AGARD-CP-515

I
I 5 ADVISORY GROUP FOR AEROSPACE RESEARCH & DEVELOPMENT I

2 7 RUE ANCELLE 92200 NEUILLY SUR SEINE FRANCE

AGARD CONFERENCE PROCEEDINGS 515

High-Lift System Aerodynamics

(LAirodynamique des /

Systkmes Hypersustent

Paperspresented and discussions recorded at the 71st Fluid Dynamics Panel Meenni .
and at the Symposium held in Banff, Alberta, Canada,from 5th-8th October 1992.

I
-@ I
c NORTH ATLANTIC TREATY ORGANIZATION

Published September 1993

Distribution and Availability on Back Cover

r --
-Ir>
'Easy access to advanced knowledge and technical information on COMPACT DISK. . . . 'I

The new "NATODISK"

The NATO-PCO DATABASE and the AGARD AEROSPACE DATABASE on CD-ROM

2 important databases of interest to scientists and engineers

working in universities, research institutes and industry

We are pleased to announce that the enhanced version of the "NATO DISK", i.e. the computer-readable COMPACT DISK
(CD-ROM) has just become available. It contains the first major update of the NATO-PCO DATABASE with its 4O.OOO
I
bibliographical records referring to non-military scientific/technical literature generated with the sponsorship of the NATO
- -
Science Committee, and for the first time a special version of the AGARD AEROSPACE DATABASE providing access to
the aerospace literature resulting from the programmes of NATO's Advisory Group for Aerospace Research and Development.

With the easy-to-follow menu options of the retrieval software, access to data is simple and fast! The data is well-structured and
the presentation is pleasant while fundonal. The software CD-ANSWER - (c) Copyright DATAWARE Technologies Inc. -
which comes with the CD-ROM, received the prize for T h e Best CD-ROM Retrieval Software of 1990". It functions in
ENGLISH, FRENCH and GERMAN. There is no need to leam a complex search language.

The CD-ROM can be used with standard Personal Computer equipment CpC =/AT or 100% compatible, minimum 512 KByte
RAM and with MS-DOS/PC-DOS version 3.0 and above) including a CD-ROM drive.

The NATO-PCO DATABASE The AGARD AEROSPACE DATABASE

The NATO-PCO DATABASE -(c) Copyright W T V GmbH, This bibliographic database generated by PCO in co-operation
Germany - is a reference database. It covers more than 20 with NASA (USA) and AGARD (France) provides
years of non-military scientific/technical meetings and information on thousands of AGARD-sponsored publications -
publications in the NATO AS1 SERIES, sponsored by the conference proceed@, AGARDographs (major works of
NATO Science Committee. The NATO AS1 SERIES which - long-lasting interest), lecture series, reports and advisory
is the official vehicle for publication of the results of NATO reports - published from about 1960 to 1991. As is the case for
Advanced Study Institutes, NATO Advanced Research the NATO-PCO DATABASE, the AGARD AEROSPACE
Workshops and other high-level scientific/t&cal meetings - DATABASE includes details not only of the complete
is published by KLUWER ACADEMIC PUBLISHERS (The publications but also of the papers contained in them (up to 40
Netherlands), PLENUM PUBLISHING Corporation (USA) or more in a conference proceedings).
and SPRINGER-VERLAG (Germany)
AGARD (The Advisory Group for Aerospace Research and
The NATO-PCO DATABASE contains full references (With Development) is an agency of NATO based in Paris. Its
keywords and/or abstracts) to more than 4O.OOO contributions mission is essentially to interchange information about R&D
from international scientists of high repute, in aerospace Within and between the NATO nations.
AGARDs areas of activity cover:
The database covers a broad spectrum of scientific disciplines: AEROSPACE MEDICINE, A WONICS, FLIGHT
LIFE SCIENCES, ECOLOGY; MEDICINE, CHEMISTRI: M E C W I C S , ELECTROMAGNETIC WAVE
GEOSCIENCES, ASTRONOMZ: MATHEUATICclL AND PROPAGATION, FLUID D W M I C S , GUIDANCE AND
P H Y , S I W SCIENCES, BENA WOURAL.SCIENCES, CONTROL, PROPULSION AND ENERGETICS,
MATERL4L.S SCIENCES, ENGINEERING, SYSTEMS and STRUCTURES AND MATERLALS, and TECHNICAL
COMPUTER SCIENCES. INFOMTION.

The price for the CD-ROM containing both databases is DM 1.140.- or USD 790.00 (excluding tax).
Updates are planned to appear on a yearly basis.

The "NATO DISK" is available from:

NATO AS1 SERIES Publication Coordination Oflice (PCO),

(Information Office of W T V GmbH, Germany)
Elcerlyclaan 2, B-3090 Overijse / BELGIUM
Tel. +32-2487.6636 Fax. +32-2687.98.82 enfiancais au verso
 #
.'k
t$

~e nouveau "DISQUE SCIENCE &' T E ~ H N OOGIE OTA"?

La base de donnbes du Bureau de coordination des publications (PCO) de I'OTAN et la
Base de donnebs abrospatiales de I'AGARD mises sur disque compact (CD-ROM):
Ces deux bases de donne& importantes pr6sentent un tr& grand interet pour les scientifiques et les
ingdnieurs qui travaillent dans les universitds, les instituts de recherche et I'industrie.
Nous avons le plaiSir d'annoncer que la version ameliorte du "Disque Science & Technologie OTAN"; c'est A dire le &que
optique compact (CD-ROM) lisible par ordinateur, est disponible depuis peu. 11contient la premibre mise &jour importante de
LA BASE DE DO"&S DU BUREAU DE COORDINATION DESPUBLICATIONS DE L'OTAN (NATO-PCO) avec
ses 40.OOO entr6es bibliographiques renvoyant a la litterature scientifique non militaire produite sous I'kgide du ComitC
Scientifique de I'OTAN et, pour la premibre fois, une version spkiale de la "BASE DE DONNkES &ROSPATIALES' de
I I'AGARD donnant a& a la littkrature aerospatiale qui d h u l e des programmes du Groupe consultatif pour la recherche et
les rCalisations aerospatiales (AGARD) de I'OTAN.

Grace la fadit6 d'exploitation offerte par les options sur menu du logiuel d'extradon, I'a&s aux donnees est simple et
rapide! La structure de ces donnees est excellente et leur presentation agreable tout en etant fondonelle. Le logiciel CD-
- -
ANSWER (c) Copyright DATAWARE Technologies Inc. ,qui accompagne le disque compact a r e p le prix du "Meilleur
logiciel &extraction de &que compact CD-ROM" pour 1990.Il fonctionne en ANGLAIS, en FRANQUS et en ALLEMAM).
Il n'impose pas I'apprelissage d'un langage de recherche compliqud.
Le disque optique peut etre utilise avec un ordinateur personnel standard (PC XT/AT 100 % compatible, avec RAM de 512
kilo-octets minimum et MS DOS/PC DOS version 3.0 minimum), comportant un ledeur de disque optique.

BASE DE DONNkES &ROSPATIALES

BASE DE DONNfiES (NATO-PCO) DE L'AGARD
La BASE D E DO"&S NATO-PCO, dont les droits Cette base de donntes bibliographique a& par le Bureau de
d'auteur sont detenus par la societe allemande WTV GmbH, coordination des publications (PCO) en cooperation avec la
est un outil de r6f6rence. Elle couvre plus de vingt wCes de NASA @tats-Unis) et I'AGARD (France) fournit des
rkunions et de publications scientifiques et techniques non informations sur des milliers de publications rkalisees sous
militaires de la s6rie AS1 de POTAN, tenues ou 6mises sous I'egide de I'AGARD et editees entre 1960 et 1991environ:
l'egide du Comit6 Scientifique de I'OTAN. Cette serie, pro& verbaw de conference, AGARDographies (travaux
dhicule officiel de publication des rQultats des instituts importants prhentant un interet sur une longue pbriode),
d'ktudes avancks et des ateliers de recherche de pointe de cycles de conferences, comptes rendus et rap orts consultatifs.
I'OTAN et autres reunions scientifiques et techniques de haut
niveau, est publik par les maisons d'tdition KLUWER
Comme dans le cas de la BASE D E DO "% ES NATO-PCO,
la BASE DE DOMES &ROSPATLXLES DE L'AGARD
ACADEMIC PUBLISHERS des Pays-Bas, PLENUM inclut le detail non seulement des publications completes, mais
PUBLISHING Corporation des ktats-Unis et SPRINGER- a m i des communications qu'elles renferment cjusqu'&40 ou
VERLAG &Allemape. plus dans un pro& verbal de conference).

La BASE D E DO"I% NATO-PCO contient les references L'AGARD (Groupe consultatif pour la recherche et les
intkgrales (y compris les mots clCs/rbum&) concernant plus rCalisations aerospatiales) est une agence de I'OTAN dont le
de 40.000 articles r6diges par des scientitiques internationaux sibge est & Paris. Sa mission consiste essentiellement &
de haut niveau. Cchanger des informations de R et D dans le domaine
akrospatial tant & l'mtkrieur des pays de I'OTAN qu'entre eux.
La base de donnees couvre un large eventail de disciplines Ses secteurs d'activitk couvrent:
scientifiques: SCIENCES DE LA ME, ECOLOGIE, LA MEDECINE AEROSPATIALE, AWONIQUE, L A
MEDECINE, CHIMIE, SCIENCES DE L A TERRE, MECANIQUE DU VOL, LA PROPAGATION DES ONDES
ASTRONOMIE, SCIENCES MATHEMATIQUES ET ELECTROMGNETIQUES, LA DYNAMIQUE DES
P W S I Q UES, SCIENCES D U COMPORTEMENT, FLUIDES, LE GUIDAGE ET LE PILOTAGE, L A
SCIENCES DES M A T E U W , TECHNOLOGIE, PROPULSION ET L'ENERGETIQUE, LES STRUCTURES
SCIENCES DES SYSTEMES ET DE L'INFORMATIQUE. ET LES MATERLQW ET L'INFORMATION TECHNIQUE.

Le prix hors taxe du CD-ROM contenant les deux bases de donnees est de 1140DM ou 790 $ US (soit environ 4OOOFF).
I1 est prCvu de prodder A une mise & jour annuelle.
I1 est possible de se procurer le "DISQUESCENCE & TECHNOLOGIE OTAN"a u p r b du:
NATO AS1 SERIES Publication Coordination Office (PO),
(Bureau d'information de la societe WTV GmbH, Allemape)
Elcerlyclaan 2, B-3090Overuse / BELGIQUE
Tel. +32-2681.6636 Fax. +32-26819882 English version overleaf
I
AGARD-CP-515
liI

ADVISORY GROUP FOR AEROSPACE RESEARCH & DEVELOPMENT

7 RUE ANCELLE 92200 NEUILLY SUR SEINE FRANCE ~

AGARD CONFERENCE PROCEEDINGS 515

High-Lift System Aerodynamics

(LAbrodynamique des
Systkmes Hypersustentateurs)

Papers presented and discussions recorded at the 71st Fluid Dynamics Panel Meeting
and at the Symposium held in Banff,Alberta, Canada, from 5th-8th October 1992.

North Atlantic Treaty Organization

Organisation du Traite de I’Atlantique Nord

I
 The Mission of AGARD

According to its Charter, the mission of AGARD is to bring together the leading personalities of the NATO nations in the fields
of science and technology relating to aerospace for the following purposes:

- Recommending effective ways for the member nations to use their research and development capabilities for the
common benefit of the NATO community;

- Providing scientific and technical advice and assistance to the Military Committee in the field of aerospaceresearch and
development (with particular regard to its military application);

- Continuously stimulating advances in the aerospacesciences relevant to strengthening the common defence posture;
- Improving the co-operationamong member nations in aerospace research and development;
- Exchange of scientific and technical information;
- Providing assistance to member nations for the purpose of increasing their scientific and technical potential;
- Rendering scientific and technical assistance, as requested, to other NATO bodies and to member nations in connection
with research and development problems in the aerospace field.

