YM sc) YMinteractionsPressureCorrection)

RM ~ M ( R M S C )

-

RMinteractions)

Pressure tare correction for axial, pitching moment, and yawing mom ent forces

are given by

where

where

Pressure Tarecorrection l P a c t +

2 P u p p e r

+ P l o w e r + 4 P ac tP u p p er + sPactPlower

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 Axial Pitching mom ent Rolling Yawing mom ent

( FS) ( F S )

( FS)

moment ( FS)

(

FS)

0.04

0 39 0.12 0.07

1.64




24

Table A.2 Summary

of

GACC pressure resolutions

Po

A P

AP

AP AP wall

freestream freestream model static rake signature

AP

MEMS

PSIA PSID PSID in. H 2 0 in. H 2 0 PSID

15 1 2.5 10 10

5

0.1080 0.0072 0.0360 0.0052 0.0052 0.0720

A m i n

(PSF)

A m i n

(PSF)

A m i n

(PSF)

A m i n

(PSF)

A m i n

(PSF)

A m i n

(PSF)

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 instrumenta-

tion 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 Josh , 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

une 2005, pp. 1-36.

3Moore, M. D., “Wake Vortex Wingtip-Turbine Powered Circulation Control High-Lift

System,” NASA/CP-2005-213509/PT2 une 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 AIA A/C EAS

Aeroacoustics Conference, AIAA Paper 98-2226, June 1998.

6Wood, N., and Nielson, J., “Circulation Control Airfoils Past, Present, and Future,”


’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,” NA SA, 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 Aerodynam ics,” Journal ofA irc ra f, Vol. 12, No. 6,19 75 ,

‘‘McCormick, B. W., Jr., Aerodynamics of

V / S T O L

Flight, Dover Publications,

”Wilson, M. B., and von Kerczek, C., “An Inventory of Som e Force Producers for Use

pp. 501-530.

Mineola, New York, 1999.

in Marine Vehicle Control,” DTNSRDC-79/097, Nov. 1979.



242

GREGORY

S.

ONES

13Betz, A., “Ein Verfahren zur direkten Ermittlug des Profilwiderstandes,” ZFM, Vol.

14Jones, B. M., “The Measurement of Profile Drag by the Pitot T raverse Method,”

”Schlichting, H., Boundary Layer Theory, 6th ed., McGraw-Hill, New York, 1968.

16Rae,W . H., and Pope, A., Low-Speed W ind Tunnel Testing, 2nd ed., Wiley, New Y ork,

1984.

17Kind, R. J., “A P roposed M ethod of C irculation C ontrol,” 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 Air-

craft,’’ Canadian Aeronautics and Space Journal, Vol. 19, No. 3, 1973, pp. 93-108.

”Jones,

G.

.,

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 Avia-

tion 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 Wind-

Tunnel 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 S tem Plane for

Submarine Applications,” David Taylor Naval Ship R&D Center, Rep. NSRDC/AL-200

(AD901-198), March 1971.

26Englar,R. J., “Low Speed Aerodynam ic 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., H emm erly, R. A., Tay lor, D. W., M oore, U. H., Seredinsky, V., Valck-

enaere, W.

G.

nd Jackson, J. A., “Design of the Circulation Control Wing S TOL D emon-

strator Aircraft,” AIAA Paper 79- 1842, AIAA Aircraft S ystems 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 S ystems and Technology Meeting, Aug. 1979.

29Kuethe, A. M ., and C how, C.-Y., Foundations

of

Aerodynamics, Bases

of

Aerody-

namic Design,

5th ed., Wiley, New York, 1998.

30Preller, R. F., and Rose,

0

J., “Langley Wind Tunnel Force Reduction Program,”

31Smith, D. L., “An Efficient Algorithm using M atrix Methods to Solve W ind Tunnel

32Ewald,B. F. R. (ed.), “Wind Tunnel Wall Correction,” AGARDograph 336, Oct. 1998.

331yer, V., Kuhl, D. D., and Walker, E. L., “Improvem ents to W all Corrections at the

16, 1925 , pp. 42-44 .

Reports and Memorandum, No. 1688, Brit. A.R.C., Jan. 1936.

NASA, CR-165650, NOV.1980.

Force-Balance Equations,” NASA , TN-D-6860, Aug. 1972.

NASA Langley 14x22-FT Subsonic Tunnel,” AIAA Paper 2003-3950, June 2003.




243

34Rose, R. E., Hammer, J. M ., and K izilos, A. P., “Feasibility Study of a Bi-Directional

Jet Flap Device for Application to Helicopter Rotor Blades,” Honeywell, Doc. No. 12081-

FR1, July 1971.

Rogers, E.

O.,

and Donnelly, M. J., “Characteristics of a Dual-Slotted Circulation


AIAA 42nd Aerospace Sciences Meeting, AIAA Paper 2004-1244, Jan. 2004.

360y le r, T. E., and Palmer, W. E., “Exploratory Investigation of Pulse Blowing for

Boundary Layer Con trol,” North American Rockwell, Rept. NR72H-12, Jan. 15, 1972.

37Walters, R. E., Myer, D. P., and Holt, D. J., “Circulation Control by S teady 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 W ing 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., “Som e 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 Vincen ti, 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 A irfoils,” NSRD C, 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.

35



Chapter

8

Trailing Edge Circulation Control of an Airfoil at

Transonic Mach Numbers

Michael

G.

Alexander,* Scott

G.

A n d e q t and Stuart K. Johnsont


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.

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

h / c =

nondimensional slot height

*Aerospace Engineer. Associate Member AIM

'Aerospace Engineer.

Copyright

005

by the American Institute of Aeronautics and Astronautics, Inc.

No

copyright is




245



246

M

G.

ALEXANDER, S.

G.

ANDERS, AND S. K. JOHNSON

x = chordwise distance, in.

y = span distance, in.

a

=


p = density, lbm/ft3

y

= ratio of specific heat

ACl/C,

= lift augmentation ratio

y / b = nondimensional span location

Subscripts

jet

=

air flow that exits nozzle

1

=

lower and lift

plenum

=

airfoil plenum

s

= slot

u =

upper

co = infinity

TE = trailing edge

0.25 = quarter chord

I. Introduction

IRCULA TION control (CC) is considered one of the most efficient methods

C

or lift augmentation at low Mach numbers.' The device augments an air-

foil'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 ~ gm e n t a t i o n . ~

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 C C 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 T E C C on a thin

airfoil section at transonic Mach num bers. 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 espectively.

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 m odel view from right rear quarter, looking upstream):

b)

CCA

airfoil section.



CC OF AIRFOIL AT TRANSONIC MACH NUMBERS

249

Table

1

Coanda radius and slot height dimensions

Coanda

1.78: 1 2.38: 1 2.98: 1

Chord, in.

rs, in.

rTE, n.

rsIC

bE/C

Guidelines: r/c

hllrs

W r s

h3vS

hl/rTE

h2/rTE

~ ~ I Y T E

Guidelines: hl r

27.82

1.44

0.25

0.052

0.009

0.024

0.039

0.051

0.14

0.22

0.29

28.09

2.57

0.19

0.091

0.007

0.02 to 0.06

0.014

0.022

0.028

0.18

0.30

0.38

0.01 to 0.08

28.36

4.02

0.15

0.142

0.005

0.009

0.014

0.018

0.23

0.37

0.48

was aligned with the slot exit to ensure the minimum exit area occurred at

x/c, f

=

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 gu idelines 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.

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 27.82 0.0350

0.0012

h2 28.09 0.0560 0.0020

h3 28.36 0.0730 0.0026

h4 28.36 0.0210 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 height-

to-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 remo-

vable 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 in to 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.




The m odel has the capability of holding six removab le, 0.050-in.-thick, high-

pressure-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.

A.

CCA Surface Static Pressures

A total of 83 external static surface pressure taps was loca ted at

y / b

=

0.5

on

the upper and lower airfoil surface (42 upper and 41 low er taps). There are two

spanw ise rows of ten static pressures taps located at

x/cref=

0.5 and 0.8 on each

upper and lower airfoil surface.

B. Coanda Surface Static Pressures

Each C oanda 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 v l b X1Cre-f

0.2

0.2

0.45

0.5

0.55

0.8

0.3

0.8

0.8

0.8

0.8

0.8



252


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 , 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 Tunne l” (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 deli-

vering a maximum of

1

lbm/s at 200 psia. Total temperature of the supp ly air was

uncontrolled and ranged from 13°F to 70°F . Each supply line was attached to a




253

Topview

Fig. 7 CCA model installation in the TDT.

control valve that regulated total pressure to the CC A m odel. A m anually oper-

ated 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 m oment coef-

ficient [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 m odel:



254


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 ob tained 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 end-

plate effects.

VII. Discussion

of

Results

The higher Reynolds number data will be presented first, because the exper-

imental 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 sia P sia To F Relft

0.3 2.7-4.1 2.6-3.8 67-94 3.6 lo5 o 5.5 lo5

0.8 3.0-4.1

2.0-2.7

95-125 7.8 lo5 o 1.0 lo5




255

Coanda Surface Slot Height

Fig.

8

Coanda surface effect, upper slot blowing, M ach

=

0.8, a

=

3

deg.

surface was capab le 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 nega-

tive 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 Coand a 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 fol-

lowed in Fig.

9

for lower surface blowing. However, the low er surface blowing

is not as effective in producing lift increm ent as the upper surface blowing over



256



Fig.

9

Coanda surface effect, lower slot blowing, M ach

=

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 .




257

Coanda Surface

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 sma ller 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


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

flow turning around the Coanda bulb as a result of the increased

C

On the

upper surface blowing (Fig. 16), as the slot size

h / c

was increased, the




259


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 augmenta-

tion 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.

How ever, 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



Fig. 13 Slot height effect, lower slot blowing, M ach =

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 he greater ACl and

ACm

generated for the upper (Fig. 20)

and lower (Fig. 21) slot blowing.

The lift augm entation 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 slo t, the

greater its augmentation ratio becomes.

C.

Nozzle Pressure

Ratio

In Fig. 24, incremental lift data are presented at Mach num bers of 0.8 and 0.3

as a function of nozzle pressure ratio

NPR).

The surface and slot height noted in




261

Coanda Surface

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 ve locity 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



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 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




263

Coanda Surface

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



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 nega-

tive 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)

Fig. 18 Lift augmentation,Coanda surface effect, upper slot blowing, Mach = 0.3, Y = 6 deg.

-I

n

9

z

cn

P

5

5

I

z

5

rn

n

cn

N

Q

cn



Slot h/c=0.0012)

Fig.

19

Lift augmentation, Coanda surface effect, lower slot blowing, Mach

=

0.3,

a = 6

deg.

N

3,

Q



a



c

v



Coanda l.781)

Coanda 2.381)

Fig. 22 Left augmentation, slot height effect, upper slot blowing Mach = 0.3,

a

= 6 deg.



Coanda 2.381)

N

?

rn

X

D

z

n

rn

v,

D

z

n

v

rn

z

v

P

Fig.

23

Left augmentation, slot height effect, lower slot blowing Mach

=

0.3,

a

= 6 deg.




_

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

i \

~

Mach AOA ; CoanddS; /c ,

............................ I= 0.3; a = 6f 1.78:1)/0.0012 I

M

=

0.8,

a =

3 f 2.98:1)/0.0012

........................................

_............

i

...................................................................

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0

-0.1

-0.2

-0.3

-0.4

-0.5

-0.6

-0.7

-0.8

Fig. 24

Upper surface blowing

L ~ ~ ~ ~ I i

c . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . .. .

.

. . . . . .

......

.... .......,...

c L

0

1

2 3 4 5

6

NPR

Lower surface blowing

0 1 2 3 4

5 6

NPR

Nozzle pressure ratio vs

A C ,

upper and lower slot blowing.



272 M G. ALEXANDER, S. G. ANDERS, AND S. K. JOHNSON

Upper surface blowing

0.8

0.1

0.6

0.5

0.4

0.3

0.2

0.1

0

0.5 1 1.5 2

2.5

3

3.5

4 4.5

UjetN inf

Lower surface blowing

cI

0

-0.1

-0.2

-0.3

-0.4

-0.5

-0.6

-0.1

I " " I " " ~ " " " " I " " ~ " " " " _

~ Mach

AoA; CoandaiSlotG / c

0.5 1

1.5 2

2.5

3 3.5 4

4.5

U je t N in f

Fig. 25 Velocity ratio vs AC upper and lower slot blowing.




273

Upper surface leading edge pressure distribution

Fig. 26 Pressure distribution, C, effect, upper slot blowing, Coanda 2.98:1), slot

( h / c

= 0.0012), Mach = 0.8, (Y = 3 deg.

-1.5

-1

-0.5

0

0.5

-0.1 0 0.1 0.2 0.3 0.4

Upper surface leading edge pressure distribution

No BlowingCµ = 0.002

Cµ = 0.003

Cµ = 0.004

Cµ = 0.006Cµ = 0.008

Cµ = 0.009

Cµ = 0.011

Cµ = 0.012Cµ = 0.014

Cµ = 0.017

Cµ = 0.02

CP

x/c

Upper Surface

Airfoil Leading Edge

-1.5

-1

-0.5

0

0.5

0.92 0.94 0.96 0.98 1

Upper surface trailing edge pressure distribution

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

CP

x/c

Coanda Bulb

Upper Surface

Aft upper surface



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.




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 R ogers 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.

Englar, R. J., “Investigations into and Application

of

the High Velocity Circulation


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 N oise Reduction,” G TRI 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. DTNSR DC/ASE D-79/07 (AD-A084-176), May 1979.

’Nielsen, J. N., and Bigger, J. C., “Recent Progress in Circulation Control Aerody-

namics,” 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 Tr ip Technique Study in the McAir Advanced Design Wind

Tunnel,” Technical Mem orandum 4395, M ay 1984. (The McAir name is used interchange-

ably with McDonnell Aircraft Company.)

“Staff, Aeroelasticity Branch, “The Langley Transonic Dynam ics Tunnel,” LWP-799

Sept. 1969.

Englar, R. J., “Two-Dimensional Transonic Wind Tunnel Test

of

Three 15-percent

Thick C irculation 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 2005-

213545, March 2005.

2

11



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'

W est Virginia University, Morgantown, West Virginia

Nomenclature

A =

area

b

=

span,

f t

C

=

coefficient of

c = chord, ft

D

= reference lengths, f t

F

= download force, lbs

h =

height,

f t

L =

length,

f t

P = pressure, lb/ft2

Re =

Reynolds number

V

=

velocity, ft/s

p

=

density, slug/ft3

Subscripts

=jet

p

=

blowing

= freestream

1, 2, 3, 4 =

reference numbers

*Graduate Research Assistant, Mechanical and Aerospace Engineering. Student Member AIAA.

'Assistant Professor, Mechanical and Aerospace Engineering. Mem ber AIAA.

'Professor, Mechanical and Aerospace Engineering. Member AIAA.


reserved.

277



278

G. ANGLE II, B. O HARA, W . HUEBSCH, AND

J.

SMITH

I Introduction

HE COA ND A effect can be described as the balance between the inertial and

T

ormal 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 separ-

ating 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 je t is relatively insensitive

to Reynolds number Re as defined below, provided it is in excess of a threshold

value of

40,000.

Thus

'*

P PcO)V,.VcO

PV2

R e = [

where

P

is the local pressure,

P ,

is atmospheric pressure,

vj

and

Vi,

are the je t

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, main-

taining the composition of the je t and increasing the bulk jet velocity. The goal of

this work is to use blowing slots to induce the C oanda effect in the leading edges

(LE) and trailing edges (TE) of the airfoil.

Param eters 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

1

2p ,

V i

b

,

=

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 and represent the je t 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 con-

structed 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 fo r wind tunnel testing.

How ever, this size was needed for the desired instrumentation fo r the two-dimen-

sional 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 CD s the adjusted download force coefficient, CD is the measured down-

load force coefficient, A is the model frontal area, and is the test section cross-

sectional area. Surface pressure readings w ere taken on this model using m ultiple

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 calcu-

lating 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 0.1

1

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:




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 ce lls. Data were collected

for each test point for a four-minute test samp le, 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 base-

line average was computed from the three baseline runs to be used as the refer-

ence 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 0.01

5 0.02

Signal Vo l tage (V)

Fig. 3 Calibration curves for the three load cells.



282

G. ANGLE II, B. O HARA, W . HUEBSCH, AND

J.

SMITH

111. Com putational 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 WV U 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 .76 in., 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 no-

slip walls. The mesh was created using unstructured triangular cells. Additional

grid poin ts were clustered around the blowing slots and immediately dow nstream

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

60

84



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 coeffi-

cient. 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


J.

SMITH

Table

1

Computational slot velocity and

correspond ing momentum coefficient

Slot

v

t/s

P

0

10

60

130

2

0

0.0002

0.0068

0.0319

0.0756

After the initial conditions w ere all set, the first step was to find an initial sol-

ution. 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 sol-

ution. Each lam inar solution was than solved again using second-order upwind-

ing 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 solv er with enhanced wall treatment

was selected bec ause it has been fo und to yield good results fo r cases dealing with

circulation control (CC ) by Ch ang et al.5 A quick study was also carried out com-

paring the different solver types available in Fluent 6.1. The RNG k-e solver

Fig.

6

Computational grid near the leading edge.




Table 2 Experimental slot pressure and

corresponding momentum coefficient

Slot

P

si

c

0

5

10

15

20

25

0

0.0116

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 five-

thousandths 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

1.02

u

g 0.98

y 1

B

n

-

-

m

.96

q

0.94

0.92

-.-

0

0.02 0.04

Blowing Coef fic ient

( )

Fig.

7

Download force variation with blowing coefficient.

0.06



286


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 dow n-

load 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 down-

load 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 place-

ment 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 mechan-

ism 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 pitch-

ing moment of the airfoil. By adjusting the blowing pressures separately, the

pitching moment can be altered. With further testing, this potential benefit can

be be tter defined. Additional experimental data can be found in A ngle e t a1.*

V.

Com putational 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 pos-

ition 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 exam ple, where the flow is circulating, with path-

line, 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

LE

only

TE

only Both

0

5

10

15

20

25

0

0 0 0

0.01 2.84

-0.35 -3.12

0.02 0.63 -1.16 .59

0.03 3.88

1.07 2.29

0.05

6.67 -1.13 5.08

0.06 9.23 -0.77 8.68




Fig. 8 Pathlines colored by Particle ID near the leading edge.

active CC studies such as the one carried out by Sw anson et a1.6 Pathlines c an only

tell part of the story for download reduction; they illustrate the path of ai r particles

but do not really show turbulence or velocity gradients. Upstream of the wing,

the flow appears to be mostly uniform and lamin ar, 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.




Fig.

12

Vector plot near the trailing edge.

shows a similar trend. The computational results were all computed a t sea-level

standard atmospheric conditions whereas the experiment was conducted in Mor-

gantown, W V, at an elevation of 1240 ft. Despite this difference, which was

accounted for in the use of force and pressure coefficients, the amount of down-

load 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

....................................................................................

-

A

-

...................................................................................................

-

2

xperimental

5 ..............................................................................

0 1

0.01

0.02

0.03 0.04

0.05

0.08 0.07 0.08

Blow ing Coef f i c ien t Cp)

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

Blo win g Coeff ic ient (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,

5

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.




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 decen t 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 prac-

tical 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.

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 W ing,” NASA

TM-102244, Feb. 1990.

10



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

C L= 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

P ,

=

total pressure

R = range

Re = Reynolds number

S

=

wing span area b

x chord)

U

= velocity

V

= velocity

PR

= total pressure ratio of engine compressor

Wo= takeoff gross weight

W1 = empty weight

y+

=

nondimensional length scale for turbulent boundary layer

*Associate Professor, Department of M echanical and Aerospace Engineering.

+Graduate Student, Department of M echanical and Aerospace Engineering.


reserved.

293



294 G.-C.HA AND C. D. PAXTON

a = angle of attack

p = density

7= efficiency

=

ratio of specific heats

E = turbulent dissipation rate

Subscripts

= freestream

=jet injection

L = landing

TO = takeoff

T

=

touch ground

I. Introduction

ACHIEVE high-performance aircraft design, revolutionary technology

T dvancement should be pursued to d ramatically 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 conven-

tional aerodynamic design limit and bring dramatic performance improvement

to aircraft.'-3 The National Aeronautics and Space Adm inistration (NASA),

U.S. Air Force, and aerospace industry have recently made great efforts to

deve lop f low con t ro l t e~hnology .~ -~o enhance lift and suppress separation,

various flow control techniques have been used, including a rotating cylinder

at the leading ed ge (L E) and trailing ed ge (TE),3,899 irculation control (CC)

using tangential blowing at LE and TE,1°-16 multi-element pulsed

jet separation c o n t r 0 1 , ' ~ -~ ~nd

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 substan-

tial improvement in aerodynamic performance, which primarily includes lift

enhancement, drag reduction, and stall margin increase (suppression of separ-

ation); 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 implem ent.

The rotating cylinder method is generally most effective when the LE or T E

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 high-

lift flap system increases noise during landing.23

relies on a local favorable pressure gradient on a curved

surface to attach the flow-the Coanda effect. Such a favorable pressure gradient

A

cc

airfoillO,l

1,15,16



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. T o make the C C 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 T E 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 favor-

able 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 regu lar 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 je t or

pulsed je t 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 syn-

thetic je t 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 show n in Ref. 19 using the periodic synthetic jet show about 35% increase

of the

C

and little increase of stall AOA, while the co-flow je t airfoil tested in

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 je t flow control airfoil

studied in Ref. 19, which will increase the aircraft weight.

The new airfoil flow control technique using the co-flow je t (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 m ixing, which enhances lateral transport of energy from the je t 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 gradi-

ent enhances je t 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.HA 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 a irfoils 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 sam e amount of mass flow, the je t

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 je t mass flow rate, throughout the mission according

to different needs. No moving parts a re 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 o r 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

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

super-

sonic 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 maneuver-

ability, 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 p resents the

CFD

results and analysis, which are the

basis fo r the wind-tunnel tests of proof of concept conducted in Refs. 25 and 28.

The detailed

CFD

da ta also provide a very useful qualitative physical insight into

the working mechanism of the

CFJ

airfoil.

and landing.

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




basel ine air fo i l

suct ion slot

nject ion

slot

297

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 CF D 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 bound-

ary 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 C FD 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 d irection around the airfoil and in the radial direction, respect-

ively. 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 down-

stream boundary extended to 30 chord length, upstream, lower and upper bound-

ry

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


Fig. 2 Zoomed m esh around the airfoil with co-flow slot.

A. 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 m inimum increase in the order of magnitude. W hen 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 con-

verged based on the mesh size.

Figure

4

show s 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.




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 isen-

tropic Mach number also indicates the approximate Mach number outside of

the wall boundary layer assuming that the total pressure loss is small.

Figure

5

show s that the CFJ airfoil creates a very strong suction effect near the

LE and the flow is accelerated from the freestream M ach num ber 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



300

1.6

1.4

5 1.2

f

= 1 -

8

0.8

c

.-

0.6

-


-

-

-

-

baseline airfoil

Fig. 4 Streamlines at angle of attack

of

20 deg.

_ _ _ _ _ _

aseline

I of low

’.

W h J I I I I I I I I I I I I I I I I I I

0 0.25 0.5

0 75

1

WChord

Fig.

5

Surface isentropic Mach number distribution at angle

of

attack

of

20 deg.




30

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


from the jet to the main flow. A high mixing rate is therefore desirable. As indi-

cated by Gre itzer et al.,27 the adverse pressure gradient enhances je t mixing.

Based on this principle, the injection slot of the CFJ airfoil is located dow nstream

of the LE suction, as shown in Fig.

5 .

After the LE suction, the pressure continu-

ously 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. How ever, 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 injec-

tion to the LE, the more effective the CFJ, but the injection must be located down-

stream of the LE suction peak.

Unlike the C C 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 favor-

able pressure condition for injection and suction and may need less energy to

pum p the same amount of jet mass flow than does a CC airfoil.

In this study, although the AO A 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 je t injection, which determines

the jet mass flow rate of the je t and the je t injection velocity. The je t mom entum

coefficient therefore varies with AOA. The jet momentum coefficient based on

the conventional definition is given by

m

vj

c -

-.5pmU&S

where m is the injection m ass 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 air-

foils, 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 gener-

ated. The thrust is primarily generated from the strong LE suction, which is the

same mechanism as a flapping bird wing generating both lift and




303

0.2

c

0

0.05

0

Fig.

7

Jet mom entum coefficient against angle of attack.

2.5 -

2 -

1.5 -

1 -

0.5 -

I

base l ine

cof low

- - - - - .

-0.5

I

I

I \

I

I

I

I

I

I

I

I

I

I

I

I

I

0 0.1 0.2 0.3

Cd

Fig.

8

Drag polar for the baseline and CFJ airfoil.



304

0.3

0.25

0.2

0.15

3

0.1

0.05

0 -

-0.05


-

-

Cdpressure

Cd f r ic t ion

- - - - - .

- _ - _ -

-

-

-

. I . .__

c

I L r r r r ~ ' ' ' ~ ' ' ' ' ~ ' ' ' ' ~ ' ' ' ' ~ ' ' ' ' ~ ' ' '

Fig.

9

Calculated drag coefficients against angle of attack for CFJ airfoil.

The d rag 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 10shows 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

/

t

0.05

0

AOA

Fig.

10

Calculated drag coefficients against angle of attack for baseline airfoil.

1.1

_ _ _ _ _ _ .aseline

1.075

coflow

1.05

t

1.025

0.975

0.95

.925

Fig. 11 Wake shape for the baseline and CFJ airfoil at (Y =

0

deg.



CONTROL METHOD USING CO-FLOW JET 307

When U is greater than

Urn,

he drag is negative and becomes thrust. W hen the

AO A is very large, the je t energy is mostly used to diffuse the flow to ma ke 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. Th e pressure d rag is there-

fore overwhelming, which is desirable for short distance landing.

B.