The highest authority within AGARD is the National Delegates Board consisting of officially appointed senior representatives
from each member nation. The mission of AGARD is camed out through the Panels which are composed of experts appointed
by the National Delegates, the Consultant and Exchange Programme and the AerospaceApplications Studies Programme. The
results of AGARD work are reported to the member nations and the NATO Authorities through the AGARD series of
publications of which this is one.

Participation in AGARD activities is by invitation only and is normally limited to citizens of the NATO nations.

The content of this publication has been reproduced

directly from material supplied by AGARD or the authors.

Published September 1993

Copyright Q AGARD 1993
All Rights Reserved

ISBN 92-835-0715-0

Printed by Specialised Printing Services Limited

40 ChigwellLane, Loughton, Essex IGlO 3TZ

ii
15-10

DATA
....... NUMERICAL-UPPER SURFACE
- F
F
-ACF
1 o EXPERIMENTAL-UPPERSURFACE
* EXPERIMENTAL-LOWERSURFACE ii. *'4.
1 ooo
:
0
:
,

0"
A

0 INTEGRATION OF
EXPERIMENTAL Cp DATA
0 N UME RlCAL RESULTS
*
0.0 0.2
I
0 ' d.2 ' 014 ' 0:s ' 0l8 ' 1
WC WC
l I I I I
0 4 8 12 16 i
ANGLE OF ATTACK. a

Fig. 14 Ltfl Coefficient Results as Compared with Fig. 15 Comparison of Surface Pressure Solution
Tunnel Balance Data & Surface Pressure with Experimantal Data (U, = 0.17,
Data Integrated Values (M,=0.17, Re = 3.44 a = 6.1 95", c p = 0.04)
Re=3.4M, & . = O . O l )

Fig. 16 isobar Solution at M,=0.17, Re=3.4M, Fig. 17 Streamline Solution at M,=0.17, Re=3.4M,
-6.1 95", a=0.04
a=6.195", c p = 0.04
 4. " 0
15-9

7. Brune, G.W. and Manke. J.W.. "An Improved Version 12. Pulliam. T.H., "Euler and Thin Layer Navier-Stokes
of the NASA-Lockheed Multielement Airfoil Analysis Codes: ARC2D, ARC3D," Notes for ComDutational
Computer Rogram," NASA CR-145323, 1978. Fluid Dvnamics User's Workshou, The University of
Tennessee Space Institute, Tullahoma. TN. (UTSI
8. Drela, M., "Newton Solution of Coupled Publication E02-4005-023-84), March 1984. pp
Viscid/Inviscid Multi-Element Airfoil Flows." AIAA 15.1-15.85.
Paper 90-1470. June 1990.
13. Thomas, P.D.. "Numerical Method for Predicting
9. Chow, R. and Chu, K.. "Navier-Stokes Solution for Flow Characteristics and Performance of
High-Lift Multielement Airfoil System with Flap Nonaxisymmetric Nozzles --Theory." NASA CR-
Separation," AIAA Pater 91-1623, June 1991. 3147. 1979.
10. Cooper, G.K.. '"The PARC Code Theory and Usage," 14. Baldwin, B.S. and Lomax, H., "Thin Layer
Arnold Engineering Development Center Report Approximation and Algebraic Model for Separated
AEDC-TR-98-24. October 1987. Turbulent Flows." AIAA Paper 78-257, January 1978.
11. Cam. J.E.. "An Aerodynamic Comparison of Blown 15. Steibrenner, J.P., Chawner, J.R., and Fouts, C.L.,
and Mechanical High Lift Airfoils." Proceedings of 'The GRIDGEN 3D, Multiple Block Grid Generation
the Circulation-Control Workshop, NASA CP-2432, System." WRDC-TR-90-3022, Vols. I & II, Wright
February 1% Patterson Air Force Base, 1990.

I / I
Fig. 12 Isobar Solution at M,=0.17, Re=3.4M,
a=6.153", c p = 0.01

Flg. 13 Velocity Vector Field Solution on Top Surface

(M,=0.17, Re=3.4M, a=6.153", &O.Ol)
 Recent Publications of
the Fluid Dynamics Panel
AGARDOGRAPHS (AG)
Design and Testing of High-Performance Parachutes
AGARD AG-319, November 1991
Experimental Techniques in the Field of Low Density Aerodynamics
AGARD AG-318 (E),April 1991
Techniques Experimentales Liees a I’Ahrodynamique a Basse Densite
AGARD AG-318 (FR), April 1990
A Survey of Measurements and Measuring Techniques in Rapidly Distorted Compressible Turbulent Boundary Layers
AGARD AG-315, May 1989
Reynolds Number Effects in Transonic Flows
AGARD AG-303, December 1988

REPORTS (R)
Shock-Wavemoundary-LayerInteractions in Supersonic and Hypersonic Flows
AGARD R-792, Special Course Notes, August 1993
Unstructured Grid Methods for Advection Dominated Flows
AGARD R-787, Special Course Notes, May 1992
Skin Friction Drag Reduction
AGARD R-786, Special Course Notes, March 1992
Engineering Methods in Aerodynamic Analysis and Design of Aircraft
AGARD R-783, Special Course Notes, January 1992
Aircraft Dynamics at High Angles of Attack Experiments and Modelling
AGARD R-776, Special Course Notes, March 1991

ADVISORY REPORTS (AR)

Air Intakes for High Speed Vehicles
AGARD AR-270, Report of WG 13, September 1991
Appraisal of the Suitability of Turbulence Models in Flow Calculations
AGARD AR-291, Technical Status Review, July 1991
Rotary-Balance Testing for Aircraft Dynamics
AGARD AR-265, Report of WG 11, December 1990
Calculation of 3D Separated Turbulent Flows in Boundary Layer Limit
AGARD AR-255, Report of WGlO, May 1990
Adaptive Wind Tunnel Walls: Technology and Applications
AGARD AR-269, Report of WG12, April 1990

CONFERENCE PROCEEDINGS (CP)

Theoretical and Experimental Methods in Hypersonic Flows
AGARD CP-514, April 1993
Aerodynamic Engine/Airframe Integration for High Performance Aircraft and Missiles
AGARD CP-498, September 1992
Effects of Adverse Weather on Aerodynamics
AGARD CP-496, December 1991

iii
 Manoeuvring Aerodynamics
AGARD CP-497, November 1991
Vortex Flow Aerodynamics
AGARD CP-494, July 1991
Missile Aerodynamics
AGARD CP-493, October 1990
Aerodynamics of Combat Aircraft Controls and of Ground Effects
AGARD CP-465, April 1990
Computational Methods for Aerodynamic Design (Inverse) and optimization
AGARD-CP-463, March 1990
Applications of Mesh Generation to Complex 3-D Codigurations
AGARD CP-464, March 1990
Fluid Dynamics of Three-Dimensional Turbulent Shear Flows and Transition
AGARD CP-438, April 1989
Validation of Computational Fluid Dynamics
AGARD CP-437, December 1988
Aerodynamic Data Accuracy and Quality: Requirements and Capabilities in Wind Tunnel Testing
AGARD CP-429, July 1988
Aerodynamics of Hypersonic Lifting Vehicles
AGARD CP-428, November 1987
Aerodynamic and Related Hydrodynamic Studies Using Water Facilities
I AGARD CP-413, June 1987
Applications of Computational Fluid Dynamics in Aeronautics
AGARD CP-412, November 1986
Store Airframe Aerodynamics
AGARD CP-389, August 1986
Unsteady Aerodynamics - Fundamentals and Applications to Aircraft Dynamics
AGARD CP-386, November 1985
Aerodynamics and Acoustics of Propellers
AGARD CP-366, February 1985
Improvement of Aerodynamic Performance through Boundary Layer Control and High Lift Systems
AGARD CP-365, August 1984
Wind Tunnels and Testing Techniques
AGARD CP-348, February 1984
Aerodynamics of Vortical Type Flows in Three Dimensions
AGARD CP-342, July 1983
Missile Aerodynamics
AGARD CP-336, February 1983
Prediction of Aerodynamic Loads on Rotorcraft
AGARD CP-334, September 1982
Wall Interference in Wind Tunnels
AGARD CP-335, September 1982
Fluid Dynamics of Jets with Applications to V/STOL
AGARD CP-308, January 1982
Aerodynamics of Power Plant Installation
AGARD CP-30 1,September 1981
Computation of Viscous-Inviscid Interactions
AGARD CP-29I, February 1981
Subsonicfhnsonic Configuration Aerodynamics
AGARD CP-285, September 1980
iv
 Foreword

The ability to design efficient high-lift systems for aircraft has become increasingly important for a number of reasons.

Firstly, during the last 10-15years, the worldwide ability to design wings for efficient transonic flight has been transformed by:
a) increased understanding of how to design for mixed subsonic and supersonic flow over the wing;
b) the increased capability and reliability of CFD codes in predicting these flows.

As a result, wing loadings have been able to increase without incumng any penalty in cruise or turn performance. The corollary
of this is that increased demands have been put on the design of the high-lift system in order to retain the same take-off and
landing performance.

Secondly, the current world military situation, with the cessation of the Cold War, demands that NATO forces are able to react
quickly to a developing situation anywhere in the world, and thus that military transport aircraft are able to operate from
unprepared and short landing strips - again making increased demands on the design of the high-lift system.

Finally, the demands made by stealth requirements on the shape of the aircraft -generating configurations such as the F117 and
the B2 - lead to geometries which are far from optimum aerodynamically, and hence which are likely also to make increased
demands on the design of the high-lift system in order to preserve the landing and take-off performance.

Consequently, it is somewhat surprising that the last AGARD Symposium concerned with high-lift systems was as long ago as
1984 - and even then the Symposium subject was split between “High-Lift” and “Drag Reduction” (AGARD CP 365).

However, recent advances in CFD techniques, and in the understanding of the complex flows that occur over high-lift systems,
suggest that more efficient systems can now be designed with a lower risk than before, and that it is therefore very appropriate
that AGARD FDP should hold a Symposium, at this time, solely dedicated to the subject of “High-Lift System Aerodynamics”.

The Symposium was aimed to address:

a) how, given the role and configuration of the aircraft, the most appropriate high-lift system can be selected;
b) how, having selected the type of high-lift system, an efficient detailed design can be produced;
c) the experimental and analysis techniques which are necessary to explore and enhance the performance of the high-lift
system.

Thus, although the aerodynamicsof high-lift systems was the dominant theme, the very pertinent aspects of weight, simplicity,
reliability, and structural and mechanical integrity, were an integral part of the Symposium, and were treated in the papers
presented. In this sense, this Symposium has attempted to take a broader view of the high-lift system than has been taken by
similar conferences in the past.

V
 Avant-Propos

La conception de systkmes hypersustentateurs performants pour a6ronefs prend de plus en plus d‘importance aujourd’hui, et
ceci pour differentes raisons.

En premier lieu, au cours des derniers 10 ii 15 ans, les capacitks de la communaute abronautique internationale dans le domaine
de la conception des voilures pour le vol transsonique performant ont CtC transformks par:
a) une meilleure comprChension de la prise en compte des Ccoulements mixtes subsoniques et supersoniques autour de la
voilure dks la phase d’ktudes;
b) un accroissement de la capacitk et la fiabilitC des codes CFD dans la pridiction de ces koulements.

Ceci a permis d‘augmenter les charges alaires sans pinaliser les performances de croisikre ou de virage. Ce dCveloppement a eu
pour corollaire I‘imposition de contraintes plus rigoureuses sur la conception des systkmes hypersustentateurs afin de conserver
les mkmes performances au ddcollage et A I’atterrissage.

En deuxitme lieu, la situation militaire mondiale actuelle, avec la cessation de la guerre froide, nCcessite une riaction rapide de
la part des forces de I’OTAN face B une dventuelle situation de crise n’importe OG dans le monde. Ceci implique la mise en oeuvre
opCrationnelled’avions de transport B partir de pistes d’attemssage courtes et non-prCparCes et entraine de nouvelles demandes
au niveau de la conception des systkmes hypersustentateurs.

Eniin, les consiquencesdes exigences de furtivitk sur la forme des aCronefs - c’est a dire la criation de configurations telles que
celles des F117 et B2 - mknent B des gComCtries qui sont loin d’btre optimales du point de vue akrodynamique. Par voie de
constquence,elles risquent d’imposer des contraintes suppltmentairessur la conception des systkmes hypersustentateurs dans
le cas oh les performances au dCcollage et a I’attemssage devraient btre conservks.

Il est donc surprenant de constater que le dernier symposium organist par I’AGARD sur ce sujet remonte a 1984 - et que le
thkme de la rkunion Ctait partage entre “I’hypersustentation”et “la rkduction de la trainte” (AGARD CP 365).

Cependant, les progrks rialisis rkemment en akrodynamique numirique, et dans I’analyse des Ccoulements complexes autour
des systkmes hypersustentateurs laissent supposerque des systkmes plus performants peuvent dbormais btre rialids a moindre
risque que par le passe. Il Ctait, par consiquent, tout B fait opportun pour le Panel FDP de I’AGARDd’organiser un symposium
consacrC au seul sujet de “I‘airodynamiquedes systtmes hypersustentateurs”.

Le symposium a examink les thkmes suivants:

a) le choix optimal de systkmes hypersustentateurs en fonction du rijle et de la configuration de I‘aCronef;
b) la rbalisation d‘une Ctude dCtaillbe optimale en fonction du systkme hypersustentateur choisi;
c) les techniques expCrimentales et d‘analyse nkessaires B l’examen dCtaiUCe et I’amClioration des performances des systkmes
hypersustentatuers.

Ainsi, bien que I’atrodynamique des systkmes hypersustentateurs Mt bien le thtme dominant de symposium, certains aspects
trks pertinents tels la masse, la simplicitk, la fiabiliti et I’intCgritC structurale et mbcanique faisaient partie integrante de la
reunion et figuraient dans les communications prCsentCes. De cette faGon, le symposium a voulu donner une vue beaucoup plus
large des systkmes hypersustentateurs, ce qui que n’a pas CtC le cas pour d‘autres confkrences organisdes sur ce sujet dans le
pass&

vi
 Fluid Dynamics Panel

Chairman: Prof. Ir J.W. Slooff Deputy Chairman:M.C. Dujanic

National Aerospace Laboratory NLR ASE-Programme HERMES
Anthony Fokkerweg 2 Bitiment Copernic
1059 CM Amsterdam 18 Avenue Edouard Belin
The Netherlands 31055 Toulouse Cedex -France

PROGRAMMECOMMI'ITEE

Mr D. Woodward (Co-chairman) Lt Col. D. Petridis

SuperintendentAP3 Hellenic Air Force General Staff
Low Speed Aerodynamics Division Branch C'- Directorate C1
X80 Bldg Section 4
Defence Research Agency Holargos, TGA 1010
Farnborough, Hants GU14 6TD Athens, Greece
United Kingdom
Prof. M. Onorato
Mr LJ. Willaims (Co-chairman) Dipartimentodi Ingegneria
Director, High-speed Research Division Aeronautica e Spaziale
Office of Aeronautics & Space Technology Politecnico di Torino
NASA (Code RJ) C. so Duca degli Abruzzi 24
Washington, DC 20546 10129 Torino, Italy
United States Prof. Dr Ir J.L. van Ingen
Prof. R. Decuypere Department of Aerospace Engineering
Ecole Royale Militaire Delft University of Technology
Chaire de MCcanique Appliquee (MAPP) Kluyverweg 1
Avenue de la Renaissance 30 2629 HS Delft, The Netherlands
B-1040 Brussels, Belgium
Dr L.P. Ruiz-Calavera
Dr L. Chan INTA
High Speed Aerodyn. Lab. - U66 Division de Aerodinamica
Institute for Aerospace Research Carretera de Ajalvir Km. 4.5
National Research Council 28850 Torrejon de Ardoz
Montreal Road Madrid, Spain
Ottawa, Ontario KIA OR6, Canada
Dr W.J. McCroskey
MJ. Bousquet Senior Staff Scientist
Division Avions-A/DET/AF' MO 151/7 US Army Aeroflightdynamics
ACrospatiale Directorate - M I S N258-1
316 Route de Bayonne NASA Ames Research Center
31060 Toulouse Cedex 03, France Moffett Field, CA 94035-1099
United States
Dr B. Wagner
Dornier Luftfahrt GmbH
Abteilung Aerodynamik
Postfach 1303
D-7990 Friedrichshafen, Germany

PANEL EXECUTIVE
Dr W. Goodrich
Mail from Europe: Mail from US and Canada:
AGARD-OTAN AGARD-NATO
Attn: FDP Executive Attn: FDP Executive
7, rue Ancelle Unit 21551
F-92200 Neuilly-sur-Seine APO AE 09777
France Tel: 33(1)47 38 57 75
Telex: 610176 (France)
Telefax: 33 (1) 47 38 57 99

vii
 Contents

Page

Recent Publications of the Fluid Dynamics Panel

...
111

Foreword V

Avant-Propos vi

Fluid Dynamics Panel vii

Reference

Technical Evaluation Report T

by D.C. Whittley
-
SESSION I INTRODUCTION
Chairman:L. Chan

-
Where is High-Lift Today? A Review of Past UK Research Programmes 1
by D.S. Woodward and D.E. Lean

-
SESSION I1 HIGH-LIFT IN 2D (1)
Chairman:J.L. Van Ingen

Experimental Investigation of Flow around a Multielement Airfoil 2

by N. Alemdaroglu

Paper 3 withdrawn
A Design Procedure for Slotted Flaps
by S . De Ponte, A. Cella and M. Marcazzan
Calculation of Maximum, and High Lift Characteristics of Multi Element Aerofoils
by W. Fritz
Navier-Stokes Computations of Turbulent Flow around High-Lift Configurations
by P. Bartsch, W.Nitsche and M. Britsch
Efficient Simulation of Incompressible Viscous Flow over Multi-Element Airfoils
by S.E. Rogers, N.L. Wiltberger and D. Kwak
Navier-Stokes Calculations on Multi-Element Airfoils using a
Chimera-Based Solver
by D.W. Jasper, S. Agrawal and BA. Robinson

-
SESSION 111 HIGH-LIFT IN 2D (2)
Chairman:B. Wagner

Numerical Solution of the Navier-Stokes Equations for High-Lift 9

Configurations on Structured Composite Grids
by T.E. Nelson, D.W. Zingg and G.W. Johnston
High Reynolds Number Configuration Development of a High-Lift Airfoil 10
by W.O. Valarezo, C.J. Dominik, R.J. McGhee and W.L. Goodman

viii
 Reference

Une MCthode d’Interaction Visquew Non-visqueux pour Ecoulements 11

Incompressibles Hypersustentes sur Profils Multi-corps en Rkgime de
DCcollement Profond
(A Viscous-Inviscid Solver for High-LiftIncompressible Flows over
Multi-Element Airfoils at Deep Separation Conditions)
par J.C. Le Balleur et M. NCron

High-Lift System Analysis Method using Unstructured Meshes 12

by K. de Cock
Prediction of the High-Lift Performance of Multi-Element Aerofoils 13
using an Unstructured Navier-Stokes Solver
by LJ. Johnston and L. Stolcis
Numerical Calculations of High Lift Flows using Structured and 14
Unstructured Methods
by R. Bailey et al.
Navier-Stokes Simulation of Flow Field around a Blown-Flap 15
High-Lift System
by R. Chow, K. Chu and G. Carpenter

SESSION IV - HIGH-LIFT EXPERIMENTS IN 3 D

Chairman:R. Decuypere
The GARTEUR High Lift Research Programme 16
by JJ. Thibert
An Experimental Investigation of the Optimum Slat Setting on a Combat 17
Aircraft Model
by I.R.M. Moir
An Experimental Investigation of Attachment-Line Transition on the 18
Slat of a Combat Aircraft Model
by B.C. Hardy
Viscous Phenomena Affecting High-Lift Systems and Suggestions for 19
Future CFD Development
by P.T. Meredith
A Study of the Use of Half-Models in High-Lift Wind-Tunnel Testing 20
by P.B. Earnshaw, A.R. Green, B.C. Hardy and A.H. Jelly

In-flight Pressure Distributions and Skin-Friction Measurements on a 21

Subsonic Transport High-LiZt Wing Section
by L.P. Yip, P.M.H.W. Vijgen, J.D. Hardin and C.P. van Dam

High-Lift and the Forward Swept Wing 22

by L.A. Walchli

-
SESSION V HIGH-LIFT CFD IN 3 D
Chairman:L. Williams
A Fast Computing Method for the Flow over High-Lift Wings 23
by K. Jacob
Calculation of Multielement Airfoils and Wings at High Lift 24
by T. Cebeci
 Reference
’
Wake Structure and Aerodynamic Behavior of High-Lift Aircraft Configurations 25
during Unsteady Maneuvers in Ground Effect
by A. Baron and M. Boffadossi

Calcul par Interaction Visqueux non-Visquew des Ecoulements Compressibles 26

Fortement D6coll6s aux Grandes Portances sur Profils d’Ailes et Voilures
(Viscous-Inviscid Calculation of High-LiftSeparated Compressible Flows over
Airfoils and Wings)
par J.C. Le Balleur

SESSION VI - DESIGN O F HIGH-LIFT SYSTEMS FOR COMPLETE AIRCRAFT

Chairman: J. Bousquet

Forty Years of High-Lift R&D - An Aircraft Manufacturer’s Experience 27

by E. Oben
High-Lift Research Application to the Design of the ATR72 Flap 28
by P. Capbern
The Aero-Mechanical Design of a Novel Fowler Flap Mechanism 29
by J.R. Mathews
Design, Development and Flight Evaluation of the Boeing YC-14 USB 30
Powered-Lift Aircraft
by T.C. Nark

High-Lift Design for Large Civil Aircraft 31

by A. Haig and R. Hilbig
High LiZt Systems for Transport Aircraft with Advanced Airfoils 32
by B. Eggleston and R.J.D. Poole
Choice and Optimization of a High-Lift System for an 33
Advanced Amphibious Aircraft
by M A . Averardo, M. de Leo and V.Russo
Round Table Discussion RTD .
. .