Energy Expenditure

The CFJ airfoil achieves performance enhancement using the powered co-

flow je t, 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 je t is usually sourced from the engin e. 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 com -

pressor. In othe r words, an extra am oun t of fuel needs to be burned to drive the

compressor with the CFJ pumping system. The loss resulting from the CFJ is

given as

Powercfj

Loss =

Powercompressor

here

yiz fj

is the ratio of the CFJ mass flow rate to the engine mass flow rate,

hcfj= h c f j / h e n g i n e , q&the efficiency, and PR is the total pressure ratio at the

inlet and exit. The hcfj s 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.

Th e 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.

Th e fuel consumption dramatically increases and the thrust is greatly reduced if a

part of the flow is bled from the engine. Th is can be seen by applying a control

volume to an engin e and, assuming that the flow at the engin e exit is expanded to

ambien t pressure, the thrust is given by

F

=

(hinlet + hfue1)vnozzle- inletvinlet

=

h n o zz lev n o zz le - inletvw 5 )

It is obv ious that the bled mass flow from the com pres sor will dec rease hno zz le,

which will directly reduce the thrust. Assuming that the engine is running on the

ground with V = 0, the thrust becomes



308


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 SF C 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 w ing weight. These

advantages will be shown in the next section.

C.

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 air-

frame-propulsion system. The military aircraft F-5E is selected as the

example, because a detailed mission analysis has been conducted by Roth

et a1.32-35 and data are available. In Refs. 32 -35 a generalized vehicle thermo-

dynamic 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 d rag to the total fuel con-

sumption, which is particularly useful for an aircraft designer in finding and opti-

mizing 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 esti-

mate 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

Mission Analysis of F-5E Aircraft Using CFJ Wing




Table

1

Work potential and fuel consumption of

F-5E

intercept mission

based on Roth’s model”

Component Work potential, hp.min Fuel consumption,

Propulsion

Compressor

loss

585

Compressor PR

585

mass flow rate

Total engine

loss

Total engine thrust

Total propulsion

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

585

Vcompressor

Airframe

27,108

7.8

89%

53.6

kg/m

113,762

191,529

305,291

55,151

19,268

32,504

26,052

9,116

9,572

3,466

26,061

5,446

186,636

8.88

37.3

62.70

100

18.06

6.32

10.65

8.53

2.99

1.14

8.54

1.78

61.11

3.10

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 formu-

l a t i o n ~ * ~ :

Table 2 CFJ wing parameters for F-5E-CFJ

Parameters Value

V C t j

60%

CFJ PR

4

Wing span

1 2

baseline

Wing weight

1 2

baseline

CFJ

CL

wing

2

baseline

CFJ

Cd

wing skin

+

wave

1 8

baseline

CP

0.1



31


Table 3 Estimation of fuel consumption

of

F-5E intercept mission using

recirculating CFJ wing

Work potential, Fuel consum ption, CFJ benefit,

Component hp.min

Compressor loss 32,003 10.48 - .6

Turbine

loss

17,714 5.8 - .2

Total engine

loss

119,386 39.11 - .81

Total engine thrust 185,905

60.89 1.81

Total propulsion 305,291

100 0

Airframe

Wave + skin

friction

Fuselage drag 55,151 18.06

0

Tail drag

19,268 6.32

0

Structure weight 22,929 7.51 1.02

Propulsion weight 9,116 2.99

0

Fixed equip. weight 9,572 3.10

0

Stores weight 3,466 1.14

0

Fuel

+

misc. 26,061 8.54 0

Store drag loss 5,446 1.78 0

Wing drag 2,03 1.5 0.66 9.98

Induced drag

weight

Total airframe

loss

153,040 50.13 9.17

Endurance 41.3

Range 37.7

and

where

WO

and

W1

are the takeoff gross weight and the weight with empty fuel

tank, and ct is the specific fuel consum ption.

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 pum p the

CFJ

increases the-

compressor loss by 18%, which generates a penalty to the total fuel consumption

of 1.8%. The largest gain for the fuel consum ption is from the dra g 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.




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 CFJOO25-

065-196 airfoil with

C

= 0.29,25928he

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 ~ ~

s 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 je t causes strong turbu-

lence 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 circula-

tion. 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 com-

bined with other flow control techniques.

The C FD simulation indicates that the CFJ airfoil performance is more sensi-

tive to injection location than to suction location. The injection location should be

as close to the L E as possible, but must be dow nstream of the L E suction peak to

make use of the adverse pressure gradient as enhance jet mixing with the main

flow.

For the NAC A2415 airfoil studied, at low AOA with moderate m omentum jet

coefficient, the CFJ airfoil will not only significantly enhance the lift, but w ill also

dramatically reduce the drag, or even generate negative drag (thrust). The mech-

anism for this is that the co-flow je t reduces the pressure drag by creating very

strong LE suction, and can generate negative pressure drag greater than the fric-

tion 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 a irfoil 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 subso-

nic 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


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 air-

frame 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) dra-

matically reduces drag and can achieve very high C L / C D t 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

Air Vehicles,” A IAA Paper 2004-3785, June 2003.

‘Sellers, W. L. I., Singer,

B .

A., and Leavitt, L. D., “Aerodynamics for Revolutionary

’Gad-el Hak, M., “Flow Control: The Future,”


Vol. 38, 2001,

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 2004-

2622, June 2004.

6Miller, D., and Addington, G., “Aerodynamic Flowfield Control Technologies for

Highly Integrated Airframe Propulsion Flowpaths,” AIAA Paper 2004-2625, June 2004.

pp. 402-418.




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.


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, W right-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 A ttack,”

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 B lowing,”

Journal

of

Aircra f ,

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,

Englar, R. J., “Circulation Control for High Lift and Drag Generation on STOL

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 M ulti-Element Airfoils,” AIAA Paper 92-2636, 1992.

”Wygnanski, I., “The Variables Affecting the Control Separation by Periodic Exci-

tation,’’ AIAA Paper 2004-2625, June 2004.

”McM anus, 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 McM anus, 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 Evalu-

ation of the Steady and Pulsed Jet Effects on the Performance of a Circulation Contro l Wing

Section,” AIAA 42nd Sciences Meeting and Exhibit, AIAA Paper 2004-0056, Jan. 2004.

25Zha, G.-C., Carroll, B., Paxton, C., Conley, A., and W ells, 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 A irfoil Circulation Augm ent 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 Astro-

nautics and Aeronautics.

27Greitzer, E. M ., Tan, C.

S. ,

and Graf, M. B.,

Internal Flow

Cambridge Univ. Press,

Cambridge, UK, 2004.

10

pp. 175-181.

16

Aircraft,”

Journal

of

Aircra f ,

Vol. 12, 1975, pp. 457-463.



31


”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 m embers Car, D. and Copenhaver, W .), “Super Diffusion Cascades

Using Co-Flow Jet Flow Control,” National R esearch Council Summ er Faculty Final Rept.

Aug. 2002.

31DeLaurier,J., “Work

on

Flapping-Wing Flight,” Lecture,

23rd

AIAA Applied Aerody-

namics Conference, June 2005.

32Roth,B. A., “A Theoretical Treatm ent of Technical Risk in M odem 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


Nomenclature

A j = blowing slot area

b = wing span, ft

c = chord length, ft

cg = cen ter of gravity, ft

CD= three-dimensional drag coefficient

CL= three-dimensional lift coefficient

CDE= equivalent drag coefficient

C = maximum lift coefficient

CM25,CM quarter chord pitching moment coefficient

CT = thrust coefficient

C, =jet momentum coefficient [see Eq. 2 ) ]

C = et m om entum coefficient, leading-edge blowing

CpChw

=

et momentum coefficient, Channel-Wing blowing

Elev = elevator deflection angle

hslot, j = blow ing je t slot height, in.

iT

= tail incidence angle, deg

m =jet mass flux, slugs/s

q

= freestream dynamic pressure

(=

V2) , psf

Pd

PD ,P , =

duct total pressure

*Principal Research Engineer, Aerospace Transportation Lab., Georgia Tech Research Institute.

'Senior Aerospace Engineer, Configuration Aerodynamics Branch.


Associate Fellow AIAA.


315



31

6 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 je t velocity, isentropic, ft/ s

W / S

=

wing loading, psf

x/c = nondimensional chordwise location

x

= moment center location, ft

xTO = takeoff distance, ft

a

=


p = freestream density, slugs/ft3

p = blowing je t density, slugs/ft3

&lipstream = slipstream deflection angle, deg

ACL= incremental lift coefficient

I. Introduction

HE ABILITY to achieve Super-STOL (short takeoff and landing) or V/

T

TOL (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 air-

craft. Although such capability has direct application for military missions such

as those of a tilt-rotor o r 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 tech-

nology 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 circula-

tion 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 flight-

tested and further developed by Custer in the mid-1960~,~mployed 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



PNEUMATIC POWERED-LIFT SUPER-STOL AIRCRAFT

31

7

Three-view

of the Cuetor Chann el 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 sa me manner a s 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

a s dem onstrated by C ~ s t e r . ~owever, 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 CLwas from redirected thrust, and less from circulation

lift augmentation.

2 )

High cruise dra g could result from the channel’s extra surface area.

3) Asymmetric thrust yields asymm etric mom ents 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 improvem ents.

4)

Channel LE and TE separation could occur at high angle of attack

a.

5 ) Poor low-speed control is available from conventional aerodynamic sur-

6 There is nose-down pitch from aft propeller loading on the w ing.

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 sur-

faces 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) oncept of Fig. 4, where tangential blowing on a

highly curved TE behind a je t eng ine 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 STO L, 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

faces at low speeds.



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 avail-

able 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 6,7

and

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

vwt

de?wilm sl

OCHllis

lnd

t u m d m o d d

Fig. 4 Previously developed CCW/upper surface blowing powered-lift concept:



320


predictions before the current wind-tunnel test data were available. These implied

CL alues approaching 9-

10

for a pneumatic channel wing aircraft with blowing

on outboard CCW wing panels at relatively low aircraft angle of attack. Higher

CL

alues 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 CL f 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 experi-

mental 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 de g

+ a).

It was already

known (Fig. 4) that thrust deflections up to 165 deg yielded by blow ing were a

Fig. 5 Conceptual PCW Super-STOL transport configuration.




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

CC W wing section were also blown for separation con trol. The em phasis 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 tech-

niques were employed to account for the installation of the active propeller in

the channel (see below). Som e 980wind-tunnel runs (including propeller calibra-

tions) 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 (incre-

mental 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 CT andC,. 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


A. Tunnel Test Results, Outboard Wing

ON

In Figs. 7a and 7 b 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 CLmax f the unblown configuration with virtually no reduction in the stall

angle, astall he C, values shown are comparable to or greater than those that

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 pro-

gress 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 / ( q S ) ,

where T is the calibrated unin-

stalled 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

) Lift-drag polars.




Channel Wing

CT

Fig. 8 Effects of prop thrust variation on lift and drag at constant blowing C,) and

Y

= 0 deg.

a comm on reference basis to determ ine 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

CD

ncludes the input thrust, which cannot reasonably be sep-

arated from the aerodynamic drag alone once the prop is in the channel. Measured

CD an therefore be (and sometimes is) negative. After the initial low values of

CT

are exceeded, C L ncreases nearly linearly with

CT,

and

CD

educes nearly lin-

early. (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

=

10deg. Th e 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 visualiza-

tion showed that the initial stall

a

=

15-17

deg) seen for most of the lift curves



324


Channel Wing C k

Fig.

9

Effects of blowing variation on lift and d rag at constant

CT

nd 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 posi-

tive 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 ma ximum slot arc of 160 deg was most effective.

Blowing of m ore than 160 deg of channe l arc was not appropriate on this model

because the last 20 deg of inboard arc was along the chan nel right next to the

fuselage, and b lowin g there w ould do little more than bou nce off the fuselage.

Th e effect on increased tail-off pitching m oment 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 a re referred

to the ch annel’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 problem s 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, a s well as for L E separation prevention.




Angle of Attack, a dcg Angle of Allack, a deg

Fig. 10 Effects of blowing, Ch and a on lift coefficient, stall angle, and drag

coefficient for the P CW model with unblown ou tbo ard wing.

Fig.

11

Effects on lift and drag of varying blown channel slot arc length at

constant C, and a t

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 C k or CT

Fig. 12 Effects of prop/nacelle location, blowing and thrust on quarter-chord

pitching m oment, 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 conven-

tional horizontal tail installed at the midvertical location on the aft fuselage.

Needless to say, not all of the lift values show n in Fig. 13 are trimmed long itud-

inally. Furthermore, for the

CT=

case with blowing on, the conventional tail of

the aircraft stalled experimentally over much of the lower

a

ange.

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 configur-

ation 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

= +2 0 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




I I

R e u m a ~ c hannel WmgONLK, Outboard

CCW

OFF,ROD N,Wd Tad

a q r c n

Fig. 13 Effect

of

thrust and/or blowing increase on lift and pitching moment

variation with

a

or channel-wing-only configuration (no outboard wing panels)

with tail at midlocation, i = 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 experi-

enced 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

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 visual-

ization 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 center-

of-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 CT = 3 case

of a tail-less PCW without outboard wing. At

C,

=

0,

moving the center-

of-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 nose-

up 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 pitch-

trim device without lift loss due to tail download.




MTF063,

i4

Pitching Moment, Prop ON,

N o Outboard

Blo-g,Tsil Power

CT-2.2,CmuChW-1.0

,

CCW

Rap4 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,

Outboard

CCW

OFF, Rop

ON Tad

Effects. Pneumahc

Channel WmgONLY.

Ou t h a rd

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


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 attachm ent caused by LE blowing on outboard CCW and channel flow

entrainment at Y

=

20 deg, channel LE not blown: a) Outboard LE slot unblown and

b) outboard LE slot blown.




331

a)

9

8

7

6

d 5

4

3

2

1

0

10

9

8

7

6

5

4

3

2

1

MTF068, Pneumatic Channel

Wng,

Phase

111

CL v s CM, Run

799,

Cp Channel=O.O,

Channel Only, Prop

ON,

CT.3.0. Tai l OFF

1

0

CM

1 2

+xmom=.375

-+Xmom=.SO

I

_ _ _ _ .

I

__ .

I... _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ I

__ .

2 1 0 1

CM

Fig.

19

Effect of a ft center of gravity location on pitching moment curves for the

tail-less

PCW

at

CT

= 3 and two blowing values, x = x,.Jc: a)

CT = 3, C, =

0,

b) CT =

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/CC W flight-test data. W hereas the prop/electric



332

a)

l

9

8

7

cL

6

5

4

3

2

I

0

R. J.

ENGLAR AND B. A. CAMPBELL

. . .

, ,

. . .

,

Solid

Line

Data Derived from CCWNSB, a1

Tunnel Data from

GTRl

MTF055

a=lW

cl=fl.fl

0 1

0.2

0.3 0.4 0.5

0.6

0.7 0.8 0.9 1.0

CCChW

y

olid

Line Data Derived

,rom

CCWIUSB,, a = l W

I

9

Tunnel Data

from GTRI MTF055

0.0 0.1 0.2

0.3

0.4

0.5

0.6 0.7 0.8 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 sym bols) 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 B aseline Tilt Rotor

a t 100 ground roll

Fig. 21

Pneumatic channel wing predicted super-STO L takeoff performance.




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 stud ies conducted at NASA L aRC 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 dem onstrates that this tech-

nology has the potential to enable

simple/reliable/effective

STOL and possibly

VTOL operations of personal and business-sized aircraft operating from remote

or sm all sites as well as increasingly dense urban environm ents. 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 con-

firmed the potential aerodynamic payoffs of this possible Super-STOL configur-

ation, 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 charac-

teristics (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 Super-

STO L approaches.



334


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 40-

45

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 signifi-

cant 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 S TO L operation.

Large nosedown pitching mom ents 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/S TO L there is no download on the wing

from prop thrust because the PC W props d o not tilt. The potential of PCW for an

integrated lift, thrust/drag interchange, and control system, all from one se t of

devices, holds promise in terms of simplicity, weight reduction, and

reliability /maintainability.

The projected operational benefits based on these early data suggest Super-

STOL 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 pro-

pulsion 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 investi-

gated)12 may further increase lift efficiency and reduce already low b lowing

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 mili-

tary 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 per-

sonal and business-type aircraft.

Future testing, evaluation, and development still need to be accom plished to

address possible pitch-trim problems, performance at higher

CT

nd 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 fol-

lowing should be experimentally investigated:

1) Use of pulsed blowing to further reduce required blowing m ass 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

vailability.

3)

Further evaluation of low-speed controllability and trim, including evalu-

ation 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 aerody-

namic/pneumatic trim and control devices (blown canards, for example).




335

The earlier mission analyses should be revised to incorporate the experimen-

tally 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 m odel with variable yaw capability shou ld 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 a t GTRI.

References

‘Pasamanick, J., “Langley Full-Scale-Tunnel Tests of the Custer Channel Wing Air-

plane,’’ NACA RM L53A09, April 1953.

’Mitchell, K. A., “Mr. Custer and His Channel Wing Airplanes,” Journal o American

Aviation Historical Society, Spring 1998.

3Blick,

E.

F. and Homer, V., “Power-on Channel Wing Aerodynamics,”

Journal

o

Air-

craft,

Vol.

8,

No. 4, 1971, pp. 234-238.

4Englar, R. J., “Circulation Control Pneumatic Aerodynamics: Blown Force and

Mom ent Augm entation and Modification; Past, Present and Future,” AIAA Paper 2000-

2541, AIAA Fluids 2000 Meeting, Denver, CO, June 19-22, 2000.

’Englar, R. J., and Applegate, C. A., “Circulation Control-A Bibliography

of



6Englar, R. J., “Development of Circulation Control Technology for Powered-Lift

ST OL 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., “Devel-

opment 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 M eeting, Aug. 1979.

’Hines, N., Baker, A., Cartagena, M., Largent, M., Tai, J., Qiu,

S.,

Yiakas, N., Zentner, J.,

and Englar, R. J., “Pneumatic Channel W ing Comparative M ission Analysis and D esign

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 Bound-

ary 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,”


Vol. 36,

‘’Jones, G.

S. ,


Through Pulsed Blowing”, AIAA Paper 2003-341 1, AIAA 21st Applied Aerodynamics

Conference, June 2003.

NO. 2, 1999, pp. 332-339.



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

C L= lift coefficient

Cl

=

rolling moment coefficient

C = pitching moment coefficient

c

=

chord, mm

c =

standard mean chord, mm

=

mean aerodynamic chord

MAC),

mm

c , = root chord, mm

c , = tip chord, mm

D = drag,

N

h

=

slot height, mm

L

=

lift, N

= 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

CLc0,=

initial lift coefficient

Cl o)=

initial rolling moment coefficient

*Postgraduate Research Student, Fluid Mechanics Research Group, Aerospace Engineering.

'Professor, Head of Department, Aerospace Engineering. Senio r Member

AIAA.


Member

AIAA.


337



338

S.P. FRlTH AND N.

J.

WOOD

2

S = reference area, m

vj

=jet velocity, m/s

a

=


p = orifice diameter to internal pipe diameter ratio

V =

freestream velocity, m /s

aC,/aC, = lift augmentation

aC,/at =

rolling moment derivative with respect to aileron deflection

A = wing sw eep angle, deg

p

= density

5= aileron deflection, deg

Subscripts

c /4

= quarter-chord position

D =

drag

= j e t

L

= lift

L = left jet on full-span model

1=

rolling m oment

m = pitching moment

LE

=

leading edge

max

=

maximum

plenum = associated with plenum parameters

R

=

right je t on full-span model

total = combined left and right je ts

trim = trim condition

p =

associated with blowing parameters

6=

aileron deflection

co = freestream

I. Introduction

IRCULA TION control (CC) has been recognized as a technique by which

C ery 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



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 con-

figurations 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 jet s

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 C C 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 T E consisted of a 6-mrg-diam brass rod, giving a T E radius-to-mean-aerody-

namic-chord ratio of 0.005C.

A narrow convergent slot provides the je t blowing. This was achieved by man-

ufacturing a “knife-edge” section that ensured that there was a contraction w ithin

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 pu sh-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 posi-

tioning of the splitter board, an inboard 5-mm section of the slot length was



340

S.P. FRlTH AND N.

J.

WOOD

round transition

\

Pressure

taps

\

a)

,Rectangu lar o I

I

Moun ting Strut

,

\

CCW

trai l ing edge

Fig. 1 Half-span cropped-delta model geometry: a) Upper surface view, b) cross-

sectional view.

permanently sealed off. The aspect ratio of the wing was calculated using

b 2 / 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, M anchester,

UK, as shown in Fig. 3. A splitter board was mounted to ensure that the wind-

tunnel 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


34

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 m aximum jet ve locities were approxi-

mately 180 m/s.

The air supply was sourced from pressurized receiver tanks fed by an A tlas-

Copco 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 m omentum coefficient C , could be calculated using

Vjriz

c

=

s

where vj is the velocity of the Coanda jet,

r z

is the jet mass flow rate, q is the free-

stream 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 o r moment curves.

Particle image velocimetry (PIV) was also performed to obtain more infor-

mation on the interaction of the je t with the freestream flow.5 A horizontal light-

sheet was fired at the TE of the CC W using an Nd:YAG laser. A megapixel CC D

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 agree-

ment 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 carpe t 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 grid-

lines of constant CL sed 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 sim ilar in layout to

the lift carpet plots. The main difference is that the offset for the blowing coeffi-

cients 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.



5

USE


...........................................................

.

Increasing

343

Blowing

............................................................

I

Fig.

5

Example of a p itching mom ent data carpet plot.

reference point (in the case of the half-span model, this is the LE), as shown

in Fig. 5 .

Th e results given in Fig.

6

show the effect of CC on the lift characteristics w ith

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

level of lift augmentation

aC,/aC,

is of the order 18-25 over the complete

range

of C

tested, although a maximum incremental augmentation of approxi-

mately 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

If 0.4

G

J 0.2

4

0,

.-

8

0

0.2

Fig. 6 C L vs C,: effect of slot height on CC.



344

S .

P. FRlTH AND N.

J.

WOOD

smaller slot height giving rise to an increase in the ratio between the m omentum

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

Coa nda surface. Th is 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 C C system ,

as also seen by Jones and E n g la ~ - .~he 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

15 de g revealed that the nosedown p itching moment of the model was less than

that obtained for the model at l0 deg . 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 dra g augmentation is significantly less than the equivalent

value for lift, suggesting an overall increase in

LID.

However, drag measure-

ments are not presented in this chapter because of an inconsistency in the data,

which may be du e 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. Th e data also demonstrate that the downstream exten t of the wake was

reduced. There is also an area of accelerated flow over the upper surface, just

........................................................................

o‘2 T

G O

J

-

E

0” -0.2

a

0,

-0.4

U

C

-0.6

-

-0.8

a -1

E

-1.2

I

3

0

c

C

a

8

p -1.4

.-

r

f -1.6

-1.8

..............................................................................

- 5 d e g r e e s

A

10

degrees

...............................................................

_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

Fig. 7 Variation of pitching mom ent with blowing for half-span model with

0.15-

mm slot height lower blowing coefficients range:

0

C,

0.01).



USE


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 je t exit. Because of restrictions w ith the apparatus it was not possible to

seed the jet and investigate the full interaction with the freestream flow.

111. Full-SpanUAV Configuration

A. Experimental Procedure

A full-span model was designed and constructed at the Goldstein Laboratory,

Manchester, to investigate any interaction of the Coanda jet s and exam ine the

possibility of roll control, as well as lift enhancement. A schematic is shown in

Fig.

9.

The main body w as constructed using modelboard. The model had a L E 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 su rface and brass for the lower surface, incorporated similar T E dimen-

sions 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 300mm 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 accura te control. The fuse-

lage section of the model was manufactured from aluminum sheet, to create a

theoretical aircraft profile and provide protection for the instrumentation (dual-

axis inclinometer and strain-gauge balance) and air supply within the model.

The model was mounted on a sting in the Avro

9 x

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 approxi-

mately 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


347

Fig. 10 Sting-m ounted model in wind tunnel.

B. Results

In quiescent conditions, both C oanda jet s performed a s expected, with the jet s

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 signifi-

cant fluctuations, although all tests w ere 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 e ffectiveness of the full-span model, in the form of

carpet plots with contours of constant C and angles of attack. A lift augmenta-

tion aCL/aC, of 17-24 was achieved, as demonstrated in Fig. 11, in which data

are shown for both Coanda jet s at the same mass flow rate, and therefore the sam e

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 augmenta-

tion is that the drag fo r such a surface is reduced when compared to traditionally

large CC Coanda surfaces.

Table

1

Test proced ure for full-span m odel

Left plenum section Right plenum section

Configuration blowing coefficient,

CI* L)

blowing coefficient, CI* R)

Symmetric

Differential 0

Differential

0.002

Differential

0.004

0- .004

0-0.004

0- .004

0-0.004

0-0.004

0 (constant)

0.002

(constant)

0.004

(constant)



348 S.P. FRlTH AND N.

J.

WOOD

1 ............................................................................

I

-

5 degrees I

Fig. 11 Variation of lift with blowing for full-span model with both system s active

blowing coefficients range: 0

Cp total)

0.008).

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

cente r of lift is located aft of the quarter chord. It is encouraging to see that the C C

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 je t mo mentum

is additive; that is, if the left jet was used at the maximum value of C the acti-

vation of the right jet would result in a similar lift curve to that obtained with

symmetric blowing.

Th e control of rolling moment by C C 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 par-

ticular 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 fro m a sec-

ondary 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 m oment curves.

Th e 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


349

0 5 10

15

0.04

5 -0.02

0

c

m

C

0.04

r

m

C

-0.06

h

-0.08

Fig. 12 Variation of pitching mom ent with blowing for full-span model with both

systems active blowing coefficients range:

0 5

Cp total)

0.008 .

constant, while increasing the je t momentum on the left system through the entire

range possible (0 5 C 5 0.004).Th e effect of the vortex is clearly ev ident, 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.