X
 T-1

Technical Evaluation Report

D.C. Whittley
17 Book’s Landing
375 Book Road
Grimsby, Ontario L3M 2M8
Canada

1. INTRODUCTION principles at work. It was not until 1972 that the theoretical
The AGARD Fluid Dynamics Panel organized a symposium treatise of A.M.O. Smith provided a clear and comprehen-
on “High-Lift System Aerodynamics” which was held in sive insight into the fundamentals of a multi-element foil
Banff, Alberta, Canada, from 5 October through 8 October, (34). He showed that it was by splitting the pressure rise
1992. This meeting represented a natural successor to a pre- over several elements that the tendency for flow separation
vious FDP symposium held in Brussels, Belgium in 1984, could be suppressed and, thereby, lift potential could be
which was entitled “Improvements of Aerodynamic increased. Along with this understanding, came a deeper
Performance through Boundary Layer Control and High-Lift awareness of the role and importance of viscous effects.
Systems”. Considerable progress has been made in the area
of high-lift systems since the Brussels conference, especially Three important developments took place during the follow-
ing ten years or so:
in the areas of experimental testing and theoretical predic-
tion (otherwise referred to as computational fluid dynamics - construction of ,some new pressurized wind tunnels in
or CFD). member states, to permit a more thorough exploration of
scale effects and encourage testing at higher levels of
In view of the current aerospace climate and corresponding
Reynolds number;
industrial interest, for the Banff symposium, it was decided
to focus attention on aerodynamic research and technology - fairly rapid development of inviscid-viscous theoretical
applicable to advanced (passive) high-lift systems for sub- methods as applied to multi-element foils and wings;
sonic transport application. Notwithstanding this emphasis, - an upsurge in experimental research to strengthen the
three papers were presented dealing with experimental test-
empirical aerodynamic data base.
ing of low aspect ratio wing planforms, suitable for combat
type aircraft and one dealing with powered-lift for transport These developments were reported at the FDP symposium,
application. Brussels, 1984.
The first paper contained a review of some past research In the past decade, continuing work has been associated with
programmes in the United Kingdom and was presented by the following:
Dr D.S. Woodward, co-chairman of the programme commit-
consolidation and development of inviscidviscous cou-
tee. Paper 3 was withdrawn so that, of the 13 papers which
pled methods based on integral boundary layer calcula-
followed, 1 1 dealt with CFD 2D methods as applied to high-
tions;
lift airfoil configurations such as single or double-slotted
flap with some form of leading edge device. The next seven a growing effort associated with computational field
papers concerned experimental testing of high-lift configura- methods which require a grid of the domain and permit,
tions in 3D. The following session of four papers was devot- for example, solutions to the Reynolds averaged Navier-
ed to CFD methods as applied to 3D high-lift wings and, Stokes equations;
finally, seven papers were presented in which industrial rep-
a build-up of reliable experimental data relating to high-
resentatives described design and development of high-lift
lift configurations showing separately the effect of
systems for complete aircraft.
Reynolds number and Mach number; also correlations
References (I) through (33) list the symposium papers in the between wind tunnel and flight measurements;
order in which they were presented.
boundary layer and wake measurements to provide a
Delegates attending the symposium numbered 105. deeper understanding of the flow physics associated with
multi-element foils.
2. BACKGROUND
It was consideration of these aspects of the subject which
Typically, the aerodynamic high-lift design process in indus-
dominated the symposium in Banff.
try involves the use of an empirical data base, theoretical
prediction methods, wind tunnel testing and flight develop-
2.2 The Design Process
ment.
Two important factors overshadow the design process:
2.1 High-Lift Aerodynamic Technology - There are serious penalties for failing to meet the flight
The very beginning of high-lift can be traced back to the design goals of the high-lift system for a projected air-
years 1917-1920 (l), but even after 30 years, most of the craft and, therefore, the subject must be afforded an
design data were compiled in just four publications (27), but appropriate emphasis and effort. For example, the design
there was little understanding of the underlying aerodynamic procedure includes extensive parametric studies at the
T-2

project design stage - time is especially important here; Although effective as an engineering tool, the ‘coupled’
optimization and refinement are required at the early method is somewhat limited in being unable to compute
development stage - costs can soar if undertqken solely much beyond maximum lift and it may have other shortcom-
in the wind tunnel; verification of the final configuration ings associated with slat/flap wells, thick trailing edges, etc.
is usually undertaken in a wind tunnel at reasonably high Hence, potentially more accurate field methods have been
Reynolds number - generally regarded as essential at sought which define the domain by means of a computation-
whatever cost. al grid and permit, for example, solutions to the Reynolds-
- The flow field of a high-lift aircraft configuration is averaged Navier-Stokes equations. Such methods tend to
place great demand on computer capability and capacity
extremely complex, including such phenomena as
which result in a longer solution time andor an increase in
boundary layer growth and separation, wake develop-
cost; these methods have not yet reached a level of maturity
ment, confluence ,of boundary layers, viscous wake inter-
to permit routine industrial use. In order to overcome these
actions, laminar separations, etc.
difficulties, some investigators have considered various
Therefore, as a result of these factors, theoretical and experi- degrees of simplification while still seeking to retain an ade-
mental aerodynamic technology is aimed toward improve- quate degree of accuracy; for example, effects of compress-
ment of the accuracy of prediction to reduce risk, and toward ibility can be omitted or one can opt for a fairly simple tur-
reduction in time and cost of the design process. bulence model. Likewise, attempts are constantly being
made to find more efficient ways to generate and refine the
3. DISCUSSION computational grid.
3.1 The Opening Review Paper
The introductory paper (1) by Woodward and Lean, DRA, 3.2.1 Structured Versus Unstructured Grids
United Kingdom, described a “National High-Lift Early work was based on the generation and use of struc-
Programme” conducted in the United Kingdom in the 1970s, tured grids. More recently, unstructured grid solvers have
and contained much data heretofore not generally available. been studied which, although requiring a greater memory
The work was aimed toward strengthening the overall aero- capacity, do allow ease of automatic grid generation about
dynamic data base used by industry so as to make empirical multi-element configurations: also, local increase in grid
methods of design more reliable. The use of such a data base density for critical areas can be easily achieved. During the
was described by Butter at the 1984 Brussels symposium symposium, some opinions were expressed regarding the
(35). relative merits of structured versus unstructured grids but it
would seem unwise to make a final judgement at the present
Systematic tests were undertaken on four models: one wall-
stage of development.
to-wall and one endplate model for 2D studies, and, one con-
stant chord and one tapered planform model (both having a
3.2.2 Navier-Stokes Solutions
variable sweep capability) for 3D studies.
In all, there were eight presentations, which described
In particular, the research programme provided a basis for numerical solutions to the Navier-Stokes equations, six of
optimization of slat and flap positions relative to the main which, as described below, used structured grid solvers.
foil. Rake measurements gave some insight into the growth Each investigator discussed the foundation of his particular
of boundary layers and wakes on the various foil elements method and made comparison with experimental or test
and thereby suggested reasons for optimum locations; of cases - generally, reasons were suggested for discrepancies
special interest were cases in which the slat wake merged in correlation.
with the wing boundary layer.
Fritz, Dornier, Germany (5). A proper treatment of the wake
The research programme went beyond the purely aerody- of the main foil becomes important at high angles of attack:
namic aspects of high-lift systems to include consideration this requires fine grid resolution and the correct prediction
of variation in structure weight as geometry was adjusted for of viscous effects within the wake. Simple turbulence mod-
best aerodynamic performance. It was shown that, by use of els failed in prediction of the eddy viscosity for such a com-
weight fractions, an optimum design could be found (in plex flow field and therefore’a two equation model is need-
terms of best fuel plus payload) taking into account both ed.
aerodynamic and structural factors.
Bartsch, Nitsche and Britsch, TU Berlin, Germany (6). As in
The authors point out that this research effort, conducted (5). the authors indicated the need for investigating turbu-
jointly by industry and government groups, played a major lence modelling more thoroughly but also pointed to a
role in overcoming a perceived shortfall in UK high-lift requirement for more detailed and well established experi-
technology at the time. mental cases against which to test theoretical predictions.
Also, there is a need to assess numerical solution errors
3.2 Theoretical Methods: High-Lift in 2D when making comparisons between theory and experiment.
Inviscidhiscous coupled methods have been available in a
Rogers, Wiltberger and Kwak, NASA Ames, United States
form suitable for industrial use since the early 1980s (36)
(7). Two different approaches for grid generation were
and (4), by de Ponte, Cella and Marcazzan, Milan Poly-
described and two different turbulence models were studied.
technic, Italy, is a good example. These methods serve to
Test cases included single, two, three and four element air-
complement and/or displace previous design procedures
foils. Predictions are generally good but correlation deterio-
which were based solely on an empirical data base - they
rates when significant areas of separated flow arise. Again,
rely on an integral boundary layedwake model. These codes further work in turbulence modelling is recommended.
are attractive on account of computational efficiency and
flexibility (that is, providing a choice in the level of sophisti- Jasper, Agrawal and Robinson, McDonnell Douglas, United
cation to suit various stages of design). States (8). Good accuracy of prediction was achieved for a
 T-3

two element airfoil without confluent boundary layer flow. investigating grid generation techniques for multi-element
Similarly, good agreement was achieved on four elements airfoils. The method was adapted to simulate separated
without separation on the slat. The authors conclude that regions in slat and wing coves and was moderately success-
treatment of separation, stall and post-stall conditions ful in predicting lift, drag and pitching moment of the three
requires better modelling of turbulence. Also, it was pointed element airfoil. Improvements will be sought by an exten-
out that accurate calculation of gap effects may require sion of the mathematical model to Reynolds averaged
modelling of transitional and re-laminarizing boundary lay- Navier-Stokes equations.
ers.
3.2.3 Viscous-Inviscid Interaction Methods
Nelson, Zingg and Johnston, Toronto University, Canada
Methods based on viscous-inviscid interaction have been
(9). Test cases were shown for a two element and three ele-
developed primarily in Europe (38) as a means to obtain rel-
ment airfoil. Grids were generated by an automated proce-
atively fast solutions and avoid reliance upon a “super com-
dure which divides the domain into blocks. Good agreement
puter” even though the approach might not capture all the
with experiment is obtained when flow is attached or regions
features of the flow.
of separated flow are small. The Baldwin-Barth turbulence
model works well in most instances but under-predicts eddy Le Balleur and NCron, ONERA, France (1 1). At ONERA,
viscosity in wake regions and over-predicts in the outer por- Le Balleur and his colleagues have persevered with the
tions of the boundary layer. method of viscous-inviscid interaction and made improve-
ments so that it is now possible to make predictions for
Chow, Chu and Carpenter, Grumman Research Center,
deeply separated flows over multi-element airfoils at incom-
United States (15). In this paper, the authors apply their
pressible speeds. This achievement is attributed to progress
method to prediction of flow around a three element airfoil
made in both coupling algorithms and generalized thin-layer
with blown flap; however, in the context of the AGARD
modellings along with manipulation of grid resolution, espe-
symposium, the computational method is of more interest
cially when close to separated flow regions. Test cases are
than the test case: the paper should be read in conjunction
chosen in which separation occurs in the cove of both slat
with a previous publication by the same authors (37). The
and flap at mid-chord of the flap. Good correlation with
method is shown to be capable of predicting both onset and
experiment for surface pressure distribution is shown over a
growth of separated regions and wakes with good degree of
broad range of flow conditions.
correlation - even with respect to drag. The authors seem to
attribute this success to a novel “stacked-C” mesh system
3.3 Theoretical Methods: High-Lift CFD in 3D
which was used to map the multi-element geometry into a The complex flow fields associated with high-lift are diffi-
single computational domain and to the fact that the flow
cult to predict in 2D and become even more so in 3D, so that
separation zone is treated with a semi-adaptive mesh gener-
comprehensive solutions in 3D are at a relatively early stage
ating procedure that utilizes slat, foil and flap trailing
of development.
streamline. The code uses an algebraic turbulence model
loosely based on the Thomas formulation of the Baldwin- Four papers were presented (23), (24), (25) and (26), the
Lomax model. first of which by Jacob, DLR, Gottingen, Germany (23)
described details of a method linking 2D viscous calcula-
Johnson and Stolcis, UMIST, Manchester, United Kingdom
tions with a 3D lifting surface theory. Such methods are both
(13). In this case, the work was based on an unstructured
fast and economical and have proved valuable for use in
grid solver but the significance lies in the fact that the
industry (36).
authors made few simplifying assumptions at the outset.
The paper. (24) by Cebeci, Douglas Aircraft, United States,
The approach consists of solving the compressible Reynolds
dealt with multi-element airfoils and wings with theory
averaged Navier-Stokes equations, using a two-equation tur-
based on interaction between inviscid and boundary layer
bulence model. The authors conclude that a two-equation
equations - such an approach serves to avoid excessive
turbulence model is the lowest order model consistent with
demands on the computer. In spite of some success in devel-
complexity of the flow physics. A baseline computational
opment of a turbulence model, the author stressed the need
method has been developed with encouraging initial results
for further improvements in order to achieve better correla-
at both low speed and transonic high-lift conditions when
tion with experiment in the presence of separated regions.
tested against a two element airfoil.
In paper (26), Le Balleur, ONERA, France, continued his
Bailey et al., DLR Braunschweig, Germany (14). ‘Some
earlier discourse in the conference (11) with regard to vis-
interesting comparisons were made between use of struc-
cous-inviscid interaction methods, extending the work to
tured and unstructured grids: in general, it was found that
include compressibility at high-lift and deeply separated
good agreement was obtained with both strategies. Like
flow conditions for airfoils and wings. The author pointed to
(13), the authors based their method on solution of the full the modelling success achieved using a two-equation turbu-
Reynolds averaged Navier-Stokes equations and used a tur-
lence model for a NACA 4412 2D airfoil up to 28 degrees of
bulence model developed by Johnson and King in which
incidence and showed good correlation for lift and drag
both convection and diffusion of turbulence is taken into within the experimental range. Two examples of 3D calcula-
account so as to allow a more accurate determination of tur-
tions were presented, a rectangular plan wing of aspect ratio
bulent stresses in separated boundary layer flow. For a single
6, and an unswept tapered wing of aspect ratio 9.5. The
foil, this turbulence model gave superior correlation and
method is capable of predicting the spanwise spread of the
may lead to similar success for more complex high-lift con-
separation line and the corresponding pressure distribution
figurations. and boundary layer displacements on the wing and in the
De Cock, NLR, The Netherlands (12). The author chose to wake. For the tapered wing, comparisons are made with
base his work on the Euler equations as a fast means of experiment for a range of incidence. So far, treatment has
T-4