350

1

0.8

0

0.2

S.P. FRlTH AND N.

J.

WOOD

Fig. 14 Variation of lift with blowing for full-span model 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 de g/s is achieved with a single downw ard aileron deflec-

tion of 10 deg.

............................................................................

......................................

/

I 5 degrees

............................................................................

............................................................................

............................................................................

Fig. 15 Variation of rolling moment with blowing for full-span model with left

system active only blowing coefficients range: 0

5

C a ~ )

0.004).



0.04

0.03

q 0.02

m

-

0'01

s o

:0.01

a 0.02

.-

0.03

-0.04

USE OF CC FLIGHT CONTROL

.............................................................................

.............................................................................

35

................... ............................ -5 m r e e s ..

-A- 10degrees

,=0.004

.............................................................................

Fig. 16 Variation of rolling moment with blowing for full-span model with right

system active only blowing coefficients range:

0 f i ~ ) .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

-

01

0

5

-0.01

a 0.02

0

.-

-

0.03

-0.04

.............................................................................

........................................-

..

-

l c ,(ToTAL) = 0.002

...........................................................................

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

Ca L)

0.004).



352

S.

P. FRlTH AND N.

J.

WOOD

Table

2

Calculated values for ro lling moment

derivatives of full-span model

acIm

acIm

- .084

rad-

-0.052

rad-

possible to equate the equivalent blowing coefficient required to generate the

same rate of roll such that

Th is gives a

C

of approximately 0.0021, equivalent to an aileron deflection of

approximately 10 deg. Th e 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 T E geometry and then on a full-

span 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 .............................................................................

0.03

.............................................................................

I

.............................................................................

:-

0.02

-

.-

g

0.01

.............................................................................

O degrees

8 +-lodegrees

= -0.01 .........

0

0.02

..................

......

.......

................. 4 1

0.03

.

TOTAL =

0,004

0.04 .............................................................................

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

C p ~ )

0.004).



USE


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 ACLx 0.1). As the CC system is considerably less complex mechani-

cally 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 generat-

ing significant rolling m oments a t low blowing coefficients. Importantly, the pro-

duction of roll moments can be superimposed on the lift generation, suggesting

minimized interaction and simple control development.

More detailed work a t even sm aller increments of C especially in the lower

blowing regions, w ill enable greater understanding of the physics involved in

CC

and the areas of higher lift augm entation. 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 acknow ledge 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.

Englar, R. J., and Applegate, C. A., “Circulation Control-A Bibliography

of


DTNSRDC Rept. 84/052, Sept 1984.

3Jones, G.

S.,


Through Pulsed Blowing,” 21 st AIAA Applied A erodynamics 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

Th e Coanda Effect,” AIAA 1st Flow Control Conference, June 2002.


Moment Augm entation and Modification-Past, Present and Future,” AIAA Fluids

2000 Conference and Exhibit, June 2000.

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.

2

7

8“



1I.C.

Experiments and Applications: Nonaerospace



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 $lS O /gall on , 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 econom y 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 direc-

tional), reduction in side-wind sensitivity, reduction in splash and spray, and

improved traction plus aerodynamic braking. One could also include an aerody-

namic m eans to reduce tire rolling resistance. Any such devices being considered

for these applications should also be sim ple and robust, contain few or no m oving

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 aerodynam ics 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 develop-

ment 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

becom e know n as circulation control (CC) aerodynam ics. Here, an airfoil’s con-

ventional 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 ) . ~his entrains the external flowfield to follow the curving jet, and thus

enhances the circulation around the airfoil and the ae rodynam ic forces produced

by it. The governing parameter is not angle of attack, but rather the blowing

mom entum coefficient:

mV,

c,

=qs

where m is the jet mass flow,

vj

the isentropic je t velocity,

S

is a reference wing

area (or frontal area

A

for a ground vehicle), and q is the freestream dynam ic

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 B LOWING

TANGENTIAL BLOWING OVER ROUNDED TRAILINQ EDGE SURFACE

t

BOUNDARY LAYER CONTROL

c p

=

IilvJIqs

MOMENTUM COEFFICIENT, Cp

FORCE BALANCE

J E T

SHEEF

Fig. 1 Basics of circulation control pneumatic aerodynam ics on a simple two-

dimensional 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 en gine thrust expended

for vertical lift (which must ex ceed weight), with very little, if any, augmentation

of aerodynamic lift occurring.

It is because of this high return on blowing, o r conversely, b ecause 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.

The se advantages led to the application of this pneumatic concept to imp rove

the aerodynamics of an already streamlined car model.5 The resulting large jet

turning over the curved rear of this vehicle is show n in Fig.

3.

Significant but dis-

tinctly 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




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

hows 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 M odels

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.



362

R. J.

ENGLAR

installed. It is actually representative of a faired cab-over-engine HV based on a

Penske racing team c ar 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

D

n

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, D s based on projected

frontal area of the vehicle

A,

including the wheels.)

B.

Based on the preceding unblown configurations with reduced drag , additional

wind tunnel tests were conducted to evaluate aerodynamic improvements

Wind-Tunnel Evaluations

of

Blown HV Configurations

1.8

1.7

1.6

1.5

1.4

1.3

1.2

1 1

CD

1

0.9

0.8

0.7

0.6

-15

-10

5 0 5 loNose right 15

Yaw Angle, y~, eg

Fig. 6 Effects of side wind on drag for various unblown HV configurations.




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 nu mber

=

2.5

x

lo 6, based on total tractor

+

railer

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 D = 0.824 at this Reynolds number, the blown configur-

ation reduced the drag coefficient to

0.459

at C,

= 0.065.

This is a 44% D

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 a t 1.0 psig (C,

=

0.13),

D was reduced by

50%.

These data are all for a smoothed bottom tractor-

trailer 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.



364 R. J.

ENGLAR

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 C D 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 jet s, 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

s 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 -

.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 per-

centages will be modified when the power expended to compress the blowing

air is included, but that will have to await a full-scale sys tems 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 increm ents caused by blowing slot configuration.




0 0.1 0.2 0.3 0.4

0.5

0.6 0.7

0.8

0.9 1

Momentum C oefficient, C p

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 effi-

cient 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 m eans of improv-

ing upon the gap problem.

Note that when com paring these data to other experim ents that have been con-

ducted 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 bound-

a r y - l a y e r 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

-

deg produces a



366

R. J.

ENGLAR

0 65

0 60

0

s

0 so

CD

0 45

0

40

0 35

0

30

O W

005

0 1 0 015 020

0 2 5 030

Momeolum

Cafl5eieot.Cp

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

Right 4.0

90°/300 1 / 2 TE ,0.375 R

p 00

O ,

Wheels ON , Left Slot Blowing Only

N = Half Chord Yawing Moment

y O ,

M s e

Straight Ahead

Coefficient

1.5

1 o

0.5

0.0

0.5

-1.0

-1.5

CN

ose Yawed Left

Nose

Lefl

' ' 8 ' ' ' ' ' 8 . .

0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.18 0.20

Momen tum Coemcient, C

Fig. 11 Directional control capability provided by HV configuration with left rear

slot blowing only.




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 oppo-

site yawing moment

C N =

+2.0). With the nose yawed left (for example,

Ic,

= - 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 Econom y Testing

The preceding model tests led to the conclusion that a full-scale proof-of-

concept 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 fabri-

cated the PHV blown test trailer, including blowers, drive motors, control

systems, and instrumentation. This configuration is shown in Fig. 12 in compari-

son to a stock (reference) Great Dane trailer. Blown tufts confirming je t turning of

90

deg around the right-side TE curved pneumatic surface are shown in Fig. 13.

A. TuningTests

Test vehicle fabrication and assembly w ere completed at GTRI and the modi-

fied trailer was then picked up by team member V olvo Trucks of North Am erica

and moved to its facility in Greensboro, NC. Here two initial tuning tests were

conducted (Fig. 14). Figure 15 show s a rear view of the pneumatic trailer

with the tufts confirming on-road flow turning. These tests verified the test equip-

ment 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 con-

ducted 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 PH V tuning test near Volvo facility in North Carolina.




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 TR C/

GTRI/Volvo/Novatek team members in strict accordance with S A E est pro-

cedures (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 TR C test track.



370

R. J.

ENGLAR

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.g91113 Figure 17 shows the pneumatic test tractor-trailer a t 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 econom y improvement, with four trailer slots blowing.




%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 respect-

able, 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 configur-

ations. Whereas the lower blowing values w ere 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 pro-

duced the aerodynamic and fuel consumption differences discussed above.

Figure 20 shows that as the configurations approached the representation of the

24

13

FEI

12

WT

= GTR l Small-scale Wind Tunnel Tests,

TRC= Full-scale Track Test at TRC

-

(from Figure 1)

LO

9

8

7

5

4

3

2

I

0

, R u n 157, 2 Side

Slots

Only

0 0.01

0.02

0.03

0.04

0.05 0.06 0.07 0.08

0.09 0.1

0.11

Blowing Momentum Coefficient, C

@

Fig.

19

Comparison

of

wind-tunnel results to TRC track results.



372

0.85

R. J.

ENGLAR

Comparative CDvs Ck , MTF 065

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 w ind-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 fro m eac h configuration ch ang e, whereas Fig. 2 2 shows

the predicted chan ge in %FEI due to each. Major problem s 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 exp lains the causes of the less-than-expected track-test

results. We have since conducted further testing to improve the final PHV con-

figurations 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 con-

figuration 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.




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

GTR I model drag reductions relative to Run 36 (configuration closest to

baseline

HV)

(positive AC, here is drag reduction).

30

2 5

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.



374 R. J.

ENGLAR

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 evalu-

ation 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 con-

figuration 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,

,

1 5 0 , ~

140.0

130.0 -

120.0 -

Source: Refs. 16 and 17

BGY=Billions of Gallons per

i i n n

A n n

1

20.0

w w

10

~ (Class 2b-8) &

Buses

Repon Date ~ (No

Military Vehicles)

o

0.0

L ,

, , ,

' ' ' '

,

. I

,

' I , d

1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020

Year

Fig.

23

Highway energy usage comparisons (billions

of

gallons per year) by vehicle

type.




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 T E je t 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

Lockheed wind tunnel.

GM Suburban test vehicle undergoing tuft flow visualization in the



376

R. J.

ENGLAR

0.8

0.7

0.6

0.5

0.4

0.3

0.2

CD,

CLfO.1

Cside

0.0

Pneumatic SUV, Lockheed LSWT T1835,10/18/02

Yaw Sweep, q=12.54 psf, V=71.7

mph,

Run 3

I

onventional

GM Suburban

0.25

0.20

0.15

0.10

0.05

-0.00

-0.1

-0.2

-0.05

-0.3

-0.10

-0.4

-0.5

-0.6

-0.7

-0.8

-0.25

-16 -12 -8 -4 0 4 8 12 16

-0.15

-0.20

Nose Right

Yaw

Angle, v, deg

ose Left

Fig. 26 Resulting aero forces and mom ents 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 a ft surfaces and a con -

tracting of the je t wake behind the vehicle. T he wind-on, blowing-on data showed

different behaviors for the different TE configurations. Greater TE turning-

surface 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

by drag increase with higher

C

This drag reduction at lower blowing is of

the ord er 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 f rom drag reduction to braking if requested

by the driver. Note a lso 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.

PSUV Drag Variation with Blowing, V=SOmph

Momentum Coefficient, Cb

Fig. 28 PSUV drag coefficient changes with varying C

0.44

0.46

0.48

0.50

0.52

0.54

0.56

CD

0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08

Momentum Coefficient, Cµ

PSUV Drag Variation with Blowing, V=50mph

Run 12, 45° Top, Bottom & Bottom Sides;90° Top Sides

Run 11, 45° Top & Bottom; 90° Sides

Run 13, 180° All 4 Sides

Run 14, 45° All Sides



378

R. J.

ENGLAR

nonoptimized pneumatic configurations was the same as that of the stock refer-

ence 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 direc-

tionally 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 pro-

duces 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.

PSUV

Aero Data, LSWT Test 1853, Run 17, Left Side Slot ONLY ,

required

to

offset v=-lOo nblown

C required

to

offset

~ = - 5 '

nblown

0 0.01

0.02 0.03 0.04 0.05 0.06

Left Slot Momentum Coefficient, Cmu

PSUV

Aero Data, LSWT Test 1853, Run 17, Left Side Slot ONLY ,

Cmu Sweep, V=50 mph, All Turning Corners are 45

0.09

. , . , . . . , . , . . . ,

Nose 1

Right

0 08

0.07

0.06

0.05

Cyaw

0 04

0.03

0.02

0.01

O

0 0.01

0.02 0.03 0.04 0.05 0.06

Left Slot Momentum Coefficient, Cmu

Fig. 29 Yawing mom ent control by b lowing the left-side slot only.




In a related application, GTRI and Novatek are also currently developing a

patented aerodynamic heat exchanger that is based on these pneumatic prin-

c i p l e ~ . ’ ~ ” ~his can be combined with the above devices to further reduce

vehicle drag by reducing the drag associated with the conventional vertical radia-

tor 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 investi-

gations to confirm the use of pneumatic aerodynamics on these vehicles. We

have verified these capabilities by designing, fabricating, and testing small-

scale PHV m odels in three separate wind-tunnel tests, and by designing, fabricat-

ing, 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

SW . It has thus been verified that these blowing concepts

can reduce aerodynamic drag, favorably modify other aerodynamic character-

istics, and thus improve the performance, stability and con trol, 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 capa-

bilities 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 approxi-

mately 23%, corresponding to about 46%

D

eduction 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 delibe rate 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 w heels 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.




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 Amer-

ican 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 Aug-

mentation and Modification; Past, Present and Future,” AIAA Paper 2000-2541, June

2000.

Englar, R. J., Smith, M. J., Niebur, C. S., and Gregory, S. D., “Development of Pneu-

matic 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.

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.

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 GTR I/DOE Pneumatic Heavy Vehicle Tuning

Tests,” GTRI Rept. A-5871, March 2002.

“Englar, R. J., “Preliminary Results of the Pneumatic Heavy V ehicle 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.

5

9



382

R. J.

ENGLAR

13Englar, R. J., “Developm ent and Evaluation of Pneumatic Aerodynamic D evices 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 V ehicle

Road-Test Configurations,” GTR I Rept. Projects A-587 1 and A-6395, Jan. 2003.

‘’Englar, R. J., “Drag Reduction, Safety Enhancement and Performance Improvem ent

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/OR NL-6958, Sept. 1999.

”“EIA Annual Energy Outlook 2000,” DOE/EIA-0383 (2000), Dec. 1999; also AEO

2001.

“Gaeta, R.

G.

nglar, R. J., and Blaylock, G., “Wind Tunnel Evaluations

of

an Aero-

dynamic 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 H eat

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

P

fficiency. This usually means that aerodynamic drag, both profile drag

and friction drag, are minimized. Design strategies for low profile drag include

*Senior Research Engineer, Aerospace and Acoustics Technologies Branch, ATAS, Georgia Tech

'Principal Research Engineer, Aerospace and Acoustics Technologies Branch, ATAS, Georgia

*Undergraduate Student, School of Aerospace Engineering.


Research Institute. Associate Member AIAA.

Tech Research Institute. Associate Member AIAA.


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 eng ine size and passenger room.

Heat exchangers used in these veh icles 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 a ir through its core to remove engine

heat. However, its location and how the vehicle is shaped around are largely gov-

erned by the engine placement and the vehicle styling. A conventional, state-of-

the-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-perform-

ance 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 oncom ing

flow. A novel approach to this problem is the aerodynamic heat exchanger (AH E)

concep t, 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 differ-

ence 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 i s 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

385

Flat

AP

at

expense of

late

high drag

Wing

P with

low drag

Fig. 2

A

wing has an order of m agnitude 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 A H E

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 G TR I and Novatek, Inc. in

2000

and it involves the

use of a very effective high-lift airfoil section to generate the pressure d ifferences

needed across a conventional automotive radiator.’ As the natural pressure differ-

ence 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 differ-

ence, the greater the flow through the wing.

The A H E 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 . ~his effect postulates that a fluid stays attached to a curved

i t o R e s w e

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.


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 curva-

ture. 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 concep t 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 configur-

ations 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 App roach

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 Veloc i ty

Jet Sheet

Increasing Mom entum Ratio, C,

Increases Velocity on upp er sur face,

thu s increases l i ft [see Coanda Effect,

Circulat ion]

Circulat ion Control produ ces up to

8000%

increase in CL relative

to m omentum force input

Phenomenal

Inc rease in P

Fig.

4

Controllable lift, thus controllable heat transfer, with pneumatic flow control.




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 ellip-

tical, 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 am oun t 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 co-

efficients. 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 radia-

tor 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 flex-

ible hoses to a three-phase electric

3600

W water heater. Water was heated and

pumped into one side of the wing and in to an inlet reservoir attached to the radia-

tor. After the water made it s way through the radiator core, it exited into an ou tlet

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 Out let Reservoir

Fig.

5

Schematic of A H E experim ent showing the coolant path through the test airfoil.



3aa

R. J.

GAETA,

R. J.


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 evalu-

ate 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 conven-

tional Visteon radiator was evaluated without blowing or airfoil frame but perpen-

dicular 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

=

s

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 s LE downward.

B. Test Articles

All radiator test articles were made to fit in the center section of the two-

dimensional 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, o r only 58% flow capacity.

2.

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-

ORNL

Very Dense Graphite Foam Core




389

Fig. 6 Conventional aluminum finned radiator core made by Visteon, showing

coolant passage in lets.

tivity properties. Although it is porous, the bulk density is such that it has a sig-

nificant pressure drop. Brass tubes were press fit into the foam to carry the coolant

through the material for heat exchan ge, as shown in Fig. 7.

3 .

A second

ORNL

supplied radiator core consisted of smaller carbon-graphite

foam fins arranged in such a way that flow could follow the serpentine, a s shown

in Fig. 8. The se were brazed to narrow water channels in a ma nner similar to the

aluminum radiator. The manufacturing of this core was such that some of the

coo lant passages were blocked off,

so

its full heat rejection potential was not rea-

lized. 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.

ORNL

PorouslSerpentine Graphite

Foam

Core

111. Results

A. Note on Measurement Uncertainties

The aerodynamic data presented in this chapter were acquired from a six-

component 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 .5% reading

accuracy. Omega thermocouples were used that were accurate to within a



390

R. J.

GAETA,

R. J.


Fig. 7 Dense ORNL carbon-graphite foam radiator core.

deg ree Fahrenheit. Perhaps the largest uncertainty was in the water mass flow rate

measurement.

An

Om ega flow meter using a turbine wheel was used to obtain the

liquid flow rate. Th e manufacturer specified the m eter 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.



392

R.

J. GAETA,

R.

J. ENGLAR, AND G. BLAYLOCK

MTF059 Pressure Increment Across Airfoil Radiator,

q=5 psf, a=O

Conventional Rad iator 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 g reater LE radius) version of the current baseline blown ellipse airfoil,

and it show s no sign of separation, reaching a C of or more. Thus great down

force potential is confirmed with blowing (no increase in airfoil angle of attack is

necessary) and this w ill 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




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 0.20

0.30 0.40

0.50

CD

Fig. 11 Drag polar as a function of blowing momentum for various AHE

configurations.

a 6 4 mph freestream velocity. Figure 12 show s coo lant 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 fab-

rication anom alies, 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 con-

figuration 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



394

R. J.

GAETA,

R. J.


0.0

-1

.o

-2.0

-3.0

4.0

=

-5.0

6

-

6.0

-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 differen ce; Visteon

core, V , = 64 mph.

removal rate via increasing air flow through the w ing. A notable exception is the

high-density carbon-graphite configuration.

Figure 14shows the heat removal for the various A HE 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 i s 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 low er; see Fig. 7). This result was intriguing and suggests that

the heat removal can be varied and/or augmented by simply varying the turbu-

lence 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 15

20 kW in normal operation for a full-sized engine. The model AH E tested here

produced roughly half of this heat rejection, but with a radiator core of less




395

1

9.0

8.0

- 7.0

6.0

8

5.0

m

-

4.0

3.0

t

8

2.0

1.o

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 A HE does the heat transfer jo b 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.


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.




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 AH E has great potential for exhibiting both controllable aerodynamic

force and low drag penalty for eng ine cooling.

5 Carbon-graphite foam enables optimal performance of the radiator core

within the AH E concept.

It is important to note that system integration issues will pose a (surmoun-

table) 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 recog-

nized 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 Jam es Klett and April McM illan 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 Augm ent Modify

Aerodynamic Forces,” U.S. Patent No. 6,179,077, Aug.

2000.

3Metral, A. R., “O n the Phenomenon of Fluid Veins and The ir Application, the Coanda

Effect”,

F

Translation, F-TS-786-RE, 1939.

4Cheeseman,

I.

C., and Seed, A. R., “The Application of Circulation Control by

Blowing to H elicopter Rotors,”

Journal ofthe Royal Aeronautical Society,

Vol. 71 , July

1966.

’Williams, R. M., and How e, H. J., “Two-Dimensional Subsonic Wind Tunnel Tests on

a

20

hick,

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 Aero-

space Engineering, U niv. of Maryland, College Park, MD , June 1973.

8Wilkerson, J. B., Barnes, D. R., and Bill, R. A., “The Circulation Control Rotor Flight

Demonstrator Test P rogram,” American Helicopter Society Paper AH S-795 1, May 1979.



398

R. J.

GAETA,

R. J.


’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

=

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

= spanwise location, m

a

=


*Professor, Department of Aerospace and Mechanical Engineering. Member A I M .

‘Research Associate, Department of Aerospace and Mechanical Engineering. Member AIAA.

Copyright 005 by

the

authors. Published by the American Institute of Aeronautics and Astro-

nautics, Inc., with permission.

40



402

H F. FASEL, A. GROSS, AND S.WERNZ

r

=

circulation, m2/s

y

= ratio of specific heats

6* = displacement thickness, m

8

=

mom entum thickness, m

h = wavelength, m

p =

molecular viscosity, k3/(m

s)

v

=

kinematic viscosit m

s

v

= eddy viscosity, m / s

p

= density, kg/m3

6 = streamwise (azimuthal) angle, deg

w =

vorticity, l/s

Y, /

Subscripts

in

=

inflow inside plenum

jet = nozzle exit

max = wall-normal maximum

wall

=

wall, surface

= span wise direction

6 = streamwise direction

03 = free stream

Superscript

=

wall coordinates

I. Introduction

AL L jets over curved surfaces have great potential for technical appli-

cations. Coanda wall jets over convex surfaces can effectively provide

aerodynamic side forces or change the circulation of an airfoil. An existing appli-

cation 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 Co anda flow experim ents are presented that may

help to shed som e light on the relevant physical mechanisms.

On e of the most intriguing phen omen a of the C oanda wall jet is the com pe-

tition/interaction of naturally occurring streamwise and spanwise vortical struc-

tures, 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. T he amp litud es and wavelengths of

the coherent structures will also determine the intensity and the frequency spec-

trum of the associated aerodynamic/hydrodynamic noise. In addition, because

both the streamwise and the spanwise structures are a consequence of hydrodyn-

amic instabilities, instability mechanisms may be exploited advantageously for

active flow control (AF C) strategies.

Tan i was among the first to report on streamwise vortices in a turbulent bound-

ary layer along a concav e wall.’ In his experim ents he observed regularly space d

spanwise modulations of the velocity profiles, which he attributed to a Gortler

instability mechanism. T o com pare with stability theory results for a laminar

boundary layer, he assum ed a constant eddy viscosity v r

=

0 . 0 1 8 ~ ~*,and a dis-

place ment thickness, S*= 1.38( 8 s the momentum thickness). Moser and Mo in2

performed direct numerical simulations (DNS) of a curved turbulent channel flow

to determine the effects of curvatu re 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. Suffi-

ciently 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 som e cases, twice as large as for a plane wall.

In the Reynolds-averaged Navier-Stokes (RA NS ) calculations considered in

this chapter , the prediction of the spreading rate dep ends on the turbulence m odel

employed. Pajayakrit and Kind3 used the Baldwin-Lomax, the Dash et al.

K - E ,

the Wilcox K- -w , and the Wilcox multiscale turbulence mo dels 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 cu rved w all je t the nondimensional vel-

ocity 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 co-

workers4 a turbulent wall je t on a circular cylinder was investigated. For this con-

figuration extensive numerical simulations, including DNS, large eddy

simulation (LES), and unstead Reynolds-averaged Navier-Stokes (URA NS)

calculations were conducted?” Th e flow parameters in the simulations were

chosen to match the ex eriment, with cylinder diameter

d

= 0.2032 m, nozzle

height

=

2.34

x

10- m, and jet-ex it velocity vjet

=

48 m/ s. Th e Reynolds

number based on jet-exit velocity and cylinder diameter was

Re

= 6.15

x

lo5

Reb= 7,080 based on jet-ex it 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 co mp uted using steady RAN S. Th is flow configuration was posed as a bench-

mark problem to the CFD community for the 2004 NASA/ONR Circulation

Control Workshop.’ Experiments by Abramson* on this particular airfoil

B



404


Table 1 Elliptic airfoil

CFD

challenge cases

Case

283

Case

321

a, eg 0 - 8

h j e t , kg/ms 0.196 0.182

CLL

0.209 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 . 0 0 3 ~ .

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 softw are Fluent on computational grids with approxi-

mately 1.6

x

lo5points. Computations with the standard and the realizable K - - E

turbulence model only yielded realistic results for the jet momentum coefficient

c p

=

0.026. The jet mom entum coefficient was defined as

with jet-exit velocity vjet, jet-mass flux hjet= pjetvjetb, nd freestream dynamic

pressure 1/2p,vk. For a higher mom entum coefficient, c p

=

0.093, the same

two turbulence models predicted the wall-jet separation slightly farther down-

stream than observed in the experiment. At the even higher momentum coeffi-

cient, c p

= 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 circula-

tion 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 dis-

tribution (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 com -

bined, they will help in understanding the different physical mechanisms involved.

A. Direct Num erical Simulations D NS)

was adopted to allow for

highly accurate DNS of turbulent Coanda wall jets for Reynolds numbers in the

An existing incompressible Navier-Stokes




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 streamw ise 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 pseudo-

spectral 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 high-

order accurate upwind schemes was developed to allow for robust com putations

of complex geometries. The convective terms are discretized with fifth-order

upwind schemes based on a weighted essentially nonoscillatory (WENO) extra-

polation and the Roe schem e,14 and the v iscous 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 deve lopm ent of coherent structures in the tur-

bulent flow upstream of separation and the impact of forcing on these structures

and on the mean flow development.

An illustration of the computational dom ain 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 term s added to the right-hand side of the

Navier-Stokes eq ua tio ns 5 For actively forcing the wall je t 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 tur-

bulent fluctuations from the 0u tfl0w .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


x [mml

Fig. 1 Com putational grid used for DNS.

the radial direction,

193

points are clustered toward the surface such that, through-

out 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

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 je t (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 spanw ise coheren t

structures that result from the inflectional instability.

T o probe this conjecture, DNS were performed, w here 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 w ith a fixed spanwise wavelength

are generated by steady volume forcing with a periodic modulation in the span-

wise direction A, = 20 mm ).

The downstream development of the streamwise structures for the three simu-

lation cases is visualized with the iso-surface plots in Fig. 2 of the time-averaged




407

b) C)

Fig.

2

DNS of tur bu len t Coanda wall je t, 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 two-

dimensional 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

wq

=

00/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 struc-

tures 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 turb ulent Coand a w all je t showing instantaneous spanwise vorticity,

spanwise average: a) “Unforced” reference; b) harmonic two-dimensional forcing;

c) steady three-dimensional forcing.



408


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 othe r 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).

Th e time-development of the spanw ise coherent structures ca n be visualized

nicely with their footprint on the wall, namely, fluctuations in the spanwise

wall vorticity qwall.how n in Fig. 4 for the three cases are time-space diagrams

of the spanwise-averaged w ~ , ~ ~ ~ ~lotted 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 fre-

quent 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 (com pare Figs. 4a and 4c). T his suggests

that the presence of streamwise vortices indeed inhibits the deve lopm ent of span-

wise coherent structures.

Fig. 4 DNS of turbulent Coanda wall jet showing time-space diagram s for

spanw ise-averaged wall-vorticity: a) “Unforced” reference: b) harmonic two-

dimensional forcing; c) steady three-dimensionalforcing.




409

b)

Fig. 5 DNS of turbulent Coanda wall jet showing effect of forcing on the mean-flow

development: a) Inverse-square of stream wise mean-velocity maximum ; and b) wall-

jet 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 = 35 deg with the unforced case.

The results from the DNS also showed that a strengthening or weakening of

the streamwise o r spanwise structures changes the downstream development of

the Coanda wall jet. For example, both the decay of the streamwise mean vel-

ocity and the radial spreading of the jet in the downstream direction are signifi-

cantly 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. How ever, most of the interacting mech-

anisms between spanwise and streamwise vortical structures have to be investi-

gated 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 Coan da flow

calculations was scrutinized in two-dimensional RANS calculations of the

Coanda w all jet experiment by W ygnanski 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

a)


b) C)

Fig. 6 Computational grid used for two-dim ensional 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 eddy-

viscosity 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 je t wrapped around the cylinder more than once.

For some of these turbulence models the jet-velocity decay and jet-half-

thickness 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

norm alized 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.



a)


411

b)

Fig.

8

Two-dimensional RANS computations

of

Coanda flow: a) Jet-velocity decay;

b)

jet half-thickness vs stream wise 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 experi-

mental velocity profiles. With the 1988

K- -W

model (with Boussinesq or EASM

Reynolds stress), very good predictions of the wall-pressure distribution were poss-

ible (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 drift-

ing separation location had reached

6

=

360 deg.

For all but the 1988 K - -W model with EA SM, jet spreading and velocity decay

were underpredicted. Based on the

DNS

results one may assume that the turbu-

lence 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 exper-

imental result when the standard 1988 K- -W turbulence model was used, this

model was then chosen for subsequent three dimensional RANS stability inves-

tigat io m 6 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 span-

wise 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 Fourier-

mode 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.



41

2


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) disturb-

ance 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 4 3

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)

1o

6 o

1o

10-10

0 50 100 150 200

6

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.



41 4

a) 10 l

10 1

10 l

10 2

L

0


no

phase shift

d2 phase shift

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 1

A; and

b) mode amplitudes.

circulation for 6

>

150deg 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 C oanda wall jet, the non linear resonance m echanisms 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.




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 conse-

quence 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 free-

stream molecular viscosity p = 1.790

x

kg/ms. Assuming a gas constant

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

.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. How ever, this

i s u l t s in a nozzle-exit Mach number .7 and requires the use of a com-

pressible code. The nozzle-inflow area

is

Ai,/c =

0.03188. The nozzle-area

ratio is 10.2.

B.

Com putational Grid

The com putational 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) close-

up of airfoil and block boundaries; and c) close-up of Coanda flow region.



416


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 sm aller 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

the mass flux hi

= hjet pinvinAin

and the total temperature (the total temp-

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

1 2

T i n -vin

YR

m+-v; =

- 1

2

y - 1 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). Show n 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

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).




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

norm alized 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

norm alized by laminar viscosity left) and turbulence kinetic energy right).



418

a)


b)

Fig. 17 RANS calculation of CC airfoil showing the wall-pressure coefficient

cp

=

201 pm)/ pjetvfet):) 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 calculationof 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 numeri-

cally:

1) The circular cylinder from the experim ents by Wygnanski and coworkers;

2 ) the NCCR 1510-7067 N C C airfoil from the expe rime nts by Abramson

and

(the workshop C FD challenge).




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 coher-

ence, but did not noticeably weaken the streamwise structures. The reason for

this is unclear and necessitates further research. Forcing of the streamwise struc-

tures weakened the spanwise structures and strengthened the streamwise structures,

as expected. The downstream development and interaction of both types of struc-

tures 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

je t spreading and velocity decay by application of AFC at the nozzle exit.

Configuration

1

was also used to evaluate turbulence models for steady RA NS

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 RA NS 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 half-

thickness, 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 ou r 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


’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 M odels,”

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 NA SAIO NR Circulation

Control Workshop, NA SA /CP 2005-213509, June 2005.

‘Abramson, J., “Two-Dimensional Subsonic Wind Tunnel Evaluation of Tw o 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 S imu lation

of Circulation Contro l 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 Bound-

ary 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,

‘‘Gross, A., and Fasel, H., “High-Order W E N 0 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,

17Spalart, P. R., and A llmaras,

S.

R., “A One-Equation Turbulence Model for Aerody-

namic 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,

‘’Fasel, H. F., von Terzi, D. A., and Sandberg , R. D., “A Methodology fo r Simulating

Compressible Turbulent Flows,” FEDSM 2003-45334, 4th ASMEIJSME Joint Fluids

Engineering Conference, July 2003; also ASME Journal

of

Applied Mechanics (to be

published).

V O ~ . 57, NO. 1, 2000, pp. 371 -403.

pp. 904-910.

pp. 933-942.



Chapter

16

RANS and Detached-Eddy Simulation of the

NCCR Airfoil

Eric

G.

Paterson* and W arren J. Bakert

Pennsylvania S tate University, University Park, Pennsylvania

Nomenclature

a

= speed of sound, ft/s

C D

=

section dra g coefficient,

F ~ / ( 1 / 2 ) p U i S

C L= section lift coefficient, F ~ / ( 1 / 2 ) p U i S

C

=

section moment coefficient, M z / (1 / 2 ) p U i S c

C,

= pressure coefficient,

( p

p o o ) / ( 1 / 2 ) p ~ L

C ,

=jet momentum coefficient, r i z ~ j / ( l / 2 ) p ~ L ~

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.

c

=

foil chord length, in.

M

=

Mac h number, U / a

M ,

=

moment about the z-axis, ftelbf

m = mass flow rate, pUjhw, lbm/s

p

= pressure, lbf/ft2

*Senior Research A ssociate, Applied Research Laboratory and Associate Professor of Mechanical

'Graduate Research Assistant, Department of Aerospace Engineering. Member AIAA.

Copyright 005 by Eric G . Paterson and Warren J. Baker. Published by the American Institute

and Nuclear Engineering. AIAA member.

of Aeronautics and Astronautics, Inc., with permission.

42



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,

m

t /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

p

=

density, lbm/ft3

u= DES blending function or cavitation number

T~ = wall-shear stress, lbf/ft2

w

= turbulent dissipation rate, ft2/s3

8 7 = curvilinear coordinates

Subscripts

00 =

freestream

min

=

minimum

=

at je t orifice

Superscripts

r

=

resolved turbulence

s =

subgrid turbulence

tot

=

total

I. Introduction

IRCULA TION control (CC ) for lift augmentation via the Coanda effect has

C een studied for many years.192 n 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



DETACHED-EDDY SIMULATION OF NCCR AIRFOIL

423

suffer from orifice fouling or shock damage, and, for applications w here stealth is

important, have limited impact on the hydroacoustic ~ignature.~he 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 w ell as other forms of flow control. Of particu-

lar interest to CC are high-performance solenoid

valve^,^ which can achieve effi-

cient pulsed blowing, a mode of CC that has been known to reduce mass-flow

requirem ents for a given performance increment. -12 However, detailed under-

standing 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 compu-

tational fluid dynamics (CFD) models must be able to simulate if such tools are

to be used in design. M ost notable are streamw ise curvature effects on the turbu-

lent boundary layer and spanwise coherence of the wall jet . Nearly the en tire range

of Reynolds-averaged N avier- Stokes (RA NS) turbulence models from algebraic

to full Reynolds-stress transport models (RSTMs) have been modified for curva-

ture effects.13-15 Unfortunately, the s tate of affairs is poor in that modifications to

algebraic and one- and two-equation models are limited in range due to empiri-

cism, whereas RSTMs have yet to convincingly demonstrate capability to

resolve subtleties in the way curvature impacts mean flow and turbulence

s t r u c t ~ r e . ' ~onetheless, numerical experim ents for a C C configuration16 have

demonstrated that baseline RSTMs can improve simulation results in comparison

with baseline two-equation m odels, especially at large je t momentum coefficients.

Moreover, this study showed that simulations using two-equation models dem on-

strated 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 je t 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 near-

wall region is treated as RAN S and the ou ter 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 surface21 and wind-tunnel data for a pulsed C C 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 sim pler 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 RAN S CFD.16

Our progress is reported in Secs. 11-VII.

11.

Geom etry, Conditions, and Data

The NCCR-1510-7067N CC foil was tested in a wind tunnel at the David

Taylor Naval Ship Research and Developm ent Center in 1977.22The geometry

was a 15% thick elliptical cambered foil with a s ingle jet orifice on the upper

surface at

x/c =

0.967. The model chord length was

c =

8 in., the slot height-

to-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 = deg. Both are assumed to have the follow ing comm on param eters: free-

stream velocity U , = 128.54 fps, freestream density

p,

= 0.07654 lbm/ft3, and

kinematic viscosity

=

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., p j / pm

=

l), the nondimensional jet-orifice velocities

can be computed as

v j / U m

= - 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 sm all.

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 m omentum-

deficit method. In addition, estimates of experim ental uncertainty are not avail-

able; however, several possible sources have been identified such as slot-height

TRAILING

EDGE

Fig.

1 Cross-sectional geometry of NCCR 1510-7067N.




425

growth because of plenum pressure, and Coanda je t 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

CFD SHIP-IOWA23 is a general-purpose parallel unsteady incompressible

RANS CFD code. The computational approach is based upon the pressure-

implicit 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 dis-

cretized using a general five-point stencil that permits a user-specified order-

of-accuracy ranging from first-order upwind to fourth-order central. Viscous

and temporal terms are discretized using second-order central and second-

order 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 com-

puting. For time-accurate unsteady simulations, global solution of the pressure-

Poisson 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

CDEs

s 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:

3)

=

max ( , a

8,)

The new length scale

t

replaces

t k w

in the destruction term of the k-transport

equation

pk3I2

Dk,,,

=

pp*kw

=

e k w

4)




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.

Th e second m odification a ims to redu ce 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 s 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 im pose d, which requires that the time step be sufficiently small to

support turbulent eddies. The coefficients n and 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 equ ations



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

) w E

+

ow:

w,= (1 ) w Z

+

mv?

w,

(1 ) w F +mv?

w,=

(1

)wtlf

+

(13)

w,, =

(1

)wtlf,

+ow:

Finally, as discussed in the following section, it is noted that CFDSHIP-IOWA

is an overset-grid capable CFD code w ith 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 coordina te direc-

tion) 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

5 x/c 5

4,

5 y / c 5

2, and

0

I

/ c

I

.2, and a near-wall spac ing of 1.0

x

the latter of which aims to

resolve the sublayer of the turbulent boundary layer with a wall spacing of

The grid system used for

RANS

simulations is shown in Fig. 2. Nested orthog-

onal 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-refine-

ment blocks were built off these streamlines for subsequent higher-fidelity simu-

lations. 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 span-

wise 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

y + = 1.



428


J.

BAKER

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

=

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



430


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 com puted 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 com-

puted using Pegasus 5.1

. ’

Mathem atical formulation of all boundary conditions

are described in the C FDSH IP-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 Po intwise, 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.

A. Steady RANS Simulation

A comparison of experimental and simulated surface pressure is shown in

Fig. 4. Relatively good agreement is dem onstrated for both cases. The largest dis-

crepancy 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 m ajority of

the chord, and a Cp minf 7 and 8 at the TE for the simulation and exper-

iment, 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,min

re 3 and 18, respectively.

Lift, drag, and moment about the z-axis centered at midchord were com puted

by integrating C and T~ on all external surfaces. All plenum surfaces were neg-

lected in the C FD integration process so that comparison could be m ade to exper-

imental va lues. Experimental values were computedz2 by 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 com ponents) are positive

and substantially larger in magnitude. Moment coefficient C M is positive (LE

down, TE up) for both cases because of the large suction peak on the Coanda

surface. Data for

CM

are not available.




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 illustrates the impact of the Co anda effect upon the overall circula-

tion. 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 cha nge in angle of a ttack on the overall flow-

field and on the locations

of

stagnation points.

Table 1 Lift, drag, and moment coefficients

CL

D

C M

Data CFD Data CFD Data CFD

Case

283 4.2 4.0 .05

0.18 2.07

Case

321 3.1 2.4 .08 0.12 1.21

CFD

omputational fluid dynamics.




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 large r in magnitude, starts at the point of wall je t sep-

aration a nd grow s into the wake. It is noted that the max imu m

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 magni-

tude and turbulent kinetic energy a re extracte d at two locations for Case

283,

as

shown in Fig. 9. Location A is slightly aft of the je t orifice, and location B is along

a

y

=

0

line. At location A, the wall je t 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



434

a)


J.

BAKER

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.,

,

0.002) in the near

orifice region. Therefore, target grid spac ing in this area should be approximately

=

0.001, which i s 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 iso-

surface of vorticity shaded by spanwise velocity were made and snapshots are




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 ~ ownstream 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

u,

treamlines through the mean field, and RMS axial velocity fi. he

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 corre-

sponding to vortices shed off the top and bottom sides of the foil. It is noted



436

a)

0.014

0.012

.

0 01

0

0.006

0.006

0.004

0.002

00

E.

C.

PATERSON AND W.

J.

BAKER

3

5

6

Vel ity mag nit ude (U2+4/2)1’2

6

Fig. 9 Extracted profiles: a) Velocity m agnitude; 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 com puted statistics.

Analysis

of

the turbulent kinetic energy is shown in Fig.

13.

Subgrid turbu-

lence

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

s significant only in the boundary layer upstream

of

the sep-

aration. 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 nd 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




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.



438

a)


J.

BAKER

Fig. 12 Statistical analysis of axial velocity: a) Mean velocity; b) root-mean-square

velocity.

slight decrease in magnitude as the T E 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 = c / U , = 3.8.

If a new length scale is defined as the vertical distance between points of mean sep-

aration at the TE, which is

d / c = 0.052,

a more appropriate shedding frequency is

computed to be 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 /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.




439

a)

Fig. 13 Analysis

of

turbulent kinetic energy: a) Subgrid, k S ; b) resolved, k ;

c) total,

k S

+

k .



440


J.

BAKER

0.75

0.5

0.25

time

(UVC)

Powerspectral density

of

axial velocity

b)

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 fo r ship

hydrodynamics. As a rough estimate, cavitation occurs when the magnitude of

minimum pressure coefficient exceeds the cavitation number:

-cp 2

(T




Given that p m = pgz, u increases linearly with depth. Substituting p m into

Eq.

( 1 3 ,

an expression for cavitation-free operation relating

Cp,,,in,

depth

z

and vehicle speed U , can be derived:

3

Using properties of water at

15°C

p

=

1000

kg/m ,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 cavita-

tion-free operation, at least for the C studied herein. Fortunately, because

dynamic pressure increases with Urn, ower C and

CL,

and therefore decreased

Cp,min, ould 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

10 20 40

50

60

Fig. 15 Cavitation-free operation curves.




J.

BAKER

contrasts w ith 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 pre-

sented through analysis of the integral forces and mom ent, velocity field, and tur-

bulent kinetic energy.

Detached-eddy simulation was undertaken for the unblown case, and demon-

strated that the m ethod is capable of resolving turbulent vortex shedding. Statisti-

cal 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 appli-

cation of DES to both C C 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

cases will permit study of three-dimensional effects and pulsed blowing, both

of which are important issues for practical application and improved understand-

ing of basic CC flow physics.

CC conf ig~ra t ion .~’In addition to providing high-fidelity flowfield data, these

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 Cen ter 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:

Aerodynam ic 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.




443

6Bushnell, D., “Application Frontiers of ‘Designer Fluid Mechanics’ Visions versus

Reality or An A ttempt 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. nd Kegerise, M., “Overview of Active Flow

Control Actuator Development at NASA Langley Research Center,” AIAA Paper 2002-

3159, June 2002.

’Jones, G. iken, 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.

nd 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 Turbulen t Boundary

Layers,” Progress in Aerospace Science, Vol. 33, 1997, pp. 1-70.

”Gatski, T., and Speziale, C., “On Explicit Algebraic Stress Models for Com plex Tur-

bulent Flows,” Journal

of

Fluid Mechanics, Vol. 254, 1993, pp. 59-78.

‘6Slomski, J., Gorski, J., Miller, R., and Marino, T., “Num erical S imulation of C ircula-

tion Control Airfoils as Affected by Different Turbulence Models,” AIAA Paper 2002-

0851, Jan. 2002.

”Strelets, M., “Detached-Eddy Sim ulation of Massively Separated Flows,” AIAA Paper

2001-0879, Jan. 2001.

“Squires, K., Forsythe, J., Morton,

S.,

Strang, W., Wurtzler, K., Tom aro, 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 Spala rt, 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 C ontrol

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 Contro l

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.

24Am es, I. A., and Menter, F., “Two-Equation Eddy Viscosity Turbulence Models for

Engineering Applications,” AIAA Journal, Vol. 32, No.

8,

1994, pp. 1598- 1605.




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, B irkhauser

Press, Cambridge, MA, 1997, pp. 163-202.

2 7 S ~ h s, . 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 Trailing-

Edges of a Flat S trut,” DTNSRDC Rept. 4241, Dec. 1975.

29Paterson, E. G. nd 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 M ar h o, +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

ri =

mass flow rate, kg/s

S

=

span, m

Re =

Reynolds number based on

U,,

c and v,

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

=

turbulence dissipation rate, m2 /s3

r )

=

distance from wall, m

w

= specific d issipation rate, 1/s

*Propulsion and Fluid Systems Department. Member AIAA.

'Propulsion and Fluid Systems D epartment.

*Marine and Aviation D epartment (retired). Mem ber A IAA.


of

the

U.S.



445



446

P.

A.

CHANG

ET

AL.

p..

= freestream fluid density, kg /m 3

T~ = wall shear stress, kg/(m . s2)

v = free stream kinematic viscosity, m 2/ s

VT

=

turbulence viscosity, m2/s

overbar) time average

I. Introduction

ECENTLY , low-speed maneuverability has become an important design

R quirement 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 (C C) air-

foils 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 y ~ , ~ver the past 25 years. True CC airfoils typically have bluff

trailing edges. These airfoils employ the Coanda effect to obtain lift augmenta-

tion by tangentially ejecting (blowing) a sheet of fluid near the trailing edge

(TE) on the upper surface. Because of the Coanda effect, the je t 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 com pared to the case of no blowing.

Because of the difficulty and expense involved in experim entally 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 momen-

tum 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 T E 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 sur-

rounding 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 o ne equation’ eddy-viscosity turbulence models. However,



FULL-REYNOLDS STRESS MODELING 447

the CC for these airfoils is essentially a blown flap method, where the jet separ-

ates 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 circula-

tion 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 enera lized for arbitrary flows and configur-

at ion^ ^ For exam ple, Slomski et al. demonstrate that standard isotropic, two-

equation turbulence models yield nonphysical solutions for a

CC

airfoil as

blowing rate increases, whereas a full Reynolds-stress turbulence model repro-

duces the correct lift/blowin rate behavior for the same airfoil. Recently,

however, Paterson and Baker 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.

1

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 resu lts reported in Paterson and Baker,” the perform-

ance 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 pressure-

velocity coupling, is utilized. Second-order upwinding is used to discretize the

convective terms in the momentum equations with second-order central differen-

cing 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 (LR R),” as well as the blended k - w / k - e

SST model. In two d imensions, the FRSM introduces an additional five equations

-three equations for each of the correlations U U U V and VV, and equations for

k

and 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 stres-

ses parallel to the wall.

Enhanced wall treatment is utilized, which solves to the wall where y 3

and uses w all 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 tw o Coanda T E shapes5 are undertaken. B oth the “nominal” circu-

lar T E foil, NCCR 1510-7067N, shown in Fig. 2, and the logarithmic spiral TE

foil, NCCR 15 10-70678, are used. The slot-height-to-chord

( h / c )

atios include

0.0015, 0.0022, and 0.0030. Incidence ang les are 0, -4, and 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 7067 airfoil.



FULL-REYNOLDS STRESS MODELING

449

I I I

0.92

0.94 0.96 0.98

1 00

X/C

Fig. 3 Comparison of circular and logarithmic-spiral TE geometry; -, circular

TE;

- -

-, ogarithm ic-spiral TE.

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

=

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

=

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 se t 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 lOP 5m 2/s and density,

p,

=

1.224 kg/m3. The chord length of the airfoil, c is 0.203 m, giving a free-

stream 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 m ass flow rate is nondimensio-

nalized as the jet mom entum coefficient

my

1/2p,u2,c

c -

Fig.

5

Grid for

h / c

= 0.0030 airfoil showing detail of C oanda jet region.




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 cali-

brated venturi m eter that was inserted in the air supply line and the je t velocity

vj

was calculated as an isentropic expansion from duct pressure to freestream static

p r e ~ s u r e . ~

Table

1

lists the cases for the circular arc TE with case numbers corresponding

to those given in Ab ra m ~ o n .~able 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



452

P.

A.

CHANG

ET

AL.

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 m ultistep 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 -

turbu-

lence model,

URFs

less than 0.1, run for about 10,000 iterations.

4) Final jet development: Com pressible 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 m odel this procedure is m odified:

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

turbu-

lence model,

URFs

less than 0.1, run for about 5000 iterations.

3) Turn on k-w/k--E SST model: Com pressible 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.




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 com puted by

where Fy is the force in the y direction, pmand

Urn

are 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 cal-

culations, the forces are given in terms of force/unit span). The presure coeffi-

cient

C

is computed by

P

c -

1/2pwU&

(3)

A. C, Variation with h / c = 0.0030, = 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-Attack Variation with h / c = 0.0030

Figures 10 and 11 show the streamlines for = -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 C oanda 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. or = 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.



454

P.

A.

CHANG

ET

AL.

Fig.

7

Streamlines for

h / c

=

0.0030

and =

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,

=

0

deg, comparing FRSM results to

experim ental resu lts: xperiment;

0

FRSM.




b)

455

Fig. 9 Surface pressure distributions for h l c = 0.0030 at =

0

deg:

-,

FRSM;

0 xperiment-top; +, experiment-bottom. a) C =

0.050,

b) C = 0.093, c)

C = 0.209.

How ever, because of the additional circulation caused by the Coanda jet , nega-

tive angles of attack can still have positive lift. There is a constant difference

between the curves of constant 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 deg, respectively. The results are consistent

with the streamline plots (Figs. 1 0 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 je t detaches relatively

earlier with respect to the experiment, but that as C increases, the je t detachment

points get closer to their experimental values.



456

P.

A.

CHANG

ET

AL.

Fig. 10 Streamlines for h / c = 0.0030 and cu = -4 deg: Upper;

C,

= 0.048; middle,

C,

=

0.093; bottom, C, = 0.189.

Fig.

11 Streamlines for h / c = 0.0030 and cu = -8 deg: Upper: C, = 0.047; middle,

C,

= 0.090; bottom, C = 0.184.




457

5

4

3

0

2

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

FRSM results to experimental results. FRSM: -, cu =

0

deg; ---, u

=

-4 deg;

- - - 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 he ights, 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 vel-

ocities 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

C L

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 com-

parison to experimental data, with only a very small underprediction of the

peak TE pressure.

These trends in the p ressure distributions indicate that for the circular TE , for

low values of I he jet detaches early com pared with experiments, resulting in

low midchord pressure differentials and lift. However, as vj increases, the jet

detachment point extends further around, eventually matching up with the exper-

imental location. In these cases, the midchord pressure differential and lift are

well predicted.



458

a)

P.

A.

CHANG

ET

AL.

b)

Fig. 13 Surface pressure distributions for

h / c

= 0.0030 at

= -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 ampli-

tudes 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 - w l k - 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




b)

459

Fig. 14 Surface pressure distributions for h / c = 0.0030 at = -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 je t detachment at the TE ,

rather than the T E pressure drawdown effect typical for other isotropic models.'