been limited to plain unflapped foils, but the method holds cate the matter still further and so, not surprisingly, the
promise for multi-element foils based on the ability to cope author spoke of the need and outlined the scope of possible
with separated regions of flow. future work.
The paper (25) by Baron and Boffadossi, Milan Polytechnic, Meredith, Boeing, United States (19) continued on this
Italy, dealt with a different aspect of high-lift, namely, the theme by citing six viscous features which affect the flow
hazards associated with persistence of vorticity generated in field of a typical multi-element lifting system of which three
h e proximity of an airfield during take-off and landing pro- might be the cause of degradation in lift with increasing
cedures. By means of a complex mathematical treatment, the Reynolds number, that is to say: viscous wake interactions,
authors were able to predict vortex distribution and strength. re-laminarization and attachment line transition. These three
The time history is shown during a typical take-off manoeu- phenomena were described in some detail and then illustrat-
vre for a chosen high-lift winghailplane configuration. ed by means of test results. It was concluded that these mat-
Effects of ground proximity are discussed. ters are of more than academic interest and that neither the
current CFD tools nor the existing experimental data base
3.4 High-Lift Experiments in 2D and 3D are sufficiently well established to meet the need of more
Nine papers were presented in this category, an invited paper thorough understanding.
described a broad range of experiments culminating in flight
Yip, NASA Langley, Vijgen, High Technology Corporation,
test (16). four papers dealt primarily with flow physics of
Hardin, Lockheed and van Dam, University of California
high-lift airfoils (2), (18), (19) and (21), and the remaining
(21). The authors described an ambitious flight research pro-
papers related to high Reynolds number testing in the wind
gramme to analyse the high-lift flap of a Boeing 737 aircraft.
tunnel and with specific combat aircraft configurations.
An outboard wing section, containing slat and triple slotted
Thibert, ONERA, France (16) - Invited. A collaborative Fowler flap elements, was heavily instrumented - including
European test programme in high-lift (known as the GAR- surface pressures, Preston tube devices and flow - visualiza-
TEUR action group AD-AG08) was performed in the years tion by means of tufts. Variation in skin friction coefficient
1985-1989: this work was described at the symposium by on the slat upper surface showed a marked variation with
J.J. Thibert, chairman of the action group. The lack of high angle-of-attack which could be interpreted in terms of re-
quality 2D and 3D experimental data was evident and, there- laminarization. Similar correlations between Preston tube
fore, a comprehensive series of tests was undertaken to help readings, surface pressures and tufts, were observed to deter-
close this gap, with special emphasis on correlation between mine the state of the boundary layer on the main wing and
2D and 3D wind tunnel tests, and between 3D tunnel and flap elements. Two well established multi-element computer
flight measurements. Further objectives included determina- codes were used to predict pressure distribution - both codes
tion of the separate effects of Mach number and Reynolds incorporated an integral boundary layer formulation.
number (with model Re overlapping flight values), a study Comparisons were made between CFD and flight, thus
of the effect of wing sweep on transition location as a func- adding a further dimension to the usefulness of the research
tion of Reynolds number, and the identification of technolo- programme overall.
gy gaps and facility needs. By way of description, suffice it
Valarezo and Dominik, Douglas Aircraft and McGhee and
to say that the research was focussed on a section of the Goodman, NASA Langley, United States (IO). The paper
A310 wing, represented by two 2D models and one half
describes tests on two high-lift configurations in the NASA
model, with some tests being duplicated in more than one
Langley low turbulence pressure tunnel at Reynolds num-
wind tunnel. These results were supported further by a series
bers in the range 5 to 16 x IO". Favourable lift increments at
of flight measurements. The author concluded that the pro-
approach angles of attack were experienced by optimization
gramme had generated a most comprehensive, well integrat-
of leading edge slat and introduction of a trailing edge
ed and accurate body of data dealing with high-lift airfoils wedge. Equally important was the fact that a set of high
and wings which will form the basis for substantial improve-
quality tests provided further evidence of Reynolds number
ment in design. and Mach number effects on high-lift optimization.
The importance of a deeper understanding of flow physics
Eamshaw and Green, DRA, United Kingdom (20). This
was raised by several speakers; four papers dealt with this
paper dealt with the use of half models in high-lift wind tun-
aspect of high-lift in particular.
nel testing. The use of half models can offer several advan-
A paper (2) by Alemdaroglu, METU, Turkey, described tages, not the least of which is to permit an increase in
detailed flow measurements in and around the slat and flap Reynolds number as compared to a full model in the same
cavities of a three element foil. Effects of Reynolds number wind tunnel. The author conducted tests on one half of an
on the mean flow and on the turbulence parameters were existing full span model which had previously undergone
investigated using pressure probes, hot-wire anemometers several tests. Various half model mounting configurations
and a laser Doppler velocimeter. were investigated and results compared to the full model
results. The paper generated a great deal of interest on
Hardy, DRA, United Kingdom (18), gave an account of
account of the very practical nature and potential usefulness
attachment-line transition and possible re-laminarization on
of the findings.
the leading edge slat of a combat type model. Similar occur-
rences have been observed on transport type wings, so the Finally, in the category of high-lift experiments, two papers
subject is of broad interest. It was determined that the dealt with low aspect ratio planforms suitable for combat
boundary layer can go through more than one change in state type aircraft. One by Walchli, USAF WAFB (22), in fact,
within an incidence range of a degree or so which makes for dealt with the X-29 forward swept wing demonstrator air-
difficulty in the prediction of Reynolds number variation. craft. Mention was made of high-lift capability required for
Wing root effects and spanwise variations serve to compli- manoeuvrability and under post stall conditions. Apart from
 T-5

the intrinsic interest of the paper itself, it served as a methods proved successful in the determination of near-opti-
reminder that the Banff symposium focussed on only one mum selection of flap profile, slat shape and pivot position.
segment of the subject of high-lift as a whole.
Mathews described theoretical methods in use at Short
The other paper associated with combat aircraft (17) was Brothers to define flap geometry at the project stage of
presented by Moir, DRA, United Kingdom and described design with special consideration of flap overhang and gap.
tests on a generic windfuselage model of aspect ratio 3.4 Also, some structural aspects of flap design were discussed
tested in the DRA Farnborough 5 metre pressurized tunnel and a simple four-bar flap linkage mechanism was shown.
with a Reynolds number range of 6 to 11 x approxi-
Eggleston and Poole, de Havilland Inc., Canada (32)
mately. The tests were designed to study leading edge slat
described R and D high lift work at de Havilland geared
optimization in conjunction with both a plain wing trailing
toward the development of future propeller driven regional
edge and a single-slotted flap. The important findings from
transport aircraft which employs more advanced airfoil sec-
the test relate to the performance of the model under condi-
tions suited to cruise speeds up to Mach 0.7. 2D models
tions of separately varying Mach number and Reynolds
were tested at high Reynolds number (6 to 10 x 10-6) in the
number, that is, the kind and quality of measurements need-
0.381~1x 1.5m test section of the IAWNRC pressurized wind
ed to help establish an advanced data base.
tunnel. Extensive use was made of half-model testing at
3.5 Design of High-Lift Systems for Complete Aircraft
lower Reynolds number.
The final session of the symposium was devoted to consider- The paper by Flaig and Hilbig, Deutsche Airbus, Germany
ation of the aerodynamic design procedures and practices (3 1) was representative of a manufacturer of large transport
found in industry, as related to transport aircraft. aircraft. It was comprehensive in scope dealing, at some
Development of a complete system involves a mix of design length, with the objectives and constraints of high-lift design
philosophy, empiricism, theory and experiment and, over the and then with the design methods in terms of theoretical pre-
years, the methodology evolved to become an established diction and wind tunnel strategy. These general considera-
and proven procedure which is peculiar to the design team tions were concluded with an outline of the high-lift design
and the company. Over the past decade, the design process process as made up of three phases, namely, pre-develop-
has undergone some change whereby less reliance is now ment, development and pre-flight. The role of theoretical
placed upon the empirical data base and more upon theoreti- prediction was clearly identified in the pre-development
cal prediction - the rate of this change is, in part, a function phase. Finally, the Deutsche Airbus design process was
of resources available and it is the larger companies which illustrated in terms of the development of the high-lift sys-
tend to move more rapidly toward acceptance of a new tech- tem for the Airbus 321. In particular, it was shown that the
nology. significantly increased lift requirements for this stretched
The paper by Obert and Fokker Aircraft, The Netherlands, version of the A320 could be satisfied by a conversion from
(27), was entitled “Forty Years of High-Lift R and D - An a single-slotted Fowler flap to a part span double-slotted
Aircraft Manufacturer’s Experience”. A review such as this, flap, even with the geometric constraints imposed by a
“minimum change” philosophy.
not only shows the progression of aerodynamic high-lift
technology at large, but also illustrates the evolution of the Finally, dealing with the category of high-lift systems for
design procedure as guided by the design team: in this case, complete aircraft, a paper was presented by Nark, Boeing,
within the framework of a medium sized manufacturer. As United States (30) relating to powered-lift for transport cate-
such, the paper represents a valuable addition to the litera- gory aircraft. The author gave a well illustrated presentation
ture for students and professionals alike. Regarding theoreti- describing the design, development and flight evaluation of
cal prediction, the author concluded that the role of CFD in the Boeing YC-14, this being the “upper surface blowing”
the design process has greatly increased but that, at present, entry in the USAF “Advanced Medium STOL Transport”
a fair degree of empirical data has to be fed into the compu- development programme. In particular, it was evident that
tational methods and therefore access to a large empirical methods and procedures for development of a powered-lift
data base remains a pre-requisite. system differ widely from those well established for a pas-
sive lift system. The paper was of special interest to many of
Next there were three papers which illustrated, in particular,
the application of CFD methods to the design of a specific the European delegates who were not aware of the technical
high-lift system: Capbem, ACrospatiale, France (28); background to the YC-14 and’YC-15 programmes.
Averado and Russo, Alenia, Italy (33), and Mathews, Short
4. CONCLUSIONS
Bros., United Kingdom (29).
Inviscid-viscous coupled CFD prediction methods are well
Capbem described the development of flap systems for the established and play an important role in the design of high-
ATR 72 aircraft and explained how numerical methods lift systems for transport aircraft (39). Under active develop-
(introduced in 1985) have served to complement existing ment, are more advanced field methods which have the
empirical and experimental methods which were used exclu- potential to deal adequately with regions of separation and
sively until then. He outlined the computational tools and the wakes but, in their present form, do not accurately predict
overall methodology and showed how this has led to a more some of the high-lift parameters of interest. These methods
simple design for the ATR 72 flap system with additional have not yet reached a level of maturity to allow widespread
benefits in terms of improved performance and lower pro- industrial use and therefore further development to this end
duction costs. is recommended (1 9).
Similarly, Averado presented a flow chart of the design pro- High quality, high Reynolds number experimental data con-
cedure used at Alenia showing the complementary roles of tinue in short supply - as needed for test cases (comparing
theoretical prediction and wind tunnel testing. The design CFD prediction with experiment) or to upgrade the existing
T-6