In this case, however, the

k-w/k--E

SST results predict low er 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 c loser to the exper-

imental results. This generates a midchord pressure distribution much closer

to the experimental values, although the peak pressure at the TE is slightly



460

P.

A.

CHANG

ET

AL.

5 1

4

3

0

2

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; _ _ _ _

, / c

= 0.0015. Experiment5 symbols: 0 / 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 turbu lence models are best for the CC

foil problem. The results presented in this paper have not been shown to be grid

independent, for example. H owever, the following trends are evident:

1)

Isotropic turbulence models. The M enter 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 logarithmic-

spiral 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




b)

461

XIC

Fig. 16 Surface pressure distributions for

h / c =

0.0022 at = 0 deg,

-,

FRSM,

0

xperiment: a)

C

=

0.053, b)

C

=

0.103, c)

C

=

0.198.

Fig. 17 Surface pressure distributions for h / c = 0.0015 at

=

0

deg for

C

= 0.052;

-,

FRSM,

0

xperiment.



462

P.

A.

CHANG

ET

AL.

b)

Fig. 18 Streamlines

for

logarithmic-spiralTE: a)

C,

= 0.041, b)

C,

=

0.080

and c)

C,

=

0.105.

a)

Fig. 19 Surface pressure distributions for logarithmic spiral cases: -, FRSM-

h / 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.




-20

463

-1 5

-5

0

0.0 0.2

0.4

0.6 0.8 1.0

xlc

Fig. 20 Surface pressure distributions comparing turbulence models for circular

TE Case 283 h / c

=

0.0030, = 0 deg, C, = 0.209); -, FRSM: A

k -m

SST; 0

experiment.

Fig. 21

Lift coefficient vs C at =

0

deg,

h / c

= 0.0030, comparing FRSM, k -m ,

and experimental results: -, experiment5;0 RSM;

A

k-w SST.



464

P.

A.

CHANG

ET

AL.

20

-1 5

-1 0

0

-5

0

0.0 0.2

0.4 0.6 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 xperiment-h/

c

=

0.0015

(Case

51);

0

experiment-h/c

=

0.0022 (Case 356).

Fig. 23 Stream lines for logarithm ic-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

k-

W / 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 m odels appear to be better-suited for application to general

CC foil problems. M esh refinement studies are needed to explore fully the per-

formance of these models, however. In addition, only the LRR FRSM was




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 predic-

tive performance.

IV. Conclusions

An extensive series of RANS calculations have been performed on two-

dimensional 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 logarith-

mic-spiral T E 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 log-

arithmic-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. ir Force’s Aeronautical

Systems Center at Wright-Patterson Air Force Base, Ohio (Origin 3900, hpc-

11), 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 Navier-

Stokes 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 Tw o Related

Cambered 15-Percent-Thick Circulation Control Airfoils,” Tech. Rept. ASED-373,

DTNSRDC, Sept. 1977.



466 P.

A.

CHANG

ET

AL.

Liu, Y., Sankar, L., Englar, R., and Ahuja,

K.,

“Numerical Sim ulations of the Steady

and Unsteady Aerodynamic Characteristics

of

a Circulation Control Wing Airfoil,” 39th


’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 A ffected by Different Turbulence M odels,” 40th AIAA

Aerospace Sciences Meeting Exhibit, AIAA Paper 2002-0851, Jan. 2002.

“Paterson, E. G. nd 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.

nd Rodi, W., “Progress in the Development of a Reynolds-

Stress Turbulence Closure,”

Journal

of

Fluid Mechanics,

Vol. 68, No. 3, 1975, pp.

”Launder, B. and Shim a, N., “Second-M oment Closure for the Near-W all Sublayer:

6

537-566.

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, T w q o o

CL

=

section lift coefficient,

L / ( q . d )

C

=

pressure coefficient, (p oo) /qoo

C =jet momentum coefficient, ( h V , ) / ( q , A )

cr3

= parameter for curvature effects

c = chord length, in.

h

= slot height, in.

k = turbulent kinetic energy per unit mass, ft . b/slug

M = Mach num ber, V / a

m = mass flow rate, slug/s

p

=

pressure, lb/ft2

q

=

dynamic pressure, p V 2 , lb/f t2

R

=

gas constant, ft . bfslug .OR

Re

= Reynolds number,

( pV,c) /p

T = Temperatu re, OR

u,

=

friction velocity,

m,

t /s

V = velocity, ft/s

u

v = Cartesian velocity components, ft/s

*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 005 by the American Institute of Aeronautics and Astronautics, Inc.

No

copyright is




469



470

R.

C. SWANSON, C. L. RUMSEY, AND S. G. ANDERS

x,y = Cartesian coordinates, in.

y =

normalized coordinate,

y ( u 7 / v )

a =


y

=

specific heat ratio

E

= dissipation rate of k, ft . b/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 00 ) condition

t = turbulent flow quantity

w =

solid surface (wall) condition

0 = total condition

00

=

freestream quantity

I. Introduction

ONVENTIONAL high-lift sys tems use slats and flaps to create the necess-

C ry 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 aerodyna-

micists 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 con-

figurations. 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



NUMERICAL SIMULATION OF CC AIRFOILS 471

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 s . ~

spiral grid that begins in the plenum and wraps around the airfoil several times

was used for the computations. Turbu lence modeling of the flow over the airfoil

and Coanda surface was carried out by applying a modified form of the zero-

equation 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 of je t momentum coefficient

C,.

The

C

was adjusted

so

that the com -

puted 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. Neverthe-

less, Pulliam et a l. were ab le to obtain good comparisons with experimental data

for all cases considered. This work demonstrated that accurate Navier-Stokes

simulation of C C airfoil flows is possible, and turbulence modeling is the key issue.

In 2002 S lomski 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 - ~

odel)

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 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 incom pressible simulation

was perform ed. 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 exam-

ined. These aspects include 1) flow conditions, 2) grid density, and

3)

turbulence

modeling. The primary purpose of this paper is to provide some gu idelines for

accurate solutions and to delineate improvem ents needed in numerical techniques

to reliably predict CC flows, eventually including pulsed jets. The two-dimen-

sional, compressible, mass-averaged Navier-Stokes equations are solved with

a finite-volume approach for discretization. The equations are solved on a multi-

block, patched grid, and a multigrid method with an implicit approximate



472

R.


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 trans-

port 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 rem arks are given.

11. Geometry and G rid

The CC geometry for the 2004 Circulation Control W orkshop” 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 corre-

sponds to a slot height of 0.024 in.

Previously, we performed calculations for the CC airfoil that was tested by

Abramson and R ogers5 (see also W ilkerson 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 cam ber 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

0

.~......................................................................................................................

0 3 j

..................

CCR

coarlirrates ...................

:I;

T

----- 103RE coarlimtes

0

. ;

.....................

~

...............................................

~

..............................................

;

0.1

0

- 0 . 1

-0.2

- 0 . 3

~

.....................

~

...............

........,........

............... ................

.......,

......................

:

; ------ i ----

......................

~

.......................

~

........ ...~......... ................

.......,

..........

...

.....

; --

I - - - -

_ _ _ - -

-;- ---- -

L - - -

-_ -

1 -

......................................................................................................................

~

.....................

~

...............

........,........

............... ................

.......,

......................

:

; ------ i ----

......................

~

.......................

~

........ ...~......... ................

.......,

..........

...

.....

; --

I - - - -

_ _ _ - -

-;- ---- -

L - - -

-_ -

1 -

......................................................................................................................

1

i

1

.2 0.4

0.6

0 8 1

0 . 4

x l c

Fig.

1

Geom etry of airfoils.



NUMERICAL SIMULATION OF CC AIRFOILS

473

size of the plenum and the je t 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 exper-

iments, 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 Abram-

son.12 Surface pressure distributions are available from the experiments. There are

no velocity profiles or Reynolds stresses to allow a detailed assessment of turbu-

lence models. Nevertheless, pressure data provide an opportunity for initial evalu-

ation 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 avail-

able specifying the error bounds of the aerodynamic coefficients. Several sources

of

error in the experimental data w ere reported by Abramson. 2 Although the exper-

iments were generally two-dimensional, there were three-dimensional effects pro-

duced 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

eg

CP

302

0.6 5.2 x lo6 0 0.0032

283 0.12 5.45 lo5 0 0.2090

321 0.12 5.45 lo5

-8 0.1840



474

R.


0

0.5

0.4

0.3

0.2

0.1

u

s o

-0.1

-0 .2

-0 . 3

-0.4

-0.5

0.5 1

0.5

0.4

0.3

0.2

0.1

0

- 0 .1

-0 .2

- 0 . 3

- 0 . 4

-0.5

0 0.5 1

x l c

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 je t 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

0.08

0.06

0.04

0.02

g o

-0.02

-0.04

-0.06

-0.08

-0 .1

-0.05

0

0.05

0.1

0.1

0.08

0.06

0.04

0.02

0

-0.02

-0.04

-0.06

-0.08

-0.1

-0.05

0

0.05 0.1

XlC

Fig.

3

Leading-edge region of m edium grid for

CC

airfoil.




475

0.95 1 1.05

0.08 0.08

0.06 0.06

0.04 0.04

0.02

0

0

0.02

0

-0.02 -0.02

-0.04 -0.04

-0.06 -0.06

XlC

0.95 1 1.05

Fig.

4

Trailing-edge region of medium grid for

CC

airfoil.

111. Numerical Method

Numerical solutions were computed with CFL3D, a m ultizone 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 accu racy.

Thus, this code has sufficient flexibility to solve either two-dimensional or three-

dimensional 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 differen-

cing. 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 mean-

flow 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 pro-

duction source term is treated explicitly, while the advection, destruction, and dif-

fusion terms are treated implicitly. For the explicit algebraic Reynolds stress

(EASM-ko) model, the nonlinear terms are added to the Navier-Stokes

equations explicitly.



476

R.


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 pre-

scribed. 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 con-

ditions are specified. Boundary conditions for the various turbulence models con-

sidered 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 C C airfoil flows are considered. The

three principal models are the one-equation Spalart-Allm aras (SA) model,14 the

Spalart- Allmaras rotation/curvature (SARC) and the two-equation

shear-stress transport (SST) m odel of M enter’0”79’8. In addition, the zero-

equation Baldwin-Lom ax (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 Bous-

sinesq eddy-viscosity hypothesis, whereas the EASM-ko model is a nonlinear

model. The equations describing these four models can be found in their respect-

ive references. H owever, there are certain de tails 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, nd

P ,

Vaiff, nd Ddiss are the contributions associated with turbu-

lence resulting from production, diffusion, and dissipation, respectively. The pro-

duction term is given by

P = C&l[l

523WV

(2)

In the SARC model P is replaced by

..*




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

rl

=

1

cr2

=

12, and

cr

=

0.6-1.0. As

cr3

is increased, the turbulence pro-

duction 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 k inetic 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 n the w

equation of the SST model is proportional to Pk. Generally, in the com putations

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 i s substituted for the strain rate. l 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

a1

k

max (a1w;RF2

vt =

(9)

where a l is a constant, w is equal to the ratio of the turbulent dissipation rate to

the turbulent kinetic energy, R =

/m

nd

F2

is a blending function. In a

recent paper by Menter et a1.,20 the R in Eq.

(8)

is replaced by S

=Jm

n

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.



478

R. C.

SWANSON,

C.

L. RUMSEY, AND S.

G.

ANDERS

VI. Jet Momentum Coefficient

A frequently used parameter in assessing the performance of C C devices is the

je t momentum coefficient. This parameter is defined as

l i j v j =

pjvj2hb

c =

,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 M ach numbers considered in

this chapter)

Thus, for a given C with

M ,

= 0.12, the C corresponding to

M ,

= 0.6 is m ore

than an order of magnitude smaller. One must keep this behavior in mind when

considering C as M , increases.

VII. Num erical 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 experim ental5 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




481

0.968 0.969 0.97 0.971 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 T E 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

P

*

0

0.05

0

-0.05 -0.05

-0.1

1 1.05 1.1

-0.1

0.85 0.9 0.95

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 1 1.02 1.04 1.06 1.08

0.06 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 , =





483

0.85 0.9 0.95 1 1.05 1.1

0.1 0.1

0.05

P

*

0

0.05

0

-0.05 -0.05

-0.1 -0.1

0.85 0.9 0.95 1 1.05 1.1

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 , =


There is improvement in the agreement w ith the SST(base1ine) model. T he

solution with the SAR C 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 5000 7500

-4

5

-5 ...................................

-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.


T o 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 conver-

gence is the slowly converging plenum solution, which is a consequence of the

very low-s eed flow in the plenum. The implem entation of low-speed precondi-

tioning,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 pre-

viously 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.

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, sol-

utions with both the S A and SST(base1ine) models indicate je t 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 sep-

aration results in a barely discernible plateauing effect on the calculated pressures

in Fig. 14, which i s not consistent with the experimental data. Figure 16 shows the

je t 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

25

-1 0

-8

-6

-4

-2

0

0

2

0.2 0.4 0.6 0.8

1

X/C

4 0

Fig. 14 Surface pressures computed

with

SA, SARC cn = 9.6), and SST urbulence

models

( M , =

0.12, (Y

=

0 deg, Re, = 5.45 X lo5,

C =





0.2

0.6

0.7

0.8

0.9 1 1.1

0.2

0.1 0.1

0 0

s

0.1 -0.1

-0.2 -0.2

-0.3 -0.3

-0.4 -0.4

X/C

0.6

0.7 0.8 0.9 1 1.1

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

Y

>,

0

0.05

0

-0.05 -0.05

-0.1 -0.1

0.85

0.9

0.95 1 1.05 1.1

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.


-1 0

-8

-6

-4

0

2

4

0 0.2 0.4 0.6

X/C

0.8

1

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 com puted 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 je t 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 pre-

sented in Fig. 18. In the Mach contours of Figs. 19 and

20

the rearward m ovement

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 je t exit, along w ith the

corresponding fine grid, are displayed in Figs. 21 and 22. The je t flow is acceler-

ated 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




487

0.85

0.1

0.05

Y

*

0

-0.05

-0 1 0.85

0.9 0.95 1 1.05 1.1

0.1

0.05

0

-0.05

-0.1

0.9 0.95 1 1.05 1.1

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 com pares 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 pre-

viously, the experimental CDvalues include the thrust effects produced by the jet ,

-0.1 0 0.1 0.2

0.2

.2°.2

0.1 0.1

0

0

-0.1 -0.1

-0.2

-0.2 -0.1 0 0.1 0.2

0.2

XlC

Fig. 19 Mach contours computed at LE with

SARC cr3

=

0


( 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

P

*

0

0.05

0

-0.05 -0.05

-0.1

0.85 0.9 0.95

1

1.05 1.1

XlC

-0.1

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 alues 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 s 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

P

*

0.034

0.036

0.034

0.032 0.032

0.966 0.968 0.97 0.972

XlC

Fig. 21 Fine grid in jet exit region.




0.008

0

007

0

006

0

005

0

004

0

003

Y

*

0

002

0 001

0

-0.001

0.966 0.968 0.97 0.972

0.038 0.038

1 1

.........

medium gr id

f ine gr id

.........

I

......................................................

j

......................... ~ ..............................

_ ........................................................ ......................... i..............................

......................... i ............................. ............................ i..............................

1

......................... ............................. ;+...................

1 1

1

- ......................... i.............................. ............................ i................

...............................................................................................

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).

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).




1

-1 0

-8

-6

-4

-2

0

0

2

4

0

0.2

0.4

0.6 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 i s 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

0.96 0.97 0.98 0.99 1

X/C

-0.1

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 -0.05

-0.1

-0.1

0.85 0.9

0.95 1 1.05 1.1

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 aerodynam ic flows of interest, it does not appear to be appropriate for the

Coanda je t flows being considered here; the SST( 1994) model performs better for

these particular low-Mach-number Coanda flows.

There is g reater sensitivity to m esh 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 w ith the SST( 1994) model exhibit less spreading than those with

the SARC(cr3

=

0 - .6) model. Mesh refinement effect on the predicted CL

and C Dwith the SST(1994) model is given in Table

3.

On the fine grid, the pre-

dicted 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 0.209 Medium

4.20

4.19 .050

0.0966

283 0.209 Fine 4.20 3.88

-

.050 0.0746

321 0.184 Medium 3.10 2.96 - .080 0.0655

321 0.184 Fine 3.10 2.41

-

.080 0.0559




493

4.5

4

3.5

3

2.5

2

0

1.5

1

0.5

0.05 0.1 0.15 0.2 0.25

c,

Fig. 28 Variation

of

lift coefficient with 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

C L

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 Slom ski 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

0.2 0.4

0.6

0.8

1

XlC

4 0

Fig. 29 Surface pressures computed with SARC cr3 =

0


( M , = 0.12,

a

= -8 deg, Re, = 5.45 X lo5,

C

= 0.184).



494

R.


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

0.95

1 1.05 1.1

0.1 0.1

0.05

P

*

0

0.05

0

-0.05

-0.05

-0.1

0.85 0.9 0.95 1 1.05 1.1

XlC

-0.1

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).




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 Tab les

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 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-1510-

7067 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 extre-

mely small recirculation region , w hich occurs only for the SS T( 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 high-

subsonic 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 overpredic-

tion of lift. In other words, all of the models have produced near-wall



496

R.


eddy-viscosity levels that are too high in the Coanda flow. A parameter c,3) in

the curvature correction term of the SAR C 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 calcu-

lations 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

-

.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 je t flows. An indication of the effects of grid refine-

ment on the results computed with the turbulence models has been given. The

SST( 1994) model has shown greater sensitivity to mesh refinement than the

SAR C(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 C C airfoil flows. Most standard

models, including SA, SARC

(c,3

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 com parison 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 con-

sidered 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 turbu-

lence 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 m ust anticipate that there will

be a reduction in computing efficiency, because of a lower degree of numeri-

cal 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.




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

X I . Y

I.

X I .

Y l C

X I Y

I. X I C

Y

I.

0.91346 0.010565 0.99999 0.0035035

0.093469 -0.0441 14 0.44534 0.087908

0.91436 0.016505 0.99939 -0.0027953 0.072353 -0.039186 0.52873 0.089357

0.91762 0.023061 0.99700 -0.0093482 0.055038 -0.034379 0.60007 0.08861 1

0.92444 0.028929

0.99243 -0.015540 0.040894 -0.029713 0.66105 0.086478

0.93516 0.031569 0.98563 -0.020621 0.029411 -0.025191 0.71316 0.083469

0.94627 0.032378

0.97807 -0.024547 0.020176

-0.020794 0.75764 0.079735

0.95247 0.032618 0.96976 -0.027987 0.012873 -0.016494 0.79558 0.075445

0.95592 0.032815 0.96123 -0.030852 0.0072776 -0.012252 0.82791 0.070840

0.95892 0.033013 0.95121 -0.033717 0.0032691 -0.0080402 0.85543 0.066064

0.96079 0.033051 0.93948 -0.036499 0.00083659 -0.0038889 0.87885 0.061361

0.96232 0.033040 0.92582 -0.039348 0.00027135 -0.0021 13 0.89877 0.056818

0.96419 0.032955

0.90993 -0.042273

0

0

0.91570 0.052486

0.96527 0.032875 0.89147 -0.045275 0.00022742 0.0021185 0.93008 0.048398

0.96646 0.032761 0.87004 -0.048349 0.00075518 0.0039056 0.94228 0.044569

0.96776 0.032607 0.84518 -0.051494 0.0031043 0.0081157 0.95262 0.041005

0.96920 0.032403

0.81636 -0.05471 1 0.0070381

0.012430 0.96137 0.037705

0.97078 0.032117 0.78295 -0.058003 0.012574 0.016818 0.96877 0.034657

0.97250

0.031733 0.74424 -0.061375

0.019838 0.021317 0.96877 0.034491

0.97439 0.031228 0.69939 -0.064825 0.029062 0.025973 0.94527 0.038834

0.97644

0.030569 0.64742 -0.068009

0.040572 0.030832 0.89801 0.038252

0.97866 0.029731 0.58721 -0.070608 0.054791 0.035930 0.83406 0.030760

0.98108 0.028721 0.5 1748 -0.072286 0.072250 0.041298 0.50044 0.030760

0.98367 0.027438 0.43673 -0.072472 0.093598 0.046950 0.50044

-

.041077

0.98643 0.025852 0.35601 -0.070647 0.11963 0.052894 0.73553 -0.041077

0.98933 0.023862 0.28917 -0.067442 0.15131 0.059114 0.91346 0

0.99231 0.021385 0.23385 -0.063413 0.18980 0.065569 0.91346 0.010565

0.99512 0.018159 0.18811 -0.058917 0.23654 0.072089

0.99753 0.014103 0.15032 -0.054098 0.29326 0.078324

0.99925 0.0091942 0.11915 -0.049115 0.36203 0.083837

References

‘Englar, R. J., and Williams, R. M ., “Test Techniques for High Lift Two-Dimensional

Airfoils w ith Boundary Layer and Circulation Control for Application to R otary Wing A ir-

craft,”

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 Appli-

cations,” 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, A RC3D,”

Notes fo r Computational Fluid Dynamics User’s Workshop, Univ. of Tennessee Space

Inst., Tullahoma, TN , March 1984.



498

R.


’Abramson, J., and Rogers, E., “High-speed Characteristics of Circulation Control Air-

foils,” AIAA Paper 83-0265, Jan. 1983.

6Wilkerson, J. B., and Montana, P. S., “Transonic W ind 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 M odel for

Separated Flows,” AIAA Paper 78-257, Jan. 1978.

‘Slomski, J. F., Gorski, J. J., Miller, R. W., and M arino, T. A., “Numerical Sim ulation

of Circulation Control Airfoils as Affected by Turbulence Models,” AIAA Paper 2002-

0851, Jan. 2002.

’Paterson, E. G. nd Baker, W. J., “Simulation of Steady Circulation Control for

Marine-Vehicle Control Surfaces,” A IAA 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. .,

and Joslin, R. D. (ed.), Proceedings of the 2004 NASA/ONR Circulation

Control Workshop, NASA/CP 2005-213509, March 2004.

”Abramson, J., “Two-D imensional 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 U ser’s Manual,” NA SA TM

1998-208444, June 1998.

14Spalart, P. R., and A llmaras, S. R., “A One-Equation Turbulence Model for Aerody-

namic Flows,” La

Recherche Aerospa tiale,

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 W all-Bounded Turbulent Flows,” International Journal of Heat and

Fluid F low, 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 urbulence 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 M odel,” Turbulence, Heat and Ma ss Transfer 4, edited by K. Hanjalic,

Y. Nagano, and M . Tumm ers, Begell House, Redding, CT, 2003, pp. 625-632.

’lTurke1, T., Vatsa, V. N., and Radespiel, R., “Preconditioning M ethods

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 C ontrol Airfoils,’’ AIAA Paper 2005-0089 , Jan. 2005.



Chapter

19

Role of Turbulence Modeling in Flow Prediction

of Circulation Control Airfoils

Gregory M cGow an,* Ashok Gopalarathnam,+ Xudong Xiao,*

and Hassan Hassans

No rth 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

r z = mass flow rate

M =

Mach number

V = velocity

eff

effective angle of attack

p

= density

w = turbulence frequency

=

enstrophy

Subscripts

j

= j e t

= freestream conditions

*Research Assistant, Department of Mechanical and Aerospace Engineering. Student Member

'Associate Professor, Department of Mechanical and Aerospace Engineering. Senior Member

'Research Assistant, Professor, Department of Mechanical and Aerospace Engineering. Me mbe r

%Professor, Departm ent of Mechanical and Aerospace Engineering. Fellow AIAA.

Copyright 005 by

th


AIAA.

AIAA.

AIAA.


499



500

G. McGOWAN

ET

AL.

I. Introduction

HERE IS a continuing need to pursue technologies that can address long-

T

erm aerodynamic goals for future aircraft. For example, long-term visions

for aeronau tics predict the need and potential for a dram atic

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 ident-

ified that the benefits of flow control are best realized by the development of sim-

plified 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.

Circu lation control (CC ) is one type of flow control that has received consider-

able 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 cen-

trifugal force (the Coanda effect), the je t remains attached along the surface of the

wing. Thus, the T E 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 well-

attended 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 m odels including algebraic, one-equation, two-equation, and

stress models to try to predict flow characteristics of various C C airfoils. None of

the models employed performed well for all je t mom entum coefficients

C

con-

sidered. The only exception is the Spalart-Allmaras model that includes curva-

ture 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 i s to consider the flow over the 103RE(103XW)

CC airfoil tested by Abramson and Rogers.6 The tests were conducted to deter-

mine 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 m odeling the exact equa tions that govern the variance of

velocity, or turbulence kinetic energy,

k,

and the variance of vorticity, or enstro-

phy, 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 (je ts, wakes



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 com plex 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 transi-

tional 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 Toll-

mien-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 con-

ditions are detailed in the cited references. C alculations are also presented fo r 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 )

s 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 d irection 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. Num erical Procedure

The numerical solution was computed using the code CFL3D.’ It is based on a

finite volume m ethod. The convective terms are approximated by upwind-biased

spatial differences, and the viscous terms are discretized using central differ-

ences. In this work, the flux difference splitting of Roe is employed. Time inte-

gration 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 pre-

scribed. At the surface of the airfoil, no-slip and adiabatic wall conditions are

employed.



502

G. McGOWAN

ET

AL.

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

1/2pmv:c

-

where

j

is the je t m ass flux per unit span, vj is the je t velocity,

poo

and Vm are the

freestream density and velocity, and j is the jet Mach number. The effective

angle of attack a , ~as 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 inver-

sely 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 0.0

0.0

.0540

302 0.0032 0.519 .2865

306 0.0110 0.979 .7980



TURBULENCE MODELING IN FLOW PREDICTION 503

0.8

0.6

0.4

0.2

0

0.2

0

0.4

0.6

0.8

1

0

0.25 0.5 0.75

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 com pares calculated and measured pressure distri-

bution 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

tur-

bulence 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 calcu lated skin-friction coefficients using the k-5

model. The transitional behav ior indicated in the figure is a result of a numerical

transition. This is typical of all turbulence models.

Figure

7

com pares 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 fo r 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




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

0.006

0.005

0.004

u

0.003

0.002

0.001

0

0 0.2 0.4 0.6 0.8 1

XlC

Fig. 6 Calculated skin friction using k -j m od el for Case 302

C

= 0.0032.

x/c

C f

0 0.2 0.4 0.6 0.8 1

0

0.001

0.002

0.003

0.004

0.005

0.006

0.007

lower surface

upper surface

Case 302



506

G. McGOWAN ET AL.

-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 - j o n fine and coarse grid for Case 302

C

=

0.0032.

1.5

-1

0.5

e

0

0.5

1

0

0.2 0.4

0.6 0.8

XlC

Fig. 8

Prediction of

C

using k - o model for C ase 306

C

= 0.011.



4

4.5

5.5

6

TURBULENCE MODELING IN FLOW PREDICTION 507

t

Case306

0 5000 10000 15000 20000 25000

I terat ion

Fig.

9

Convergence history using k -j m od el 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

ransi-

tional/turbulence model to analyze the flow for Case 306 because such a model

will result in reduced eddy viscosity and earlier separation. The m odel, 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.

Th e transitional m echanism 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.




509

Y

*

Fig. 12 Streamline pattern around separation point for Case 306

C,

= 0.011 with

laminar cavity.

Case 306

turbulent cavi ty

Y

*

I I I I I I I I I I L

1 1.05

XlC

Fig. 13 Streamline pattern around separation point for Case 306

C,

= 0.011 with

turbulent cavity.



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 ) p U k S

C, pressure coefficient, p

, ) / 1 / 2 ) p ~ k

C ,

et mom entum coefficient,

~ U ~ / I / ~ ) P U ; S

c

chord length, in.

h slot height, in.

k turbulent kinetic energy (TKE), f t2/s2

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 polynom ial 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

Copyright 005 by Warren J. Baker and Eric G. Paterson. Published by the American Institute

and Nuclear Engineering. Member A I M .

of Aeronautics and Astronautics, Inc., with permission.

513



514 W.

J.

BAKER AND E. G. PATERSON

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

3 freestream

at jet-slot exit

I. Introduction

HE CONCEPT of circulation control (CC) using the Coanda effect is a

T

henomenon involving a two-dimensional wall bounded jet passing along

a curved surface. The jet itself is introduced via a slot, which expels the jet , typi-

cally, tangentially to the curved surface. This je t adds m omentum 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. T he 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 aerodynam ic values such as lift, total drag, and local skin fric-

tion coefficient. Figure 1 shows an example of a Coanda jet CC setup with a

single slot.

The performance benefits of CC have been show n in many experiments since

the early 1 9 7 0 ~ . l - ~ncreases in lift of as much as 10 times the typical flap system

Fig.

1

Tra iling-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 attrac-

tive for certain naval applications, in particular the replacement of current actua-

tion 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 i s 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 suffi-

cient 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 simu-

lation tools using m ultiple data sets. These data sets include a two-dimensional

CC experiment using the NCC R 15 10-7067N,* a low-aspect-ratio, tapered,

control surface for marine app lications, CCFOIL,3 and the General A viation Cir-

culation 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 Navier-

Stokes (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 y ield-

ing 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 cross-

section of the model is shown in Fig.

2.



51

6

W.

J.


UPPER

STEADY

MAMFoLo

ACTUAtOA UPPER

mSm SLOT

Fig. 2 Cross-section of the GAC C 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 non-

dimensional je t velocity U j / U W 8 34:

For all cases studied, the angle of attack was 0 deg. Experimental data are

availab le for u er slot steady blowing, lower-slot steady blowing, and dual-

assist blowing. The most recent experimentation completed focuses on pulsed

actuation, and initial data from pulsed testing are available.6 Tab le 1 summ arizes

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)’ave 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? Tw o 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?

BP

-

111.

Computational Methods

The flow code used for the current work, CFD SHIP, is a general-purpose, par-

allel, unsteady, incompressible, RANS CF D code. The computational approach is



STEADY CC SIMULATION FOR GACC WING

51

7

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

1)

Surface pressure distribution CL 1.2)

2) Lift vs mass flow rate

1) “Negative lift configuration,” lift vs blowing

2) “Negative lift configuration,” drag polar

1) Drag polar (slot height 0.01 in. and 0.02 in.)

2) Drag polar (matched slot

C ,

0.0, 0.004,

3) Drag vs angle of attack

4) Angle

of

attack vs

L I D

Pulsed upper-slot blowing

Steady lower-slot blowing

coefficient

Dual-slot assist steady blowing

0.005, 0.009,

0.021, and 0.0041)

Fig. 3 C , vs C ,

for

previous C FD simulations and experiment



518

W.

J.


Fig.

4

Drag polar for previous CFD sim ulations and experiment

based upon structured, overset-grid, higher-order finite-difference, and pressure-

implicit split-operator (PISO) numerical methods. Production turbulence model

uses a linear closure and the blended

k -w /k -E

SST two-equation m odel.* Effi-

cient parallel computing is achieved using coarse-grain parallelism via MPI dis-

tributed 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 orthog-

onal 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 inter-

polated 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




a)

519

Fig. 5 Overset computational dom ain including the plenum: a) Overa ll view;

b)

foil

view; c) plenum view.



520 W.

J.


spacing ranged between 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 T E view is show n in Fig. 6, because the computational

domain i s very similar to that show n in Fig. 5 in all regions except the TE. The

difference in grid point number between the tw o 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 w ith 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

and 2.00 x

0.M

0

4.M

0.96 0.97

0 .a 0.99

1 1.01

1.m

XlC

Fig. 6 Trailing-edge view of grid without plenum.




521

Table 2 Total grid points for the

fine, medium, and coarse grids

Grid

Total

grid

points

Fine

381,810

Medium

193,980

Coarse

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 m edium 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 l op6 respectively. Because of smooth-

ing of some of the com putational 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 con-

ditions were applied to the upper and low er 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 Upoly s a tenth-order polynomial

curve fit of a typical CC jet profile seen in a previous RANS results for the GACC



522

W.

J.


S 0 0

Dose

9s7

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) (3)

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

where y/1 is the nondimensional distance along the boundary. T o acquire tangen-

tial 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




523

t

0.001

~ . ' . ~ . ' ' ' ~ . ' ' . L

0

0.5 1 15

U

Fig. 8 Velocity profile prescribed for steady blowing boundary condition at the jet-

slot 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 bound-

ary condition for steady blowing with the grid including the plenum is given

in Eq. (7):

(7)

yetx ramp

VI. Com putational Resources

All simulations were executed on an IBM SP Power 3 machine w ith 64 nodes.

Each node contains sixteen, 375 MH z Power

3

processors. Each C PU has 64 kB

level-1 cache and 8 MB level-2 cache memory along with 1 GBRAM ach pro-

cessor 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://ww w.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 com-

pleted) 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



524

om

om

OD

W.

J.


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.

A.

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 iter-

ation; 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




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 w ithout the plenum was chosen for further simulations. A w ide

range of blowing coefficients was studied, and results were compared to exper-

iment and FUN2D simulations. Experimental data included the surface pressure

distribution for C, 0.059. The corresponding results from CFDSHIP are com -

pared 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 investi-

gation 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~ p er im en ta tio n .~

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, .091. At higher values of C, where no experimental

data have been recorded, the results vary from FUN2D solutions. The variations

in FUN 2D and CFDSHIP resuls at the highest blowing coefficient are observable



526

W.

J.


XlC

x k

Fig. 11 Total velocity contours and streamlines

of

the baseline case for

computational dom ains a) without the plenum and b) with the p lenum.

by investigating the total velocity contours, shown in Figs.

14

and

15,

respect-

ively. 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

“ d r a w dow n e f f e ~ t ” . ~ntil more experimental data are obtained,

it

is difficult

to know which of the FUN2D and CFDSHIP simulations is more accurate.




527

X

Fig. 12 Surface pressure distribution for experiment and simulation,C = 0.059.

Fig. 13 Lift

vs C

for experim ent and simulations.



528

W.

J.


Fig 14 Mach contours for FUN2D simulations with

C, =

0 162:

Fig 15 Total velocity contours for CFDSHIP simulations with

C,

= 0 162




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 wave-

length of the oscillation increases. Figure 17 shows that the surface pressure

NarrdhrerrpknalThre

Fig. 16 a) Lift force and b) drag force h istories for a wide ran ge of C,.



530

W.

J.


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

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 je t orifice

and terminating at the jet separation. The second region of high TKE denoted as

b) in Fig. 18, originates near the je t 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 je t 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 iter-

ations for each grid. The blowing coefficients used in the earlier work were now



532

W.

J.


Table 3 Grid size and runtime characteristics for grid study

Coarse Medium Fine

Grid points

97,575 193,980 381,810

Seconds/time step 1

o

2.8

6.5

W all-clock hours 3.6 9.9 16.7

CPU hours 29.1 79.1 133 .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 dis-

tributions 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

c,

Fig.

19

Lift

vs

C, for experiment and sim ulations for grid study.



534

W.

J.


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




a)

535

Fig. 22 Velocity contours for a) coarse,

b)

medium, and c) fine grids with

C,

= 0.59.



536

W.

J.


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 approxi-

mately 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 low er 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 “drawdow n 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 reattachm ent 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 sol-

utions 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 approxi-

mation based on R eynolds number yielded a near-wall spacing of 2.00 x

For the fine grid, near-wall spacing was set at 2.00

x

lo p 6 , with the coarse-

grid near-wall spacing set at 4.00 x lop6.Even with the increased near-wall fide-

lity chosen, the medium and coarse g rids proved to be deficient at higher blowing

coefficients. W ithout a method to determine proper near-wall spacing require-

ments 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 esti-

mation p rocess.

VIII.

Conclusions

The GACC wing was studied using non-time-accurate, RAN S 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

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 opti-

mizing 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 dom ain, 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 Bibli-

ography of DTNSRDC Research and Selected Outside References,” DTNSRDC Rept.

77-0076, Sep. 1977.

’Abramson, J., “Two-Dimensional Subsonic Wind Tunnel Evaluation of Tw o Related

Cambered 15-Percent Circulation Control Airfoils,” DTNSRDC A SED-373, Sept. 1977.

3Rogers, E. O., and Donnelly, M. J., “Characteristics of a Dual-Slotted Circulation


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 Appli-

cations,” 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,” 21 st Applied Aerodynamics Conference, AIAA Paper 2003-

341 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,

‘Menter,

F.,

“Two-Equation Eddy Viscosity Model for Engineering Applications,”

’Suhs, N., Dietz , W., Rogers,

S.,

Nash,

S.,

and Onufer, J. T., “PEGASU S User’s Guide

1994, pp. 1-21.

AIAA

Journal,

Vol. 32, No.

8,

1994, pp. 1598-1605.

Version 5. lg ,” Tech. Rept., NASA, May 2000.



Chapter 21

Computational Study of a Circulation Control

Airfoil Using FLUENT

Gregory M cGowan* and Ashok Gopalarathnamt

No rth 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

= 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 a nd Aerospace Engineering.

'Associate Professor, Department of M echanical and Aerospace Engineering.

Copyright 005 by

the



539



540

G. McGOWAN AND A. GOPALARATHNAM

p

=

density

Subscripts

duct

=

stagnation conditions inside plenum

fc

=

conditions at flow-control boundary

3

=

freestream conditions

J

= slot-exit conditions

I. Introduction

ECENT research in the Applied Aerodynamics G roup at the North Carolina

R

tate University (NCSU ) has led to the developm ent of an automated cruise-

flap system.lq2 he cruise flap, introduced by P f e n n i ~ ~ g e r , ~ ~ ~s a small trailing-

edge (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 au tomation

is achieved by indirectly sensing the leading-edge (LE) stagnation-point location

using surface pressure measurements and deflecting the flap

so

that the stagna-

tion-point location is maintained at the optimum location near the LE of the

airfoil. Maintaining the stagnation point a t the optimum location results in favor-

able 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 NL F 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 capab ility for generating very high lift at takeoff and landing con-

ditions. Toward this objective, it is of interest to study an approach that integrates

aerodynam ic 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 Coanda-

based 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 d rag reductions through ex tensive

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



STUDY OF CC AIRFOIL USING FLUENT

54

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 w as the FLUENT flow solver. The results are com-

pared to CFD and experimental data obtained from a recent study by Jones

et al? of a General Aviation CC (GA CC ) 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 adap tive

CC airfoil concept, the effects of CC on the LE stagnation-point location were

also exam ined in the current w ork.

The following section gives an explanation of the geometry under examin-

ation 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 wind-

tunnel 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 geom etry chosen for the current research w as the GA CC airfoil, designed

by Jones.15 The GACC airfoil was derived from a

17

GAW(

1)

airfoil by mod-

ifying 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

current research.

used in the



542 G. McGOWAN AND A. GOPALARATHNAM

The wind-tunnel experimen ts were conducted by Jones et al.15 in the Basic Aero-

dynamic Research Tunnel (BART),which is located at the NASA Langley

Research Center in Hampton, Virginia. The BART tunnel has a physical test-

section size of 28

x

40

x

120 in. The G ACC 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 wind-

tunnel 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.




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 curre nt study the solution is assumed to be steady and is not run time-

accurate. 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 FLU EN T analysis. There was an attempt to run these problems with the seg-

regated (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 afte r a few thousand iterations. In order to com pare with the

FUN2D results of Ref. 15, the one-equation Spa lart- Allmaras turbulence

model was chosen for the current work. Wall functions were not used in the

FLUENT calculations.


FLU EN T does not allow the user to input the freestream Ma ch numbe r and

Reynolds number directly. Instead, the freestream velocity and operating

pressure were calculated using Eqs. (1-3) and provided as inputs for the ana-

lyses. Th e Ma ch 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

=

w c

An approximate method was developed to estim ate the velocity required a t the

flow control boundary U f c ) o achieve a desired

C,,

CPdesired.his method

assumes incompressible flow throughout the duct, and was derived by solving

the continuity equation. The equation for

U f c

rom this approximate method is



544

given in Eq.

(4):


Once FL UENT converged, an integration was performed across the slot exit as

shown in Eq. 5 ) to ob tain the actual

C

of the je t at the slot. This C,, however, is

different from

CPdeslred

ecause the Ufc for the latter is set using an approximate

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 isen-

tropic 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 hori-

zontal 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 differ-

ences are very small, approximately 3 at the highest blowing coefficient,

care must be taken to ensure consistency in the C FD solutions and experiments:

riZ =

PJUJAJ

(6)

Fig.

5

Comparison

of

Cphkgm dith

CpbeotroPic

or

a

=

0;

the stra ight 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 free-

air conditions is compared with the results presented in Ref.

15.

In the

second part, the predicted results for the GAC C airfoil with tunnel walls are pre-

sented 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 com-

parison 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

3.5

3

2.5

CI 2

1.5

1

0.5

0

1 0

5

0

5 10 15

degrees)

Fig. 6

experimental results from Ref.

15

(data points and curve

fits

for each

Cp .

Comparison of NCSU FLUENT results from the current work with Langley

−10 −5 0 5 10 150

0.5

1

1.5

2

2.5

3

3.5

4

α (degrees)

Cl

Fluent

Calculations

Experimental Results

Curve Fit to

Experimental Data

Cµ= 0.078→

Cµ= 0.047→

Cµ= 0.024→

Cµ= 0.008→

←Cµ= 0.0

←Cµ= 0.060

←Cµ= 0.041

←Cµ= 0.025

←Cµ= 0.015

←Cµ= 0.007

←Cµ= 0.0




Table

1

FLUENT results for the free-air cases

Lift coefficient,

CL

Blowing coefficient

(C,)

=

-5 d e g a = O d e g

=

5 deg

0.000

0.008

0.024

0.047

0.078

0.090

0.382

1.082

1.979

3.045

0.666 1.193

1.009 1.486

1.646 2.080

2.544 2.7 19

3.206 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 mcom-

pressible-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 predic-

tions 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 FLU ENT results are presented in two parts. In the

first part, the effects of increasing

C

for a constan t angle of attack are presented.

The second part exam ines the effects of angle-of-attack changes and their influ-

ence 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 blow ing 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 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 subsec tion, the FLU ENT results for the GA CC airfoil with the effect of

wind-tunnel upper and lower walls are presented . Figures

10

to 12show the influ-

ence of the walls on the CF D solution. These figures present the predicted Cl as a

function of C for a = 0, 5 , and - deg, respectively. Figure 10 also includes a



548

a)


Fig. 8 Circulation control effects on flow field at C, = 0.047 for various values ofa

a)

cu

=

-5

deg;

b)

cu

=

0

deg; c)

cu

=

5

deg.

comparison w ith 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;




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 three-

dimensional effects will be extremely important at larger blowing rates.

The results for the with-walls simulations consistently show that for low

blow ing coefficients, the Cl values are predicted to be lower than those for the



550


Fig. 10 FLU ENT 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

c

i 0.06 0.07 C

Fig.

11

FLU ENT prediction of wind-tunnel wall effects for varying values of C, at

a = 5 deg.

0 0.02 0.04 0.06 0.08 0.10.5

1

1.5

2

2.5

3

3.5

4

Cµ

Cl

Fluent (with walls)Fluent (free−air)FUN2D Jones et al. (free−air)

Experiment Jones et al.




551

Fig.

12

FLU ENT 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 Ta ble 2.

C. Stagnation-Point Location

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 s imilar approach cou ld be exten ded for use with C C airfoils.

Table 2 FLUEN T resu lts for cases with w ind-tunnel walls

Lift coefficient, CI

Blowing coefficient

cc

=

-5 d e g a = O d e g

=

5 deg

0.000

0.008

0.024

0.047

0.078

0.09 1

0.388

1.063

1.892

2.893

0.702 1.247

1.027 1.491

1.671 2.070

2.475 2.711

3.044 3.178



552

1.15

1.1

SIC

1

OE

1


+

=

0.000

+

=

0.008

Cw

=

0.024

-

C =

0.047

+ = 0.078

0 5

1

..............................

//

..........................

Fig.

13

Circulation control effects on

LE

stagnation-point ocation.

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 fo r 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 pre-

dictable 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 je t 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 pre-

dictions do not match the CFD and experimental results of Ref. 15 exactly, the

overall trends are followed very closely. Throughout the range of blowing coeffi-

cients, with the exception of the no-blowing case C, = O.O> FLUENT consist-

ently predicted a slightly lower overall lift coefficient.




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.

Th e influence of C C on the LE stagnation-point location was examined. I t was

shown that changes in blowing rate and angle of attack result in systematic

changes to the stagnation-point location. Th is observation indicates that it is poss-

ible to use a closed-loop control system that is driven by sensing the stagnation-

point location.

Acknowledgments

Th e authors would like to acknowledge the fund ing 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

‘ M c A v o ~ ,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.

*MCAVOY,. 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 Thick-

ness,” Journal of the Aeronautical Sciences, April 1949, pp. 227-236.

’Davidson, I. M., “The Jet Flap,”

Journal of the Royal Aeronautical Soc iety, 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 Jo urnal, May 1965, pp. 143-151.

’Lissaman, P. B.

S. ,

A Linear Solution f o r 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,” NA SA TN D-7252, April 1973.

981 -988.




l1

Abramson, 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 W all 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.

“ka gle , C. M., and Jones, G. S., “A Wind Tunnel Model to Explore Unsteady Circula-

tion Control for General Aviation Applications,” AIAA Paper 2002-3240, 2002.

”Jones,

G. S .


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,* Lakshm i

N.

Sankar,+ Robert J. Englar,'

Krishan K. Ahuja,$ and Richard Gaetall

Georgia Institute

of

Technology, Atlanta, Georgia

Nomenclature

a = angle of attack

Aje t=

area of jet slot, ft2

CL,Cl = lift coefficient

C,, Cd= drag coefficient

C, =jet momentum coefficient

f = pulsed jet frequency, Hz

m =jet mass flow rate, slugs/s

s= wing area, ft2

St =

Strouhal number

C,, = averaged jet mom entum coefficient

Lref= length reference, in.

TJet To,,,,

=

temperature and total temperature of the jet, K

Pj,, = pressure of the jet, psia

V , = freestream velocity, ft/s

V,,, = je t velocity, f t/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 A IAA .

TResearch Engineer, Georgia Tech Research Institute. Senior Member AIAA.

Copyright

005

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

D

ir 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

a i r ~ r a f t . ” ~epending on the type of aircraft, the dominant source varies between

flap, slat, and landing gear.4 Furthermore, these high-lift system s 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 w ing

(CCW) technology. This technology and its aerodynamic benefits have been

extensively investigated over many ears through experimental studies.596A

limited number of numerical analy~e$~-~ave also been carried out. Work has

also been done on the acoustic characteristics studies of C C 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, flap-

gaps, 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,596 as 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 conven-

tional high-lift system, and CC is deployed. The large curvature of the upper

surface produces a large je t turning ang le, 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,



EVALUATION OF STEADY AND PULSED JET EFFECTS

559

Supsrwidcal Contour C C W

Flap

Fig.

1

Dual radius CCW airfo il 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 dynam ics (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 . l ~he objective of this study is to isolate and quantify the effects

of parameters such as leading edge (LE) blowing, freestream velocity, jet slot-

height, and frequency on the performance of two-dimensional steady and

pulsed C C 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 semi-

implicit 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 so lver have been made to model



560

YI LIU ET AL.

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~~~ne-equation model. In this work, all the calculations

were done using the Baldwin-Lom ax model. The effects of the turbulence

model are discussed in Ref. 16.

B. Com putational 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 , ~ ~ , ~ ~nd 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.


defined as follows:

In CCW studies, the driving parameter is the momentum coefficient C,,

mVjet

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 2 a / (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.




561

At the jet slot exit, the jet is assumed to be subsonic, and the following con-

ditions are specified: total temperature of the jet Tojet, mom entum 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 exam ple 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

Pjet

7:tAjet

/ J 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, hich can be easily solved from Eq. (6).

After the qet s calculated, the other jet flow variables, such as yetand

pjet, can be obtained from Eq.

5 ) .

These parameters are also nondimen-

sionalized 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 M ach 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 CC W 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 experim ental measurem ents 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. Excel-

lent agreement with measured da ta from the experimen ts by Englar is evident. It

is seen that very high lift can be achieved by C C 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 augm entation

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 experi-

enced with a conventional airfoil, but occurs at higher Cl,mawalues, 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




563

4 -

-

CI, Measured

I, Com puted

I

0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4

CP

Fig. 3 Variation of the lift coefficient with the m omentum coefficients ata = 0 deg.

C,=O.1657

/-:

,=0.0740

EXP, C, 0.0

EXP, C, 0.074

EXP, C, = 0.15

-CFD

6

-4

-2 0 2 4 6 8 10 12 14 16

Angle

of Attack

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 conven-

tional 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




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 s tall 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

3

u

2.5

.-

i 2

- - -

LE Blowing,

C

=

0.04

2

LE Blowing,

Cy

.08

TE Blowing,

Cy

.04

5 1.5

Os5

0 2 4 6

8

10 12 14 16 18 20 22 24

Angle of Attack degrees)

Fig.

6

Lift coefficient

vs

angle of attack for the LE blowing case.



566 YI LIU ET AL.

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 num erical 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 free-

stream 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 mom entum 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 1.2 1.4 1.6 1.8 2

(Vm-cfd)

I

(Vm+xp)

Fig. 7 Lift coefficient vs freestream velocity

Cp

= 0.1657, h

=

0.015 in., and

Vm,exp

=

94.3

fps).




567

0.15

3

Q

U

0.1

E

8

0.2

-

0 1

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

(V--cfd)/ (V-exp)

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

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 % hen 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 m ass flow rate as

possible, a thin jet is aerodynam ically more beneficial than a thick jet. How ever,

of 94.3 fps.



568

YI LIU ET AL.

1

+ c p = o .o 4

I

- p

=

0.1657

3- -1

I

0.006 0.009 0.012 0.01 5 0.018

Jet Slot Height (inc h)

Fig. 9 Lift coefficient vs jet slot height

V ,

=

94.3 fps).

----

+ p = 0.04

-Cp =

0.1657

0.15

- .

0.006

0.009

0.012 0.015 0.018

Jet Slot Height (inc h)

Fig. 10 Drag coefficient vs jet slot height V ,

=

94.3 fps).




569

0 .006 0.009 0.012 0.01 5 0.01

20

8

Y

5

+ c p = o .o 4

-Cu = 0.1657

+ p = 0.04

v = 0.1657

L

= 0.001

a

a

i?i

0.0005

Fig.

11

Variations

of

the

LID

characteristics with the jet slot height

V ,

=

94.3 fps .

4

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 coeffi-

cient. 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 sum mary, a smaller je t slot height is preferred from an aerodynamic design

perspective. However, as previously mentioned, a large r jet slot height is pre-

ferred 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

Du ring the past five years, there has been increased interest in the use of pulsed

jets, and “massless” synthetic jets for flow control and performance enhance-

ment. Wygnansky and colleagues28929 tudied the effects of eriodic excitation

on the control of separation and static stall. Lorb er et al e o and W ak e and

Lurie31 have studied the use of directed synthetic jet s fo r 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 durin g 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 , Oy ler and Palmer33 experimentally studied

the pulsed blowing of blown flap configurations. More recently, some numerical

simulations em ployin g a pulsed je t hav e also been reported fo r separation control

of high-lift systems,34 and traditional rounded T E

CC

airfoils with multiport

blowing.35 Most of the studies abo ve were focused on the use of low mom entum

coefficients or zero-mass blowings to control the boundary layer separation or

static and dyn am ic stall. Only a fe w studies33 considered the use of pulsed jets

for lift augmentation, at smaller mass flow rates co mp ared to steady jets.