empirical data base for moderate to high aspect ratio config- 1. Woodward DS and Lean DE
urations. Where is High-Lift Today? - A Review of Past UK
Research Programmes
There remains an inadequate understanding of the flow
physics relating to high-lift systems and, therefore, further 2. Alemdaroglu N
specialized wind tunnel and flight programmes are needed to Experimental Investigation of the Flow about a Multi-
explore some,of the fundamental issues associated with tur- Element Airfoil
bulence, boundary layers and wakes which are, as yet, not
fully amenable to computational analysis (19), (40). There 3. Graham JMR and Giannakidis G
was clear consensus that improvement is needed in turbu- Numerical Simulation of the Separated Flow in Wing
lence modelling and that a better understanding of flow Coves
physics would lead to improvement in CFD methods. 4. De Ponte S, Cella A and Marcazzan M
A gap remains between the Reynolds number achieved in A Design Procedure for Slotted Flaps
the wind tunnel as compared to flight. It is unlikely that new 5. Fritz W
high Reynolds number wind tunnel facilities will become Maximum, and High-Lift Characteristics of Multi-
available in the near future, but this shortfall can be offset, in Element Aerofoils
part, by more intensive use of CFD prediction methods.
Such a capability helps in the formation of wind tunnel pro- 6. Bartsch P, Nitsche W and Britsch M
grammes and interpretation of wind tunnel and flight mea- Navier-Stokes Computations of Turbulent Flow around
surements: in this latter role, CFD becomes a kind of phan- High-Lift Configurations
tom flow visualization in providing basic insight to the flow 7. Rogers SE, Wiltberger NL and Kwak D
processes at work. Efficient Simulation of Incompressible Viscous Flow
Little indication was given concerning performance over Multi-Element Airfoils
improvement expected as a result of advancement in high- 8. Jasper DW, Agrawal S and Robinson BA
lift technology (say, in terms of C, max of U D for second Navier-Stokes Calculations on Multi-Element Airfoils
segment climb). Similarly, the conference did not bring to using a CHIMERA-Based Solver
light any new or unorthodox ways to improve performance
but inferred that improvement would come as a result of 9. Nelson TE, Zingg DW and Johnston GW
careful refinement. Even quite modest improvement in high- Numerical Solution of the Navier-Stokes Equations for
lift can lead to significant gain in performance of transport High-Lift Configurations on Structured Composite Grids
type aircraft ( 19). 10. Valarezo WO, Dominik CJ, McGhee RJ and Goodman WL
In short, the design and development of a modem high-lift High Reynolds Number Configuration Development of a
system requires vigorous use of the wind tunnel as High-Lift Airfoil
described, for example, by Obert (27) and Flaig (31). CFD 11. Le Balleur JC and Ntron M
methods play an important support with respect to design -
Une MCthode d’Interaction Visqueux non-Visqueux
at the pre-development stage (3 l), and with respect to inter-
pour Ecoulements Incompressibles HypersustentCs sur
active analysis - at the development and pre-flight stages
Profils Multi-corps en RCgime de DCcollement Profond
(36).
12. De Cock KMJ
The conference served to bring together mathematicians,
High-Lift System Analysis Method using Unstructured
aerodynamicists and test engineers to help foster a better
mutual understanding of strengths and weaknesses in each of Meshes
the disciplines. The present status of high-lift technology for 13. Johnson LJ and Stolcis L
transport aircraft was clearly set forth and areas in need of Prediction of the High-Lift Performance on Multi-
further attention became evident. Element Aerofoils using an Unstructured Navier-Stokes
Solver
5. PROGRAMME COMMITTEEMEMBERS
Dr D.S. Woodward (Co-chairman), DRA, United Kingdom. 14. Bailey R, Radespiel R, Demier A, Rossow C-C, Longo
Mr L.J. Williams (Co-chairman), NASA, United States. JMA, Ronzheimer A and Kroll N
Prof. R. DeCuypere, Ecole Royale Militaire, Belgium. Numerical Calculations of High Lift Flows using
Dr L. Chan, IAR, Canada. Structured and Unstructured Methods
M.J. Bousquet, ACrospatiale, France. 15. Chow R, Chu K and Carpenter G
Dr B. Wagner, Dornier Luftfahrt GmbH, Germany. Navier-Stokes Simulation of Flow Field around a Blow-
Lt Col. D. Petridis, Hellenic Air Force, Greece. Flap High Lift System
Prof. M. Onorato, Politecnico di Torino, Italy.
Prof. Dr Ir J.L. van Ingen, Delft University of Technology, 16. Thibert JJ
The Netherlands. The GARTEUR High Lift Research Programme
Dr L.P. Ruiz-Calavera, INTA, Spain. 17. Moir IRM
Dr W.J. McCroskey, US Army Aeroflightdynamics, United An Experimental Investigation of the Optimum Slat
States. Setting on a Combat Aircraft Model
6. REFERENCES 18. Hardy BC
(Note that references ( I ) through (33) list the symposium An Experimental Investigation of Attachment-Line
papers in the order in which they were presented.) Transition on the Slat of a Combat Aircraft Model
 T-7

19. Meredith PT 30. Nark TC

Viscous Phenomena Affecting High-Lift Systems and Design, Development, and Flight Evaluation of the
Suggestions for Future CFD Development Boeing YC-14 USB Powered Lift Aircraft
20. Earnshaw PB, Green AR, Hardy BC and Jelly AH 3 1. Flaig A and Hilbig R
A Study of the Use of Half-Models in High-Lift Wind High-Lift Design for Large Civil Aircraft
Tunnel Testing
32. Eggleston BE and Poole RJD
21. Yip LP, Vijgen PMHW, Hardin JD and van Dam CP High-Lift Systems for Transport Aircraft with Advanced
In-Flight Pressure Distributions and Skin-Friction Airfoils
Measurements on a Subsonic Transport High-Lift Wing
Section 33. Averardo MA, de Leo M and Russo V
Choice and Optimization of a High-Lift System for an
22. Walchli LA
High-Lift and the Forward Swept Wing Advanced Amphibious Aircraft

23. Jacob K 34. Smith AM0

A Fast Computing Method for the Flow over High-Lift Aerodynamics of High-Lift Airfoil Systems. AGARD-
Wings CP-102, April 1972
24. Cebeci T 35. Butter DJ
Calculation of Multi-Element Airfoils and Wings at Recent Progress on Development and Understanding of
High-Lift High-Lift Systems. AGARD-CP-365
25. Baron A and Boffadossi M 36. Dillner B, May F W and McMasters JH
Wake Structure and Aerodynamic Behavior of High-Lift Aerodynamic Issues in the Design of High-Lift Systems
Aircraft Configurations during Unsteady Maneuvers in for Transport Aircraft. AGARD-CP-365
Ground Effect
37. Chow R and Chu K
26. Le Balleur JC Navier-Stokes Solution for High-Lift Multi-Element
Calcul par Interaction Visqueux non-Visqueux des Airfoil Systems with Flap Separation. AIAA-9 1- 1623
Ecoulements Compressibles Fortement DCcollCs aux
Grandes Portances sur Profils d’Ailes et Voilures 38. King DA and Williams BR
Developments in Computational Methods for High-Lift
27. Obert E
Forty Years of High-Lift R&D - An Aircraft Aerodynamics, R.Ae.S. Aeronautical Journal, AudSept
Manufacturer’s Experience 1988

28. Capbern P 39. Kusunose K, Wigton L and Meredith P

High-Lift Research: Application to the Design of the A Rapidly Converging Viscous/Coupling Code for
ATR72 Flap Multi-Element Airfoil Configurations, AIAA-9 1-0177
29. Mathews JR 40. Gatner PL, Meredith PT and Stoner RC
The Aero-Mechanical Design of a Novel Fowler Flap Areas for Future CFD Development as Illustrated by
Mechanism Transport Aircraft Applications, AIAA-9 1- 1527-CP
 1-1

WHERE IS HIGH-LIFT TODAY? - A Review of past UK

Research Programmes
D. S. WOODWARD (DRA Farnborough UK)
D. E. LEAN (DRA Farnborough UK)

SUMMARY

Some of t h e h i s t o r y of t h e development o f s l o t t e d h i g h - l i f t
s y s t e m s i s reviewed i n t h i s p a p e r . I n p a r t i c u l a r , t h e N a t i o n a l
H i g h L i f t Programme r u n i n t h e UK d u r i n g t h e 1970s i s reviewed
i n some d e t a i l . I n a d d i t i o n , t h e r e s e a r c h programme i n h i g h
l i f t s i n c e t h e c o m p l e t i o n of t h e N a t i o n a l High L i f t Programme
i s described q u a l i t a t i v e l y a n d r e f e r e n c e s g i v e n . The c o n t e n t s
c o v e r t e c h n i q u e s of h i g h l i f t t e s t i n g , r e s u l t s o f p o s i t i o n a l
o p t i m i s a t i o n of s l a t s a n d f l a p s , t h e d e r i v a t i o n o f a s i m p l e
p r e d i c t i o n method s u i t a b l e f o r u s e w i t h a p r o j e c t o p t i m i s a t i o n
method, a n d t h e d e s c r i p t i o n , w i t h r e s u l t s , of a method f o r
i n t e r p r e t i n g aerodynamic and w e i g h t d a t a on h i g h l i f t s y s t e m s
i n t h e c o n t e x t of a complete a i r c r a f t .

1 INTRODUCTION

Barnes1 r e c o r d s t h e amusing s t o r y of how t h e s l o t t e d wing came t o be

invented:

" I n A p r i l 1 9 1 1 F r e d e r i c k Handley-Page r e a d t o t h e A e r o n a u t i c a l
S o c i e t y of G r e a t B r i t a i n , a p a p e r e n t i t l e d 'The P r e s s u r e s on P l a n e
and Curved S u r f a c e s moving t h r o u g h t h e A i r , , ... H e noted t h a t , with
a wing o f m o d e r a t e Aspect R a t i o ( 6 . 2 5 ) , l i f t i n c r e a s e d a p p r o x i m a t e l y
l i n e a r l y up t o 1 0 deg i n c i d e n c e , b u t t h e n l e v e l l e d o f f , w h i l e t h e
l i f t of a s q u a r e wing (Aspect R a t i o = 1 . 0 ) c o n t i n u e d t o i n c r e a s e w i t h
i n c i d e n c e up t o n e a r l y 4 0 deg .... Using t h e wind t u n n e l a t Kingsbury
i n 1 9 1 7 , Handley-Page and h i s a e r o d y n a m i c i s t R.O. Boswall t r i e d t o
combine t h e low-drag a d v a n t a g e of h i g h a s p e c t r a t i o w i t h t h e d e l a y e d
s t a l l of low a s p e c t r a t i o , by s e p a r a t i n g a normal narrow wing i n t o
d i s c r e t e s q u a r e a r e a s by means o f c h o r d w i s e s l o t s , b u t f o u n d t h e
r e s u l t d i s a p p o i n t i n g and i n c a p a b l e of b e i n g improved by v a r y i n g the
p r o p o r t i o n s o f the s l o t s and t h e i r d i s t r i b u t i o n s p a n w i s e . A t some
p o i n t i n t h e s e e n t i r e l y e m p i r i c a l e x p e r i m e n t s , someone - w h e t h e r
Handley-Page h i m s e l f , o r Boswall, o r one of t h e c a r p e n t e r s , i s n o t
c l e a r - h a d t h e i d e a of c u t t i n g a s p a n w i s e s l o t p a r a l l e l t o t h e
l e a d i n g - e d g e , s l o p i n g upwards and r e a r w a r d s , a t a b o u t t h e 1 / 4 - c h o r d
line. The f i r s t e x p e r i m e n t , u s i n g a RAF15 a e r o f o i l s e c t i o n , g a v e a
s p e c t a c u l a r l i f t i n c r e a s e of 25%; w i t h a n improved s h a p e of s l o t i n a
RAF6 s e c t i o n , t h i s was r a i s e d t o o v e r SO%, w i t h o n l y a s l i g h t
increase i n drag ..... Numerous e x p e r i m e n t s c o n t i n u e d d u r i n g 1 9 1 8 and
1 9 1 9 , u s i n g v a r i o u s s h a p e s of s i n g l e s l o t , whose c h o r d w i s e l o c a t i o n
was found t o be c r i t i c a l on t h i n a e r o f o i l s ....
Handley-Page r e a l i s e d e a r l y on t h a t t h e s l o t t e d wing was an
extremely valuable invention, so he kept t h e p r i n c i p l e s t r i c t l y
secret u n t i l p a t e n t a p p l i c a t i o n s had been a c c e p t e d by b o t h B r i t i s h
and U n i t e d S t a t e s P a t e n t O f f i c e s . To a v o i d p r e m a t u r e d i s c l o s u r e , h e
d i d n o t s e e k p r o t e c t i o n f o r t h e o r i g i n a l f i x e d s l o t u n t i l h e was
r e a d y a l s o t o c o v e r a moveable s l o t , which c o u l d be opened and c l o s e d
by t h e p i l o t .....
D u r i n g t h e whole p e r i o d of e a r l y development o f t h e Handley-
Page s l o t t e d wing, a p a r a l l e l i n v e s t i g a t i o n h a d been made, q u i t e
1-2