In earlier work,17 it has been sho wn that the pulsed jet w ith square-wave

form is more efficient than the traditional sinusoidal form, and that the square-

wave-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 m om entum coefficient; and Fig. 14 the variation with the average

mass flow rate.

At first glance, Figs.

13

and 14 app ear to show opposite trends. Figure 14

appe ars 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 appe ars to show the opposite

trend-the steady je t appears to be always more efficient than a pulsed je t,



3

2.5

2

1.5

1

0.5

0

-


571

-Steady Jet

-Pu lsed Je t , f = 40 Hz

- . Puls ed Jet, f = 120 Hz

~

0.02 0.04 0.06 0.08 0.1 0.12

I

Time-Averaged Mom entum Coef fic ient , CpO

Fig.

13

Incremental lift coefficient

vs

time-averaged mom entum coefficient.

Pulsed J et ,

f

= 40

Hz

Puls ed Jet, f = 120 Hz

Pulsed Jet,

f

= 400 Hz

14

0

0.0002 0.0004 0.0006 0.0008 0.001 0.0012 0.0014 0.0016

Time Averaged Mass Flow Rate (siu g kec )

Fig.

14

Incremental lift coefficient vs tim e-averaged mass

flow

rate.



572

YI LIU ET AL.

u

c” 1.2-

.-

E

8

0.8-

E

u

and produces a large

ACl.

To resolve this “apparent” inconsistency between

Figs. 13 and 14, four points A , B, C,

D

are show n 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 com pare d to a steady jet. T his is consistent w ith 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 exam ple,

as shown in Fig. 15, when the frequency is equ al to 400 Hz, the pulsed je t 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, a s a conse-

quence, the pulsed jet will have all the benefits of the steady jet at considerably

lower mass flow rates.

+Pulsed Jet, Ave. C,=0.04

-Steady Jet, C,=0.04

1

0 4 ” ’ ’ ’ 8 8 8 8 8 8 8 8 8 8 8 8 8 8 1

0

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

I I I I

0

1.414 2.828

Strouhal Number f Chord /

Vinf)

Fig.

15

Tim e-averaged lift coefficient vs frequency and Strouhal number.




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

f i e f

VC

St =

7)

In the present study, for the baseline case, Gef s 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 par-

ameters 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 fre-

quency. 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 Tab le 3, the S trouhal number is fixed at 1.41 and the freestream vel-

ocity 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 free-

stream 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 Tab les 2 and

3,

it is seen that the com puted 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 pro-

vided the S trouhal number i s fixed. Tab le 1, on the other hand, show s 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-averaged lift coefficient for the case where

U,

and

LreP

re

fixed, and the Strouhal number is varied with the frequency

Baseline Half frequency Double frequency

Frequency, z

200

100

400

Freestream velocity U,, fps

94.3

94.3 94.3

Chord of the Airfoil befn.

8 8 8

Strouhal number

1.41

0.705 2.82

Computed average lift 1.6804 1.5790 1.8026

coefficient CJ



574 YI LIU ET AL.

Table 2 Com puted time-averaged lift coefficient for t he case where S trou ha l

number and

L,f

ar e fixed, an d

U ,

an d the frequency a re varied

Baseline Half velocity Double velocity

Frequency, Hz

200 100 400

Freestream velocity U,, fps 94.3 47.15 118.6

Chord of the airfoil

Gef,

n.

8 8 8

Strouhal number 1.41 1.41 1.41

Computed average 1.6804 1.6601 1.7112

lift coefficient, Cl

Table

3

Com puted time-averaged lift coefficient for t he case where Str ouh al

number and U , fixed, and Lrefan d frequency a re varied

Baseline Double chord Half chord

Frequency, Hz

200 100 400

Freestream velocity U,, fps 94.3 94.3 94.3

Chord of the airfoil Lef,n. 8 16 4

Strouhal number 1.41 1.41 1.41

Computed average lift coefficient,

Cl

1.6804 1.7016 1.6743

. . A ..

1.24

50 100 150 200 250 300 350 400 450

Frequency

Fig. 16 Time-averaged lift coefficient vs frequency: Case 1: Strouhal numb er 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; Lrefnot fixed.




575

more dominant effect on the average lift coefficient of the pulsed jet than just

the frequency.

IV.

Conclusions

Th e Navier-Stokes simulations are used to model flow over the CCW con-

figurations 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 a ttack. It can also eliminate vortex shedding in the T E region, a potential

noise source. The lift coefficient of the CC airfoil is also increased with angle

of attack like the conven tional sharp T E airfoil. H owever, 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 sup-

pressed 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 mom entum coefficient, the performance of the C C 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 fo r a given

C

This may increase the power require-

men ts 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 je t than just the frequency. Thus, the same performance of a pulsed je t

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, G rant-NAG 1-2146.

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 F ields 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 Cir-

culation 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 Contro l Airfoils,”


’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.

“Eng lar, R. J., and Huson, G. G., “Developm ent of Advanced Circulation Contro l 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 /CEA S 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, AlA A 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, G A, 2003.

”Liu,

Y.,

Sankar, L. N., Englar, R. J., and Ahuja, K. K ., “Num erical Sim ulations 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-Com ponent Navier-Stokes

Equations by Generalized Implicit Method,”

Journal

of

Computational Physics,

Vol. 24, 1977, p. 372.

”Kwon, J., and Sankar, L. N., “Num erical Study of the Effects of Icing on Finite Wing

Aerodynamics,” AIAA Paper 90-0757, Jan. 1990.




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 A llmaras,

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., “N umerical 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 S tall 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.

330y le r, 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 F low w ith 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. rmy Research Laboratory, Aberdeen Proving Ground, Maryland

Nomenclature

D = drag force, N

d =

reference diameter, m

f =

et 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 = je t velocity, m /s

W = vector of conservative variables

y+ = normal viscous sublayer spacing

x y,

z

= axial, normal (vertical), and horizontal axes

a

=


*Aerospace Engineer. Associate Fellow

AIAA.


of

the

U.S.



579



580 J. SAHU

I. Introduction

determination of aerodynamics is critical to the low-cost

A

evelopment 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-correct-

ing, gun-launched projectiles for infantry soldiers.

Because of the sm all projectile diameter d = 0.02 to 0.04m), 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 0.3mm) synthetic microjet actuators have been shown successfully to modify

subsonic flow characteristics and pressure distributions for simple airfoils and

cylinders.394 he synthetic jets (fluid being pumped in and out of the jet cavity

at a high frequency of the order 1000Hz) are control devices (Fig. 1 with

zero net mass flux and are intended to produce the desired con trol of the flowfield

through mom entum effects. M any parameters such as jet location, jet velocity,

and je t actuator frequency, can affect the flow control phenom enon. 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 p ractical 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 CF D techniques are developed and applied for the design and

analysis of these microadaptive flow control systems for steering a spinning pro-

jectile for infantry operations.

CCURATE

Pulsat ing

Synth et ic Jet

Diaphragm

Fig.

1

Schematic

of

a synthetic jet.



FLOW CONTROL ON SPINNING PROJECTILE

581

The control of the trajectory of a 40 m m spinning projectile is achieved

by altering the pressure distribution on the projectile through forced asymmetric

flow separation. Unsteady or time-accurate C FD modeling capabilities are devel-

oped and used to assist in the design of the projectile shape, the placement of the

synthetic actuators, and the prediction of the aerodynam ic force and mom ents 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 characte rize the unsteady nature of the syn thetic je t interaction flow-

field 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 Navier-

Stokes (RANS) approach. RAN S 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 turbu-

lent 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 m ajor challenge.

Both RANS and hybrid RA NS/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-energy-

containing 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 syn-

thetic 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 com-

puted results obtained.

11 Computational M ethodology

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 turbu-

lence models were used for the computation of turbulent flows. The hybrid

RA NS/L ES approach based on limited numerical scales (LNS)6 is well suited

to the simulation of unsteady flows and contains no additional empirical con-

stants 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 je t, the rest of the flowfield being occu-

pied by a coarser, RANS-like mesh. Dual time-stepping was used to achieve the

desired time accuracy. In addition, special jet boundary conditions were devel-

oped and used for numerical modeling of synthetic jets. The grid was actually

moved to take into account the spinning motion of the projectile.

A.

Unsteady Jet Boundary Conditions

One particular boundary condition (BC ) used in the present simulations of the

unsteady jet s is an “oscillating jet” BC . In its basic fo rm, 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@), whe ref is the frequency of the oscillation. Thus, the oscil-

lating 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 je t is on. The user also provides a repetition




583

time period (e.g., the time period corresponding to one spin rotation of the pro-

jectile). Within each time period, therefore, there are sets of start and end times

that define when the jet is on. During “off”periods, the am plitude 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 ensem -

ble-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 large-

eddy 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

scales has been recently developed by Metacom p Technologies.’ T his 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 numeri-

cally 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 auto-

matically 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

k



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 fo r accurate simulation of projectile flowfield in the presence of

unsteady jets. The first step here was to obtain a converged solution for the pro-

jectile without the jet . The converged jet-off solution was then used as the starting

condition fo r the computation of time-accurate unsteady flowfield for the projec-

tile with synthetic jet s. The je t locations on the projectile are show n in Fig.

3.

The

je t cond itions were specified at the exit of the je t for the unsteady (sinusoidal vari-

ation 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 = 0 to 4 deg. The jet width was 0.32 mm, the jet slot half-

angle was 18 deg, and the absolute peak jet velocities used were 3 1 and 6 9 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 je t 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 C FD ++ code' was designed in

the beginning to be able to run on a wide variety of hardware platforms and

Fig. 2 Projectile geometry.




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 aerody-

namic coefficients on both a nonspinning and a spinning projectile. Limited

experimental data (from Ref. 10and private communication with J. McMichael,

GTRI) exist only fo r 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 CF D++ code at sub-

sonic Mach numbers,

M =

0.11 and 0.24, and at angles of attack

= 0

4

deg. The preconditioned version of the CFD++ cod e was used to obtain an effi-

cient numerical solution at low speeds. For modeling of the unsteady synthetic

jets, both unsteady RANS and a hybrid RA NS /LE S 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 , = 7 m/s) and several angles of attack from 0 to 4 deg

using both the unsteady

RANS

and the hybrid R AN S/LE S approaches. The syn-

thetic jets are on all the time for these nonspinning cases. These three-dimen-

sional 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 je t operations, were saved at regular inter-

mittent 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 je t 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 vorti-

city (evident in computed pressure contours, Fig.

8).

Verification of this con-

clusion 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 flow-

fields resulting from the synthetic jet flow control.




587

Fig. 5 Velocity magn itudes,M = 0.11, Y = 0 deg.

The net lift force F,) was determined by time-averaging the actual time his-

tories of the highly unsteady lift force (an example show n in Fig. 10 for various

ang les of attack) resulting from the jet interaction at zero-degree ang le 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

x =

0

deg.

shows both low- and high-frequency oscillations in the predicted lift force at

different angles of attack, = 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.




u

c

0 6

Q)

Q)

t

2

0.4

it:

.-

L

Q

5

0

589

-

w EXPERIMENT

+CFD

Hybrid

RANSILES)

CFD RANS)

-

w

o.2A-

w w

o.8

time-histories result from the interaction

of

the jet with wake and the resulting

unsteady wake flowfields.

B.

Spinning Projectile

Of m ore interest is the spinning projectile case for the real-world applications.

Num erical computations have been made in this case for actual flight condition at

Time ms)

Fig.

10

Time-histories of computed lift force at angles of attack cu = 0 2 and 4 deg,

hybrid RANS/LES model, M

=

0.11.



590 J. SAHU

Y

t

Jet-on1

a

=3.73ms

t =O .............

ypfa

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 = 0 deg, and a spin rate of

67 Hz. The atm ospheric 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 et oper-

ated 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 pro-

cessor-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 oper-

ations 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.

Com puted 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 projec tile rotates coun-

terclockwise (looking from the back of the projectile). The particle traces ema-

nating 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 RAN S as

well as the hybrid RA NS /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, j

=

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,

= 0 deg.

Time ms)

Fig. 13 Computed lift and side forces, unsteady RANS,

M =

0.24, vj = 69 m /s,

= 0

deg.



592 J. SAHU

unsteady nature of the flowfield. When the je t 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 je t 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 corre-

sponding 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.




593

0.4

0.3

0.2

t

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 =

0

deg,

P

= 67 Hz.

the syn thetic jets. The C FD results are plotted over only one spin cycle; each sub-

sequent 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 (three-

fourths) 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 pproximately.

Figure 15 shows the computed lift force, again averaged over many spin

0 4

0.3

j

0.2

l

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, =

0

deg,

P

= 67

Hz.



594 J. SAHU

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

=

0 deg, P = 67

Hz.

cycles (10 ,2 0, 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 show s the com puted time-averaged lift force after 50

and 60 spin cycles for jet velocities 3 1 and 69 m/s, respectively. It clearly show s

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 aero-

dynam ic effect of tiny synthetic jets as a means to provide the control authority

needed to maneuver a projectile at low subsonic speeds. Com puted results have

been obtained for a subsonic projectile for both nonspinning and spinning cases

using a time-accurate Navier- Stokes computational technique and advanced




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 show n that microadaptive flow control with tiny syn-

thetic jets can provide an affordable route to lethal precision-guided infantry

weapons.

References

‘Sahu ., Heavey, K. R., and Ferry, E. N., “Computational Fluid Dynamics for Multiple

Projectile Configurations”, Proceedings of the 3rd Overset Com posite Grid and Solution

Technology Symposium,

Oct. 1996.

’Sahu

., Heavey,

K.

R., and Nietubicz, C. J., “Time-D ependent Navier-Stokes Com-

putations for Submunitions in Relative Motion,”

6th International Sym posium on Compu-

tational 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., K ibens, V., Parekh, D., and Glezer, A., “The Dynamics of Flow Reattach-

ment

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,


’Pulliam, T. H., and Steger, J. L., “On Implicit Finite-Difference Simulations of Three-

Dimensional 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,

‘‘finehart, C., McM ichael, J.

M.,

and Glezer, A., “Synthetic Jet-Based Lift Generation

“Sahu, J., “Unsteady Numerical Sim ulations of Subsonic Flow over a Projectile with Jet

V O ~ .20

1998,

pp. 457-462.

and Circulation Control on hisymmetric Bodies,” AIAA Paper 2002-3 168 June 2002.

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 TH E “Coanda Effect/CC Workshop in Hampton, Virginia (March

16- 17,

A 004)”’

the question was posed, “What are the roadblocks to further devel-

opment?” Those roadblocks may be a result of a failure to address certain

deficiencies or an inability to find solutions. Exam ples of operationa l deficiencies

are insufficient quantity of CC air, heavy, com plicated air pumps, heavy, energy-

wasting 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 pum p

and a novel fan to supply C C air.

These projects are proposed com mercial 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.

contain ade-

quate 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 appli-

cations to novel nonaeronautical situations.

The overview papers in this book and other available

*Consultant.

Copyright

005

by Terence

R.

Day. Published by the American Institute of Aeronautics and


599



6

T.

R

D Y

11. Applications

Oscillating Channel Flow Including Self Oscillating Channel Flow

.

Coanda Effect)

Although this phenom enon has been understood for quite some time,4 it appar-

ently has been a curiosity with little vision for many useful applications. The geo-

metry of a rectangular channel that enables jet self-oscillating flow must be

relatively precise to work at all. Gas jets in a channel w ill oscillate by imposition

of a pressure change alternating either side of the jet . W ith 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 turbu-

lence. A second reason for employing oscillating channel flow is that as the je t

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 ca n 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 pro-

prietary, 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.



NON ERON UTIC L PPLIC TIONS

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

je t to remain attached to one wall longer than on the opposite wall.

A useful feature is that as the je t oscillates, one side may b e routed through a

labyrinthine pathway with walls coated with a che mical that may possess a large

surface area for longer interaction time and then returned to the inflow to the fan.

Makeu p 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 je t entrains dirt and then enters an annular suction slot. The a ir ascends,



602

T.

R

D Y

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 do es not collapse onto the low-pressure vortex b ecause an even

lower pressure is generated between the bag and the bowl. The vortex flows

inwardly t o form a centr al vortex, which then return s through the fan to recircu-

late. In this way most of the air is recirculated, minim izing the quantity of dirt

needing removal by a filter. Some nondomestic versions need no filter.

Som e other features are proprietary. Th e Coanda effect and the sim ple, 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.




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 verti-

cally 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 o r 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 C oan da surface, especially when the air speed i s rela-

tively 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 je t is the working fluid and that sam e 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. Th e body can be translucent with a

circular fluorescent tube inside. Excellent Coanda mixing enhances air-

conditioned 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

D Y

Fig. 6 Smoke visualizes

flow.

following results are for the Jetfan water pump performance and demonstrate the

unique characteristics of the impelle r com pared to other pum p impellers; they are

highly indicative of similar characteristics for the fan version, that is, no stall

wi thout s tators or a di ff~ser .~

“The visual inspection of the onset of cavitation ind icated that over the range

of flow rates tested, cavitation first appears at a rotational speed of 33 rpm.

Above this speed, cavitation bubbles were observed to fo rm 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 reabsorp tion corresponds to a

Fig. 7 Injection-moldedJetfan.




605

line drawn at right angles rom the LE

o

the convex su ace

o

the adjacent

blade. This reabsorption indicates that the pressure is rising as the water

enters the blade passage.

In

comparing the pe ormance

o

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

o

a complex volute or stator blades, which are

com-

monly used to direct the ow

o

water rom the rotating impeller into the outlet

pipe in many pu mp designs.”

These fans and w ater pumps are useful in producing fan o r pump flows of suf-

ficient 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” charac teristic. 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 o r 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 ship-

ping, 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 m ade 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.



6 6

T.

R

D Y

Fig. 9 Jetfan water pumps.

to be made with high-performance geometry and with the ability to be rotation-

ally ex tracted from the m ould instead of employing investment casting and sub-

sequent milling for precision.

G.

Wind Turbines and Orbital

Pump

Full-span and tip b lowing6 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, includ-

ing 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 i s another good example, but CC there is not, strictly speaking, critical

to the aircraft performance. Many proposed applications, including some suc-

cessfully achieved, are risky, because o ther aircraft systems may be com promised

generally or occasionally. If the only reason that CC development has stagnated

Fig. 10 Manufacturing tooling.




607

Fig.

11

Axial fan rotational extraction from mould, enab ling blade overlap .

is that of insufficient control air, then CC application to w ind 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 con trol 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 high-

volume, relatively low-speed, low-noise, low-wear, fills and exhausts simul-

taneously, and can function as either a compressor or high-volume air pump, or both.

The O rbitalpump 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

D Y

Fig.

13

Nylon O rbitalpump.

It is likely that shifting the rear stagnation point and attenuating o r elimina ting

tip vortices in wind turbines and fan blades is as valid as it is for aircraft wings.

Applying C C to wind turbines may have o ther benefits. A smaller diameter wind

turbine may achieve the sa me efficiency as a larger one. This would reduce man-

ufacturing 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

W IG 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.




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 C C 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. Sm all models of 2 ft in diameter cannot carry a com pres-

sor 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 a ir 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 fol-

lowing 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 em ployed 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 dram atic 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.



61

T.

R

D Y

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.




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 under-

water 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 O rbitalpump will fill that need.

Fig.

20 View of TE slot.



612

T.

R

D Y

111. Conclusions

There are many o ther 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. Th is 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 develop-

ing 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 A utomotive 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/ON R Circulation Control Work-

Englar, R. J., “Development Of The A-G/Circulation Control Wing Flight Demon-

stration 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,” 1 th 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,

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 S urface Tip Vortex Attenuation System,” US Patent N o.

5,158,251, Oct. 27, 1992.

shop, NASA CP 2005-213509, M ar. 2005.

2

pp. 1243-1247.




61

3

’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 Archi-

tecture), 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. 245 469

Angle II, G. 277

B

Baker, W. J. 421 513

Blaylock, G. 383

C

Campbell, B. A. 315Cerchie, D. 113

Chang III, P. A. 445

D

Day, T. R. 599

E

Ebert, M. P. 445

http://77898_03.pdf/

http://77898_17.pdf/

http://77898_06.pdf/

http://77898_22.pdf/

http://77898_08.pdf/

http://77898_08.pdf/

http://77898_18.pdf/

http://77898_09.pdf/

http://77898_16.pdf/

http://77898_20.pdf/

http://77898_14.pdf/

http://77898_11.pdf/

http://77898_05a.pdf/

http://77898_17.pdf/

http://77898_24.pdf/

http://77898_17.pdf/

http://77898_17.pdf/

http://77898_24.pdf/

http://77898_17.pdf/

http://77898_05a.pdf/

http://77898_11.pdf/

http://77898_14.pdf/

http://77898_20.pdf/

http://77898_16.pdf/

http://77898_09.pdf/

http://77898_18.pdf/

http://77898_08.pdf/

http://77898_08.pdf/

http://77898_22.pdf/

http://77898_06.pdf/

http://77898_17.pdf/

http://77898_03.pdf/



Index Terms Links

Englar, R. J. 23 167 315 357

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. 499Huebsch, W. 277

I

Imber, R. 69

J

Johnson, S. K. 245

Jones, G. S. 191

http://77898_02a.pdf/

http://77898_06.pdf/

http://77898_11.pdf/

http://77898_13.pdf/

http://77898_14.pdf/

http://77898_22.pdf/

http://77898_15.pdf/

http://77898_12.pdf/

http://77898_14.pdf/

http://77898_22.pdf/

http://77898_19.pdf/

http://77898_21.pdf/

http://77898_05a.pdf/

http://77898_05a.pdf/

http://77898_05a.pdf/

http://77898_05a.pdf/

http://77898_09.pdf/

http://77898_19.pdf/

http://77898_09.pdf/

http://77898_03.pdf/

http://77898_08.pdf/

http://77898_07a.pdf/

http://77898_07a.pdf/

http://77898_08.pdf/

http://77898_03.pdf/

http://77898_09.pdf/

http://77898_19.pdf/

http://77898_09.pdf/

http://77898_05a.pdf/

http://77898_05a.pdf/

http://77898_05a.pdf/

http://77898_05a.pdf/

http://77898_21.pdf/

http://77898_19.pdf/

http://77898_22.pdf/

http://77898_14.pdf/

http://77898_12.pdf/

http://77898_15.pdf/

http://77898_22.pdf/

http://77898_14.pdf/

http://77898_13.pdf/

http://77898_11.pdf/

http://77898_06.pdf/

http://77898_02a.pdf/



Index Terms Links

L

Liu, Y. 557

Loth, J. L. 3

LutzTaubert 113

M

Marino, T. 445

McGowan, G. 499 539

Munro, S. E. 167

O

Owen, F. K. 105

Owen, A. K. 105

P

Paterson, E. G. 421 513

Paxton, C. D. 293

R

Rogers, E. 69

Rumsey, C. L. 469

S

Sahu, J. 579

Sankar, L. N. 567

http://77898_22.pdf/

http://77898_01.pdf/

http://77898_05a.pdf/

http://77898_17.pdf/

http://77898_19.pdf/

http://77898_21.pdf/

http://77898_06.pdf/

http://77898_04.pdf/

http://77898_04.pdf/

http://77898_16.pdf/

http://77898_20.pdf/

http://77898_10.pdf/

http://77898_03.pdf/

http://77898_18.pdf/

http://77898_23.pdf/

http://77898_22.pdf/

http://77898_22.pdf/

http://77898_23.pdf/

http://77898_18.pdf/

http://77898_03.pdf/

http://77898_10.pdf/

http://77898_20.pdf/

http://77898_16.pdf/

http://77898_04.pdf/

http://77898_04.pdf/

http://77898_06.pdf/

http://77898_21.pdf/

http://77898_19.pdf/

http://77898_17.pdf/

http://77898_05a.pdf/

http://77898_01.pdf/

http://77898_22.pdf/



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

http://77898_17.pdf/

http://77898_09.pdf/

http://77898_18.pdf/

http://77898_05a.pdf/

http://77898_05a.pdf/

http://77898_15.pdf/

http://77898_12.pdf/

http://77898_05a.pdf/

http://77898_19.pdf/

http://77898_10.pdf/

http://77898_10.pdf/

http://77898_19.pdf/

http://77898_05a.pdf/

http://77898_12.pdf/

http://77898_15.pdf/

http://77898_05a.pdf/

http://77898_05a.pdf/

http://77898_18.pdf/

http://77898_09.pdf/

http://77898_17.pdf/



INDEX

Index Terms Links

A

Acoustic optimization, noise reduction

and 174

Active flow control (AFC) 403

Advanced CCW airfoils 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

http://77898_06.pdf/

http://77898_15.pdf/

http://77898_02a.pdf/

http://77898_02a.pdf/

http://77898_02a.pdf/

http://77898_14.pdf/

http://77898_14.pdf/

http://77898_14.pdf/

http://77898_14.pdf/

http://77898_14.pdf/

http://77898_14.pdf/

http://77898_14.pdf/

http://77898_02a.pdf/

http://77898_02a.pdf/

http://77898_02a.pdf/

http://77898_02a.pdf/

http://77898_14.pdf/

http://77898_14.pdf/

http://77898_14.pdf/

http://77898_14.pdf/

http://77898_14.pdf/

http://77898_14.pdf/

http://77898_14.pdf/

http://77898_02a.pdf/

http://77898_02a.pdf/

http://77898_02a.pdf/

http://77898_15.pdf/

http://77898_06.pdf/



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

http://77898_07b.pdf/

http://77898_07a.pdf/

http://77898_09.pdf/

http://77898_04.pdf/

http://77898_04.pdf/

http://77898_04.pdf/

http://77898_04.pdf/

http://77898_18.pdf/

http://77898_04.pdf/

http://77898_10.pdf/

http://77898_05a.pdf/

http://77898_05b.pdf/

http://77898_07a.pdf/

http://77898_05a.pdf/

http://77898_04.pdf/

http://77898_04.pdf/

http://77898_06.pdf/

http://77898_03.pdf/

http://77898_03.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_03.pdf/

http://77898_03.pdf/

http://77898_06.pdf/

http://77898_04.pdf/

http://77898_04.pdf/

http://77898_05a.pdf/

http://77898_07a.pdf/

http://77898_05b.pdf/

http://77898_05a.pdf/

http://77898_10.pdf/

http://77898_04.pdf/

http://77898_18.pdf/

http://77898_04.pdf/

http://77898_04.pdf/

http://77898_04.pdf/

http://77898_04.pdf/

http://77898_09.pdf/

http://77898_07a.pdf/

http://77898_07b.pdf/



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 525

Blowing momentum 110

Blowing, boundary layer control,

circulation control 115

Blown airfoils, two-dimensional drag 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

http://77898_07a.pdf/

http://77898_09.pdf/

http://77898_09.pdf/

http://77898_09.pdf/

http://77898_09.pdf/

http://77898_09.pdf/

http://77898_20.pdf/

http://77898_04.pdf/

http://77898_05a.pdf/

http://77898_07a.pdf/

http://77898_07a.pdf/

http://77898_18.pdf/

http://77898_21.pdf/

http://77898_22.pdf/

http://77898_01.pdf/

http://77898_01.pdf/

http://77898_22.pdf/

http://77898_21.pdf/

http://77898_18.pdf/

http://77898_07a.pdf/

http://77898_07a.pdf/

http://77898_05a.pdf/

http://77898_04.pdf/

http://77898_20.pdf/

http://77898_09.pdf/

http://77898_09.pdf/

http://77898_09.pdf/

http://77898_09.pdf/

http://77898_09.pdf/

http://77898_07a.pdf/



Index Terms Links

Boundary layer control, suction,

circulation control (CC) high lift

generation 3

history of 4

C

Cavitation 440

CC propeller 53

CC. See circulation control.

CC / jet deflection 51

CC-disc 85

CC-valve 91

CCW airfoils, advanced 40CCW. 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 48

CFD techniques 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

http://77898_01.pdf/

http://77898_01.pdf/

http://77898_16.pdf/

http://77898_02b.pdf/

http://77898_02b.pdf/

http://77898_03.pdf/

http://77898_03.pdf/

http://77898_02a.pdf/

http://77898_02a.pdf/

http://77898_02a.pdf/

http://77898_11.pdf/

http://77898_02b.pdf/

http://77898_02a.pdf/

http://77898_11.pdf/

http://77898_07b.pdf/

http://77898_15.pdf/

http://77898_15.pdf/

http://77898_07b.pdf/

http://77898_11.pdf/

http://77898_02a.pdf/

http://77898_02b.pdf/

http://77898_11.pdf/

http://77898_02a.pdf/

http://77898_02a.pdf/

http://77898_02a.pdf/

http://77898_03.pdf/

http://77898_03.pdf/

http://77898_02b.pdf/

http://77898_02b.pdf/

http://77898_16.pdf/

http://77898_01.pdf/

http://77898_01.pdf/



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) 85

specifications 86

Circulation control

aerodynamic heat exchanger (AHE) 383

airfoil

computational fluid dynamics

(CFD) 106

concepts 106

development, CFD techniques 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

http://77898_02a.pdf/

http://77898_05b.pdf/

http://77898_15.pdf/

http://77898_15.pdf/

http://77898_07b.pdf/

http://77898_03.pdf/

http://77898_03.pdf/

http://77898_14.pdf/

http://77898_04.pdf/

http://77898_04.pdf/

http://77898_02a.pdf/

http://77898_19.pdf/

http://77898_21.pdf/

http://77898_17.pdf/

http://77898_18.pdf/

http://77898_04.pdf/

http://77898_04.pdf/

http://77898_04.pdf/

http://77898_18.pdf/

http://77898_18.pdf/

http://77898_18.pdf/

http://77898_18.pdf/

http://77898_18.pdf/

http://77898_18.pdf/

http://77898_04.pdf/

http://77898_04.pdf/

http://77898_04.pdf/

http://77898_18.pdf/

http://77898_17.pdf/

http://77898_21.pdf/

http://77898_19.pdf/

http://77898_02a.pdf/

http://77898_04.pdf/

http://77898_04.pdf/

http://77898_14.pdf/

http://77898_03.pdf/

http://77898_03.pdf/

http://77898_07b.pdf/

http://77898_15.pdf/

http://77898_15.pdf/

http://77898_05b.pdf/

http://77898_02a.pdf/



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

http://77898_18.pdf/

http://77898_18.pdf/

http://77898_18.pdf/

http://77898_07a.pdf/

http://77898_07b.pdf/

http://77898_07a.pdf/

http://77898_22.pdf/

http://77898_08.pdf/

http://77898_08.pdf/

http://77898_08.pdf/

http://77898_08.pdf/

http://77898_08.pdf/

http://77898_08.pdf/

http://77898_15.pdf/

http://77898_04.pdf/

http://77898_04.pdf/

http://77898_01.pdf/

http://77898_04.pdf/

http://77898_05b.pdf/

http://77898_10.pdf/

http://77898_01.pdf/

http://77898_01.pdf/

http://77898_05a.pdf/

http://77898_05b.pdf/

http://77898_05a.pdf/

http://77898_05a.pdf/

http://77898_05a.pdf/

http://77898_05a.pdf/

http://77898_05b.pdf/

http://77898_05a.pdf/

http://77898_01.pdf/

http://77898_01.pdf/

http://77898_10.pdf/

http://77898_05b.pdf/

http://77898_04.pdf/

http://77898_01.pdf/

http://77898_04.pdf/

http://77898_04.pdf/

http://77898_15.pdf/

http://77898_08.pdf/

http://77898_08.pdf/

http://77898_08.pdf/

http://77898_08.pdf/

http://77898_08.pdf/

http://77898_08.pdf/

http://77898_22.pdf/

http://77898_07a.pdf/

http://77898_07b.pdf/

http://77898_07a.pdf/

http://77898_18.pdf/

http://77898_18.pdf/

http://77898_18.pdf/



Index Terms Links


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

http://77898_12.pdf/

http://77898_12.pdf/

http://77898_12.pdf/

http://77898_01.pdf/

http://77898_01.pdf/

http://77898_01.pdf/

http://77898_01.pdf/

http://77898_06.pdf/

http://77898_24.pdf/

http://77898_24.pdf/

http://77898_24.pdf/

http://77898_24.pdf/

http://77898_02a.pdf/

http://77898_02a.pdf/

http://77898_02a.pdf/

http://77898_02a.pdf/

http://77898_02b.pdf/

http://77898_02a.pdf/

http://77898_02a.pdf/

http://77898_02a.pdf/

http://77898_02a.pdf/

http://77898_02a.pdf/

http://77898_02a.pdf/

http://77898_02b.pdf/

http://77898_02b.pdf/

http://77898_02a.pdf/

http://77898_02a.pdf/

http://77898_02a.pdf/

http://77898_02a.pdf/

http://77898_02a.pdf/

http://77898_02a.pdf/

http://77898_02b.pdf/

http://77898_02a.pdf/

http://77898_02a.pdf/

http://77898_02a.pdf/

http://77898_02a.pdf/

http://77898_24.pdf/

http://77898_24.pdf/

http://77898_24.pdf/

http://77898_24.pdf/

http://77898_06.pdf/

http://77898_01.pdf/

http://77898_01.pdf/

http://77898_01.pdf/

http://77898_01.pdf/

http://77898_12.pdf/

http://77898_12.pdf/

http://77898_12.pdf/



Index Terms Links


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

http://77898_02b.pdf/

http://77898_02b.pdf/

http://77898_02b.pdf/

http://77898_02b.pdf/

http://77898_02b.pdf/

http://77898_02b.pdf/

http://77898_02a.pdf/

http://77898_01.pdf/

http://77898_01.pdf/

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_03.pdf/

http://77898_03.pdf/

http://77898_03.pdf/

http://77898_03.pdf/

http://77898_03.pdf/

http://77898_03.pdf/

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_01.pdf/

http://77898_01.pdf/

http://77898_02a.pdf/

http://77898_02b.pdf/

http://77898_02b.pdf/

http://77898_02b.pdf/

http://77898_02b.pdf/

http://77898_02b.pdf/

http://77898_02b.pdf/



Index Terms Links


dual-slotted low aspect ratio wing

(CC hydrofoil) 93

exploratory investigations,

NSWCCD 69

miniature oscillatory valve

(CC-valve) 91

self-driven rotary thruster

(TIPJET) 73

wings, (CCW) 36

conventional wings, noise reduction

comparison 182

demonstrator design 5

noise reduction, experiments 168

Coanda, ceiling fan 603

Coanda, device 26

Coanda effect 25 278

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

http://77898_03.pdf/

http://77898_03.pdf/

http://77898_03.pdf/

http://77898_03.pdf/

http://77898_02a.pdf/

http://77898_06.pdf/

http://77898_01.pdf/

http://77898_06.pdf/

http://77898_24.pdf/

http://77898_02a.pdf/

http://77898_02a.pdf/

http://77898_09.pdf/

http://77898_09.pdf/

http://77898_09.pdf/

http://77898_09.pdf/

http://77898_09.pdf/

http://77898_09.pdf/

http://77898_24.pdf/

http://77898_24.pdf/

http://77898_24.pdf/

http://77898_24.pdf/

http://77898_24.pdf/

http://77898_24.pdf/

http://77898_24.pdf/

http://77898_24.pdf/

http://77898_09.pdf/

http://77898_09.pdf/

http://77898_09.pdf/

http://77898_09.pdf/

http://77898_09.pdf/

http://77898_09.pdf/

http://77898_02a.pdf/

http://77898_02a.pdf/

http://77898_24.pdf/

http://77898_06.pdf/

http://77898_01.pdf/

http://77898_06.pdf/

http://77898_02a.pdf/

http://77898_03.pdf/

http://77898_03.pdf/

http://77898_03.pdf/

http://77898_03.pdf/



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) 421

computational methods, unsteady RANS 425

NCCR airfoil


grid generation 427

http://77898_24.pdf/

http://77898_07a.pdf/

http://77898_10.pdf/

http://77898_10.pdf/

http://77898_10.pdf/

http://77898_10.pdf/

http://77898_10.pdf/

http://77898_10.pdf/

http://77898_04.pdf/

http://77898_05a.pdf/

http://77898_06.pdf/

http://77898_07b.pdf/

http://77898_07b.pdf/

http://77898_07b.pdf/

http://77898_11.pdf/

http://77898_16.pdf/

http://77898_16.pdf/

http://77898_16.pdf/

http://77898_16.pdf/

http://77898_16.pdf/

http://77898_16.pdf/

http://77898_16.pdf/

http://77898_16.pdf/

http://77898_11.pdf/

http://77898_07b.pdf/

http://77898_07b.pdf/

http://77898_07b.pdf/

http://77898_06.pdf/

http://77898_05a.pdf/

http://77898_04.pdf/

http://77898_10.pdf/

http://77898_10.pdf/

http://77898_10.pdf/

http://77898_10.pdf/

http://77898_10.pdf/

http://77898_10.pdf/

http://77898_07a.pdf/

http://77898_24.pdf/



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 404


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 28

Elliptical airfoil flow 4 144

Equal lift condition 182

Equivalent drag 201

http://77898_16.pdf/

http://77898_16.pdf/

http://77898_16.pdf/

http://77898_16.pdf/

http://77898_16.pdf/

http://77898_15.pdf/

http://77898_15.pdf/

http://77898_15.pdf/

http://77898_15.pdf/

http://77898_13.pdf/

http://77898_07b.pdf/

http://77898_02a.pdf/

http://77898_03.pdf/

http://77898_03.pdf/

http://77898_02a.pdf/

http://77898_01.pdf/

http://77898_05b.pdf/

http://77898_06.pdf/

http://77898_07a.pdf/

http://77898_07a.pdf/

http://77898_06.pdf/

http://77898_05b.pdf/

http://77898_01.pdf/

http://77898_02a.pdf/

http://77898_03.pdf/

http://77898_03.pdf/

http://77898_02a.pdf/

http://77898_07b.pdf/

http://77898_13.pdf/

http://77898_15.pdf/

http://77898_15.pdf/

http://77898_15.pdf/

http://77898_15.pdf/

http://77898_16.pdf/

http://77898_16.pdf/

http://77898_16.pdf/

http://77898_16.pdf/

http://77898_16.pdf/



Index Terms Links

Exploratory investigations, circulation

control technology workshops,

NSWCCD 69

F

F-5E aircraft, co-flow jet method and 308

Flight control

circulation control

full-span UAV model 345

half-span model 339

wing and 337

Flight testing

circulation control and 12Grumman Aerospace A-6A 16

Flow attachment, STOL aircraft


Flow control, conventional flap airfoil,

circulation control experiments

and 118

Flow prediction, turbulence modeling 499

FLUENT flow solver 539

experiments 541

numerical approach 542


grid details 542

solver settings 543

test results 545

http://77898_03.pdf/

http://77898_10.pdf/

http://77898_12.pdf/

http://77898_12.pdf/

http://77898_12.pdf/

http://77898_01.pdf/

http://77898_01.pdf/

http://77898_11.pdf/

http://77898_05a.pdf/

http://77898_19.pdf/

http://77898_21.pdf/

http://77898_21.pdf/

http://77898_21.pdf/

http://77898_21.pdf/

http://77898_21.pdf/

http://77898_21.pdf/

http://77898_21.pdf/

http://77898_21.pdf/

http://77898_21.pdf/

http://77898_21.pdf/

http://77898_21.pdf/

http://77898_21.pdf/

http://77898_21.pdf/

http://77898_21.pdf/

http://77898_19.pdf/

http://77898_05a.pdf/

http://77898_11.pdf/

http://77898_01.pdf/

http://77898_01.pdf/

http://77898_12.pdf/

http://77898_12.pdf/

http://77898_12.pdf/

http://77898_10.pdf/

http://77898_03.pdf/



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


http://77898_21.pdf/

http://77898_21.pdf/

http://77898_22.pdf/

http://77898_13.pdf/

http://77898_17.pdf/

http://77898_17.pdf/

http://77898_17.pdf/

http://77898_12.pdf/

http://77898_12.pdf/

http://77898_07a.pdf/

http://77898_07a.pdf/

http://77898_07a.pdf/

http://77898_07a.pdf/

http://77898_07a.pdf/

http://77898_07a.pdf/

http://77898_07a.pdf/

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_07a.pdf/

http://77898_07a.pdf/

http://77898_07a.pdf/

http://77898_07a.pdf/

http://77898_07a.pdf/

http://77898_07a.pdf/

http://77898_07a.pdf/

http://77898_12.pdf/

http://77898_12.pdf/

http://77898_17.pdf/

http://77898_17.pdf/

http://77898_17.pdf/

http://77898_13.pdf/

http://77898_22.pdf/

http://77898_21.pdf/

http://77898_21.pdf/



Index Terms Links

GACC (Cont .)


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 518 generation, GACC wing and 518

study, circulation control stimulation

test results and 530

Grumman Aerospace A-6A 16

H

Half-span CCW model, circulation

control flight control 339

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_05a.pdf/

http://77898_23.pdf/

http://77898_22.pdf/

http://77898_21.pdf/

http://77898_16.pdf/

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_01.pdf/

http://77898_12.pdf/

http://77898_12.pdf/

http://77898_01.pdf/

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_16.pdf/

http://77898_21.pdf/

http://77898_22.pdf/

http://77898_23.pdf/

http://77898_05a.pdf/

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_20.pdf/

http://77898_20.pdf/



Index Terms Links

Heat transfer, aerodynamic heat

exchanger and 392

Heavy vehicles (HV), pneumatic

aerodynamic technology 357

pneumatic test results 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

http://77898_14.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_07a.pdf/

http://77898_07b.pdf/

http://77898_07b.pdf/

http://77898_24.pdf/

http://77898_23.pdf/

http://77898_02b.pdf/

http://77898_18.pdf/

http://77898_18.pdf/

http://77898_02b.pdf/

http://77898_23.pdf/

http://77898_24.pdf/

http://77898_07b.pdf/

http://77898_07b.pdf/

http://77898_07a.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_14.pdf/



Index Terms Links

J

Jet momentum coefficient 478

Jet-based flow control 579

computer simulations 581

hybrid RANS / LES turbulence

model 583

unsteady jet boundary conditions 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

http://77898_18.pdf/

http://77898_23.pdf/

http://77898_23.pdf/

http://77898_23.pdf/

http://77898_23.pdf/

http://77898_23.pdf/

http://77898_23.pdf/

http://77898_23.pdf/

http://77898_23.pdf/

http://77898_22.pdf/

http://77898_05a.pdf/

http://77898_15.pdf/

http://77898_22.pdf/

http://77898_03.pdf/

http://77898_03.pdf/

http://77898_22.pdf/

http://77898_15.pdf/

http://77898_05a.pdf/

http://77898_22.pdf/

http://77898_23.pdf/

http://77898_23.pdf/

http://77898_23.pdf/

http://77898_23.pdf/

http://77898_23.pdf/

http://77898_23.pdf/

http://77898_23.pdf/

http://77898_23.pdf/

http://77898_18.pdf/



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 193

NCCR airfoil, detached-eddy simulation (DES) 421


unsteady RANS 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 wings vs.

conventional wings 182

experiments 168

http://77898_07a.pdf/

http://77898_02b.pdf/

http://77898_03.pdf/

http://77898_02b.pdf/

http://77898_05a.pdf/

http://77898_07a.pdf/

http://77898_16.pdf/

http://77898_16.pdf/

http://77898_16.pdf/

http://77898_16.pdf/

http://77898_16.pdf/

http://77898_16.pdf/

http://77898_16.pdf/

http://77898_16.pdf/

http://77898_16.pdf/

http://77898_06.pdf/

http://77898_06.pdf/

http://77898_06.pdf/

http://77898_06.pdf/

http://77898_06.pdf/

http://77898_06.pdf/

http://77898_06.pdf/

http://77898_06.pdf/

http://77898_16.pdf/

http://77898_16.pdf/

http://77898_16.pdf/

http://77898_16.pdf/

http://77898_16.pdf/

http://77898_16.pdf/

http://77898_16.pdf/

http://77898_16.pdf/

http://77898_16.pdf/

http://77898_07a.pdf/

http://77898_05a.pdf/

http://77898_02b.pdf/

http://77898_03.pdf/

http://77898_02b.pdf/

http://77898_07a.pdf/



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


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

http://77898_06.pdf/

http://77898_06.pdf/

http://77898_06.pdf/

http://77898_06.pdf/

http://77898_06.pdf/

http://77898_06.pdf/

http://77898_24.pdf/

http://77898_24.pdf/

http://77898_24.pdf/

http://77898_24.pdf/

http://77898_24.pdf/

http://77898_24.pdf/

http://77898_24.pdf/

http://77898_24.pdf/

http://77898_24.pdf/

http://77898_24.pdf/

http://77898_02b.pdf/

http://77898_23.pdf/

http://77898_23.pdf/

http://77898_02b.pdf/

http://77898_24.pdf/

http://77898_24.pdf/

http://77898_24.pdf/

http://77898_24.pdf/

http://77898_24.pdf/

http://77898_24.pdf/

http://77898_24.pdf/

http://77898_24.pdf/

http://77898_24.pdf/

http://77898_24.pdf/

http://77898_06.pdf/

http://77898_06.pdf/

http://77898_06.pdf/

http://77898_06.pdf/

http://77898_06.pdf/

http://77898_06.pdf/



Index Terms Links

NSWCCD, Naval Surface Warfare

Center, Carderock Division 70

Numerical method, circulation control

airfoils and 475

Numerical simulation 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 600Outboard wing ON, STOL aircraft


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

http://77898_03.pdf/

http://77898_18.pdf/

http://77898_18.pdf/

http://77898_18.pdf/

http://77898_18.pdf/

http://77898_18.pdf/

http://77898_18.pdf/

http://77898_24.pdf/

http://77898_24.pdf/

http://77898_11.pdf/

http://77898_20.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_20.pdf/

http://77898_11.pdf/

http://77898_24.pdf/

http://77898_24.pdf/

http://77898_18.pdf/

http://77898_18.pdf/

http://77898_18.pdf/

http://77898_18.pdf/

http://77898_18.pdf/

http://77898_18.pdf/

http://77898_03.pdf/



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 49

X-wing aircraft 35

Coanda device 26

channel wing (PCW) 52 319

flap performance 193 airfoil performance 216

blown airfoils, two-dimensional

drag 200

equivalent drag 201

experiments 207

Coanda slot setup errors 212

GACC

airfoil design 202

balance limits 208

http://77898_13.pdf/

http://77898_02a.pdf/

http://77898_02b.pdf/

http://77898_02a.pdf/

http://77898_02a.pdf/

http://77898_02a.pdf/

http://77898_02a.pdf/

http://77898_02a.pdf/

http://77898_02a.pdf/

http://77898_02b.pdf/

http://77898_02b.pdf/

http://77898_02a.pdf/

http://77898_02a.pdf/

http://77898_02b.pdf/

http://77898_11.pdf/

http://77898_07a.pdf/

http://77898_07a.pdf/

http://77898_07a.pdf/

http://77898_07a.pdf/

http://77898_07a.pdf/

http://77898_07a.pdf/

http://77898_07a.pdf/

http://77898_07a.pdf/

http://77898_07a.pdf/

http://77898_07a.pdf/

http://77898_07a.pdf/

http://77898_07a.pdf/

http://77898_07a.pdf/

http://77898_07a.pdf/

http://77898_11.pdf/

http://77898_02b.pdf/

http://77898_02a.pdf/

http://77898_02a.pdf/

http://77898_02b.pdf/

http://77898_02b.pdf/

http://77898_02a.pdf/

http://77898_02a.pdf/

http://77898_02a.pdf/

http://77898_02a.pdf/

http://77898_02a.pdf/

http://77898_02a.pdf/

http://77898_02b.pdf/

http://77898_02a.pdf/

http://77898_13.pdf/



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 376

Powered lift and engine thrust deflection 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

http://77898_07a.pdf/

http://77898_07a.pdf/

http://77898_07a.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_11.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_02b.pdf/

http://77898_02b.pdf/

http://77898_02b.pdf/

http://77898_02b.pdf/

http://77898_23.pdf/

http://77898_07b.pdf/

http://77898_22.pdf/

http://77898_22.pdf/

http://77898_07b.pdf/

http://77898_23.pdf/

http://77898_02b.pdf/

http://77898_02b.pdf/

http://77898_02b.pdf/

http://77898_02b.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_11.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_13.pdf/

http://77898_07a.pdf/

http://77898_07a.pdf/

http://77898_07a.pdf/



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


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) 73

Separation control experiments 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

http://77898_15.pdf/

http://77898_16.pdf/

http://77898_15.pdf/

http://77898_16.pdf/

http://77898_15.pdf/

http://77898_15.pdf/

http://77898_16.pdf/

http://77898_23.pdf/

http://77898_24.pdf/

http://77898_03.pdf/

http://77898_05a.pdf/

http://77898_01.pdf/

http://77898_07a.pdf/

http://77898_21.pdf/

http://77898_21.pdf/

http://77898_07a.pdf/

http://77898_01.pdf/

http://77898_05a.pdf/

http://77898_03.pdf/

http://77898_24.pdf/

http://77898_23.pdf/

http://77898_16.pdf/

http://77898_15.pdf/

http://77898_15.pdf/

http://77898_16.pdf/

http://77898_15.pdf/

http://77898_16.pdf/

http://77898_15.pdf/



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 374

Stability augmentation 54

Steady and pulsed jet effects


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 318

Custer channel wing aircraft 316

experiments on

evaluation and techniques 320

predictions vs actual 331

http://77898_23.pdf/

http://77898_23.pdf/

http://77898_23.pdf/

http://77898_23.pdf/

http://77898_13.pdf/

http://77898_02b.pdf/

http://77898_22.pdf/

http://77898_22.pdf/

http://77898_22.pdf/

http://77898_22.pdf/

http://77898_22.pdf/

http://77898_22.pdf/

http://77898_22.pdf/

http://77898_22.pdf/

http://77898_22.pdf/

http://77898_22.pdf/

http://77898_22.pdf/

http://77898_11.pdf/

http://77898_11.pdf/

http://77898_11.pdf/

http://77898_11.pdf/

http://77898_11.pdf/

http://77898_11.pdf/

http://77898_22.pdf/

http://77898_22.pdf/

http://77898_22.pdf/

http://77898_22.pdf/

http://77898_22.pdf/

http://77898_22.pdf/

http://77898_22.pdf/

http://77898_22.pdf/

http://77898_22.pdf/

http://77898_22.pdf/

http://77898_22.pdf/

http://77898_02b.pdf/

http://77898_13.pdf/

http://77898_23.pdf/

http://77898_23.pdf/

http://77898_23.pdf/

http://77898_23.pdf/



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 4

demonstrator design 6

STOL, short takeoff and landing 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

http://77898_11.pdf/

http://77898_11.pdf/

http://77898_11.pdf/

http://77898_11.pdf/

http://77898_11.pdf/

http://77898_11.pdf/

http://77898_01.pdf/

http://77898_01.pdf/

http://77898_11.pdf/

http://77898_22.pdf/

http://77898_05a.pdf/

http://77898_11.pdf/

http://77898_07b.pdf/

http://77898_03.pdf/

http://77898_03.pdf/

http://77898_08.pdf/

http://77898_08.pdf/

http://77898_08.pdf/

http://77898_08.pdf/

http://77898_08.pdf/

http://77898_08.pdf/

http://77898_03.pdf/

http://77898_03.pdf/

http://77898_07b.pdf/

http://77898_11.pdf/

http://77898_05a.pdf/

http://77898_22.pdf/

http://77898_11.pdf/

http://77898_01.pdf/

http://77898_01.pdf/

http://77898_11.pdf/

http://77898_11.pdf/

http://77898_11.pdf/

http://77898_11.pdf/

http://77898_11.pdf/

http://77898_11.pdf/

applications of circulation control technologies

Documents