independently, by the German engineer-pilot, G.V. Lachmann. He had

transferred from the cavalry to the flying corps in 1917, and had
stalled and spun-in during an early training flight, breaking his
jaw. In hospital he had time to ponder on the cause o f his accident
and on means of preventing stall, concluding that a cascade of small
aerofoils within a normal wing profile might have better properties
than the equivalent single aerofoil. To study this idea, he made a
simple wing model and rigged up an electric fan to entrain cigarette
smoke over it, so as to obtain a crude form of flow visualisation.
Satisfied with the result, he drafted a patent specification as early
as February 1918, some weeks before Handley-Page's first application,
but.. . was unable to convince the German Patent Office that the
invention would work .... It was quite by chance that he saw an
account of a demonstration by Handley-Page on 21 October 1920, and
immediately challenged his priority in the invention. Still the
German Patent Office refused to consider his claim....unless he could
prove his invention experimentally. He then approached Professor
Ludwig Prandtl at Gottingen, who agreed to do the necessary wind
tunnel tests for 1000 marks (€50); Lachmann had nothing like this sum
to spare, but his mother lent it to him, and the result ... convinced
the patent examiners, who then granted the application backdated to
February 1918."

With an origin like that, it is perhaps not surprising that the

aeronautical community should, for 54 years, have a totally incorrect view
of the physical principles underlying its operation. Until the publication
of A . M . O . Smith's classic paper2 in 1972, it was accepted widely that the
slot performs as a boundary layer control d e ~ i c e ~- -and
~ this is not true.

Smith showed clearly that the device works because it manipulates the
inviscid pressure distribution - and that is why it is such a powerful
effect. The data of Foster, Ashill and Williams7 illustrated in Fig 1
shows the flow over a typical high-lift aerofoil with a leading-edge slat
and a single-slotted flap. It will be seen that the trailing-edge of the
slat sits in the high-velocity region of the flow around the leading-edge
of the main wing; because of this, the pressure co-efficient at the
trailing-edge is significantly negative and thus the pressure rise on the
slat is reduced. The same happens at the trailing-edge of the main wing
due to the high velocities around the leading-edge of the highly-deflected
flap. In addition, the circulation around the slat induces a downwash on
the main wing; this downwash clearly reduces with distance from the slat,
so that it modifies the local velocities most strongly near the leading-
edge of the wing, reducing its peak suction markedly. The same mechanism
operates near the leading-edge of the flap. As a result of this, the
pressure rise to the trailing-edge of the overall wing is split up into a
number of smaller pressure rises;. when each of these is sufficient to just
cause separation of the boundary layer, the overall pressure rise can
clearly be very large.

Fig 1 also shows the boundary layers flowing through the slots.
Clearly if the two elements ( e g slat/wing or wing/flap) are too close
together, the boundary layer on the aft element will merge with the wake
from the upstream element.to form a thick boundary layer, and separation,
far from being delayed, will be provoked. Thus, in contrast to earlier
ideas, the flow through the slots acts to i n c r e a s e the adverse viscous
effects rather than to alleviate them.
 1-3

2 THE UK NATIONAL HIGH-LIFT PROGRAMME

2.1 Introduction

In 1968, K.J. Turner completed a secondment to the Defence Staff in

the British Embassy in Washington; during his three years in the US, he had
become increasingly aware of the advances being made by the US Industry in
the exploitation of high-lift devices. The comparisons between the "slat +
3-slot flap" of the B727-100, and the "droop-nose + 2-slot flap" of the HS
Trident 1, or between the 'plain leading-edge + l-slot flap" of the BAC
1-11/500 and the "slat + 2-slot flap" of the DC9-30, are indicative of the
contemporary differences which worried Turner. At the same time, he was
aware of the progress being made in the US in the understanding of the
basic physics of how slotted wings worked. On his return to the UK, he put
forward the concept of a "National High-Lift Programme" to be run as a
collaborative exercise between the RAE (now DRA) and Industry. Government
would provide most of the funding through RAE, whose role would be to
participate strongly, but also to manage the programme and attempt to
distil the general conclusions from the mass of data.

The programme had two major objectives:

(1) to provide sufficient data to enable the most appropriate high-

lift system to be selected for any given project;

(2) having made this selection, to provide adequate procedures and

information to permit the detailed design of systems that were
efficient in aerodynamic, mechanical, structural, and weight terms.

These objectives lead to a programme that had a number of distinctive

features:

(a) a research programme into the structural and mechanical design

of high-lift systems would be run in parallel with the aerodynamic
programme in order to:

(i) provide a "research" environment in which new, innova-

tive, lighter, and more reliable mechanisms could be
invented;
(ii) provide data on the weights of high-lift systems of
varying complexity.

This research programme would cover high-lift system installations on

a typical civil transport wing, and also on the wing of a typical
"swing-wing" combat aircraft

(b) the aerodynamic programme would aim to produce data at two

levels:

(i) data of a fundamental nature which would cast light on

the dominant physical processes . involved and might lead
to improved theoretical prediction methods;
(ii) data of direct use -in project design.

(c) all the high-lift devices were to have the shapes of their
'cut-lines" eg regions "m" and 'n" in Fig 1, designed, using the
A.M.O. Smith multi-aerofoil program, to have "sensible" pressure
distributions without large, and very sharp, suction peaks; the
1-4

locations of the various elements to be achieved in the structural

and mechanical design studies, were also to be decided on the basis
of "sensible" pressure distributions calculated from the A.M.O. Smith
program.

2.2 The Structural/Mechanical Design Programme

The complete set of leading- and trailing-edge devices embodied in

the programme are defined in Tables 1 and 2 (located at the end of the
text) and shown schematically in Fig 2. The tables indicate which of these
devices were considered in the context of:

(a) a civil transport aircraft,

(b) a military combat aircraft.

Within the structural design studies issues of reliability, fail-safe

design, and weight- and cost-saving were all addressed in addition to the
normal stiffness and strength considerations. In some cases stiffness was
particularly important in order to ensure that the designed aerodynamic lap
and gap settings were maintained within tolerance under load.

At the end of the structural design programme, each of the designs

was re-visited again in the light of the experience gained in the rest of
the programme, and the designs and the weights were revised. This lead to
a weight prediction method* that formed an important part of some of the
final assessment work.

2.3 The Aerodynamic Research Programme

The aerodynamic research programme was conducted on four different

wind tunnel models:

(i) a Two-dimensional model - shown in Fig 3; it had suction on

the side walls to remove the side-wall boundary layer and ensure very
good two-dimensional flow; it had 2 rows of chordwise pressure
tappings -
one at mid-span, and one near the wall, so that the two-
dimensionality of the flow could be checked; results were obtained by
integration of the pressure distributiong for lift and pitching
moment, and by wake traverse for drag; the geometric Aspect Ratio of
the undeployed planform was 3.2;

(ii) an Endplate Model - shown in Fig 4 ; the model was hung from
an overhead mechanical balance, the results from which formed the
main output, but there was also a chordwise row of pressure tappings
at mid-span, and it was intended that profile drag could be obtained
by wake traverse; the geometric Aspect Ratio of the undeployed
planform was 3.0;

(iii) a variable sweep Swept Panel Model .- shown in Fig 5; this

1/2-model was mounted on struts from an underfloor mechanical balance
using a false wall as a reflection plane, one strut being enclosed in
the false wall; the wing could be set at a number of sweeps in the
range 15 deg to 36 deg and had one row of pressure tappings which was
streamwise at 31 deg sweep; the wing and high-lift devices had the
same profiles normal-to-the-leading-edge as Models (i) and (ii); the
geometric Aspect Ratio of the undeployed planform at 31 deg sweep was
5.0;
 1-5

(iv) a variable sweep 1/2 model with a tapered planformMode1 477 -

shown in Fig 6; the wing could be set at 1/4-chord sweeps of 20 deg,
30 deg, and 45 deg: this model was mounted on an underfloor
4-component mechanical balance, which was in turn hung from the floor
turntable, and used the tunnel floor as a reflection plane without
any form of boundary layer control; it had pressure pipes run
spanwise so that holes could be drilled in them at a number of
spanwise locations to enable the spanwise development of the pressure
distribution to be explored; the wing and high-lift devices had the
same streamwise profiles at 30 deg sweep, as Model (iii) at 31 deg
sweep; the geometric Aspect Ratio of the undeployed planform at 30
deg sweep was 7.0.

All four models featured slat, and flap, supports that permitted
extensive variation of the fore-and-aft and height positions of the flap
and slat elements. Within the programme these movements were conven-
tionally known as "lap" and "gap" even though that is not a precise
description; for simplicity this nomenclature will be retained in this
paper. Within the Programme, it was soon found that the definition of
angles, origins, and Cartesian axes describing the locations of the
elements of the high-lift system was a major problem, because initially
each member had his own way of doing it. The systems chosen are defined in
Figs 7, 8, and 9.

Boundary layer profiles were measured on all four models at some time
during the programme using traverse mechanisms mounted external to the
models; these caused some interference with the flow, but in no case was
this excessive.

The fourth model formed the contribution from RAE, and the other
three models formed the contribution from industry. About a quarter of the
way through the Programme, two other models were added to the aerodynamic
programme -a "wall-to'-wall" two-dimensional model and a complete model
with a planform typical of a strike fighter - and these were aimed at the
evaluation and development of high-lift systems based on Boundary-Layer-
Control (BLC) by blowing with low-pressure a2r derived from the fan stages
of a turbofan engine. The results of this additional part do not form part
of this paper.

3 RESULTS FROM THE AERODYNAMIC RESEARCH PROGRAMME

3.1 Technique Development - Comparison of the Results from

the Different Models

3.1.1 Two-dimensional vs Endplate Models

It was known from previous testsl0,l1 that the two-dimensional tech-

nique used gave very good two-dimensional conditions - but the test tech-
nique, and the subsequent data reduction, were extremely time-consuming -
and therefore expensive. The Endplate model, on the other hand, provided
the data required quickly, easily and accurately from the balance. The
question nevertheless remained - was the flow on the Endplate model two-
dimensional?

On both models, flow visualisation indicates two-dimensional flow,

and comparison of the forces obtained from the integrated pressure
distributions at the centre, and near the wall, or the Endplate, shows less
than 10% reduction in lift as the end of the wing is approached. However,
1-6

comparison of the centre-line pressure distributions in Fig 10 shows that

these are radically different. It is clear, despite the effects of the
Endplates in maintaining the loading across the span of the wing, that the
downwash from the trailing vorticity increases quite strongly across the
chord, leading to higher loadings near the leading-edge, and lower loadings
towards the trailing-edge.

At first sight, it seems probable that this flow development rules

out the use of the Endplate model for serious high-lift research. However,
if we form the ratios of the lift at optimum locations of the slat and flap
to those achieved at off-optimum l o c a t i ~ n s ~ -
~ ras~ ~ * ~ ~in Table 3
shown -
it can be seen that the Endplate Model seems to produce the same optimum
locations as the two-dimensional model. This is an important result
because of the financial and time implications on high-lift research - and
it underpins much of the National High-Lift Programme.

Table 3
Comparison of the Effect of Flap and Slat Movements on
the Endplate and "Wall -to-Wall" Models
---
XS ZS Xf Zf CL 2-D CL E/P CL 2-D CL E/P

% %
% %
-- - ~
CL 2-D top row
~~

SLAT L1 AT 25 DEGREES
-1.5 2.5 4.15 3.85 1.000 1.000
-1.0 2.0 4.16 3.84 1.002 0.997
-0.5 1.5 4.02 3.67 0.969 0.953
0.0 1.0 3.52 3.20 0.848 0.833
FLAP T2 AT 20 DEGREES
0.0 2.3 4.13 3.95 1.000 1,000
2.0 2.3 4.09 3.89 0.990 0.985
4.0 2.3 3.97 3.80 0.961 0.962
0.0 0.5 4.05 3.86 0.981 0.977
0.0 5.0 4.11 3.92 0.995 0.992
FLAP T7 AT 45 DEGREES
-1.0 2.0 4.86 4.62 1.000 1.000
-0.5 2.0 4.81 4.58 0.990 0.991
+0.5 2.0 4.70 4.48 0.967 0.970
--- -1.0 1.0 4.81 4.58 0.990 0.991

The strong trailing vorticity behind the Endplate model has another -
and more unfortunate - consequence. The vorticity distorts and stretches
the wake as it passes downstream, with the result that the drag calculated
from a wake traverse varies considerably with the downstream position of
the traverse, as shown in Fig 11. Because of this, the results from Wake
Traverse measurements are of little use, and their collection was quickly
discontinued within the Programme.

As a consequence, information on the profile drag characteristics of

high-lift devices had to be derived from measurements of overall drag
obtained from the balance on which the Endplate Model was mounted. By
correlation between these drag values and true profile drag measurements on
 1-7

the "Wall-to-Wall" model for the same configuration, values of the vortex
drag on the Endplate Model could be derived. Unfortunately, the values of
vortex drag so-derived exhibited some dependence on device and deployment
angle; this meant that drag comparisons between similar configurations ( e g
between single-slotted flaps with 84%, 90%, & 96% shroud position, (Ll/Tl,
L1/TzfL1/T3]) could be made with some confidence, but that the
establishment of the profile drag increment between, say, a single-slotted
flap at 20 deg and a triple-slotted flap at 45 degf(L1/T8(45) - Ll/T2(20)),
involves some uncertainty.

3.1.2 Two-dimensional vs Swept Panel Model

The Swept Panel Model was included in the Programme to give

information on the effect of sweep on the performance of high-lift devices
and, more importantly, on the optimum location of slats and flaps. In
order to interpret the data it was important to know the relationship
between the pressure distribution on this wing and the two-dimensional
wing; clearly with such a simple wing the two pressure distributions ought
to coincide when both are referred to the velocity normal to the leading-
edge (simple-sweep theory). Fig 12 shows the same type of discrepancy as
shown by the Endplate model - and for the same reason. The low Aspect
Ratio - chosen to maximise the Reynolds number in a small tunnel - has
resulted in a significant variation, across the chord, of the downwash from
the trailing vorticity. Fig 13 supports this by showing that, for each of
slat/wing/flap, the pressure distributions at the various sweeps collapse
reasonably well when referred to the velocity normal to the leading-edge;
it is reasonable to assume that the residual differences are due to the
changing Aspect Ratio with sweep.

3.1.3 Two-dimensional vs Model 477

The planform of Model 477 was chosen to:

(a) have a high enough Aspect Ratio to provide some part of the
span which would give a close approximation to "Infinite Swept-Wing"
conditions;
(b) resemble closely the part of a typical transport aircraft wing
outboard of the crank.

Once again, it was important to establish how well these aims had been met.
Referring the pressure distribution at mid-semi-span to the velocity normal
to the 1/4-chord sweep, produced a fairly poor agreement with two-
dimensional. However, when the local pressure was referred to the velocity
normal to the l o c a l sweep, very good agreement was obtained - as shown in
Fig 14. This concept is further supported by Fig 15 which shows the
agreement between the "local-sweep normalised" pressure distributions at
sweeps of 20 deg, 30 deg, and 45 deg.

3.2 Optimisation of Leading-edge Slats

Section 1 described, in qualitative terms, the effects on the

pressure distributions on the slat and wing arising from their mutual
interaction. Considering one of the "standard" NHLP configurations
[L1(25)/T2 (20)] consisting of:
1-8

(a) the 12.5% chord leading-edge slat (L1) set at 25 deg,

(b) the 33% chord trailing edge flap with the flap shroud at 90%
chord (T2) set at 20 deg,

then Figs 16a and 16b show the response of:

(a) the trailing-edge pressure,

(b) the peak suction,

on the slat as the trailing-edge is moved over a grid of points (Xs,Zs),

such that -5% < Xs < 0% and 0% < 2s < 4 % - see Fig 7. The calculations
have been done at a constant fixed incidence close to that for the maximum
lift observed experimentally.

Fig 16a demonstrates the strong variation of the trailing-edge

pressure to the movement of the slat through the flow-field around the
leading-edge of the wing; for example, with the slat 1% chord ahead of the
wing leading-edge (Xs = -1%) the slat trailing-edge pressure changes from
Cp =-6 at a height 2 s = 4 % above the leading-edge, to Cp = -1.3 at a height
1% below the leading-edge ( 2 s = -1%). In contrast, the peak suction on the
slat (Fig 16b) varies relatively slowly with change of height, but changes
rapidly with variation of the lap (Xs) - for instance, at a height of
Zs=l%, moving the slat forward from Xs = 0% to Xs = -5% reduces the peak
suction from Cp = -29 to Cp = -19.

The variation of the slat peak suction on its own is important

because the levels are very high, and hence compressibility effects on
maximum lift can be experienced down to Mach numbers as low as 0.15 . The
difference between the peak suction and the trailing-edge pressure
represents the pressure rise experienced by the boundary layer over the
upper surface of the slat, and hence determines the thickness of the slat
wake, or even separation of the flow on the slat. The variation of this
pressure rise determined from Figs 16a and 16b is shown in Fig 17, and Fig
18 shows the corresponding pressure rise over the wing. Together these
show that, as the slat is moved:

(i) vertically ( i e varying Zs, constant XS):

(a) the pressure rise over the wing exhibits a m i n i m u m value;

clearly this is the best position for the slat in order to
delay separation of the boundary layer on the wing,
(b) the pressure rise over the slat exhibits a m a x i m u m value
at about Zs = -l%, and decreases as 2 s increases - primarily
due to the negative change in the trailing-edge pressure

(ii) horizontally forwards ( i e decreasing Xs, constant Zs):

(a) the level of the wing pressure rise i n c r e a s e s markedly,

(b) the level of the slat pressure rise d e c r e a s e s markedly.

From these results, we can extract contours of constant slat and wing
pressure rise, and constant slat peak suction, and plot these in the Xs,Zs
plane. Suppose we assume that, to avoid boundary layer separation,

(a) the slat pressure rise IACpJ must be <22

(b) the wing pressure rise lACpwl must be <11
(c) the slat peak suction ICp,l must be <26
 1-9

then, with these assumptions, it can be seen that the optimum location of
the slat must lie within the boundaries indicated on Fig 19a. Similar
boundaries, for a slat angle of 20 deg, are drawn in Fig 19b. The experi-
mentally determined maximum lift contours are shown in Fig 2014; One of
these contours is also shown on both parts of Fig 19, and it can be seen
that, despite the simplicity of the model, it correctly indicates,

(a) the size and direction of the migration of the optimum point,
as the slat angle is changed,
(b) the reduction in the area of the contour, as the angle is
reduced from 25 deg to 20 deg, suggesting that the CLmax at the
optimum location is less at 20 deg than at 25 deg.

This confirms that the primary effect of the slot ‘is to manipulate the
inviscid pressure distribution.

However, it will also be apparent that the simple model used above,
does not give a very good representation of the shape of the experimental
maximum lift contours - in particular it gives no indication of the closure
of the contours at the top. This, of course, is not surprising in view of
the simplistic separation criteria used which clearly ignore the shape of
the pressure distribution, but these criteria also fail to take in to
account that the viscous layers on the wing and flap cannot be considered
in isolation, but that each is affected by the overlying wake from the
upstream element. Figs 21 and 22 show the development of the viscous
layers over the wing+slat when fitted with a single-slotted flap (Fig 21)
and with a double-slotted flap (Fig 22). In both cases, it can be seen
that the slat wake merges with the wing boundary layer causing significant
increases in the momentum deficit.

It is relatively easy, using modern CFD methods and coupling

techniques15, to couple a boundary-layer code to an inviscid code for
calculating the viscous flow around multiple aerofoils,. but this will not
yield good estimates of the variation of maximum lift with slat (or flap)
position unless the boundary-layer method is capable of dealing with the
merging of the boundary layer and wake.

3. 3 Optimisation of Trailing-edge Flaps

The physical processes involved in the determination of the optimum

location of trailing-edge flaps are exactly the same as those for leading-
edge slats - but they appear different because the balance between the
viscous and inviscid effects is very different. For trailing-edge flaps,
the viscous effects are much larger, and the inviscid effects are much
smaller than for slats.

If we replace the lifting effect of each element by a single vortex,

then these will be situated in the vicinity of the 1/4-chord of the
respective elements, and it will be clear that the vortices of the slat and
wing are relatively close together and there will be a significant
interaction between them leading to a considerable variation of downwash
from the slat across the chord of the wing; on the other hand those of the
wing and flap are relatively far apart, and the downwash from the wing will
vary little across the chord of the flap. Therefore to get any worthwhile
reduction in pressure rise over the flap, requires the leading-edge of the
flap to be very close to the trailing-edge of the wing. Clearly the limit
to their proximity comes from the merging of the flap boundary layer with
the wing wake.
1-10

Fig 23 shows the maximum lift contours from the optimisation of the
T2 flap at 20 degl’; the optimum position is clearly around zero lap and
the lift falls away as the flap is moved forwards or rearwards from this
position. Fig 24 shows that the optimum location depends on the leading-
edge device fitted, because, of course, the slat wake adds to the thickness
of the wing wake over the flap. At higher flap angles, eg Fig 25 for T2 at
40 deglg, the loss in lift as the flap is moved rearwards is more
pronounced, and large scale separation of the flap flow sets in for quite
small rearward movements. However, this flow also exhibits an even more
interesting phenomena; it appears that the wing wake, flowing over the
flap, also has the effect of reducing the peak suction of the flap -
presumably through its displacement effect. Thus, for some flap positions
(identified on Fig 25) the flow over the flap is separated at low incidence
when the wing wake is thin, but, at higher incidence as the wing wake
thickens, the flap flow suddenly re-attaches producing a characteristic
non-linear increase in lift and negative pitching moment, and a reduction
in drag. The incidence at which the re-attachment occurs gets higher a3
the flap moves away from the wing, until ultimately the flow fails to re-
attach at all, and a large decrement in maximum lift results. This type of
behaviour has also been observed during the optimisation of double-slotted
flaps17 and triple-slotted flaps1*.

It will be obvious that increases in Reynolds number will thin the

wing wake and exacerbate this behaviour; indeed this has been found to be
one of the prime sources of adverse Scale effect on high-lift s y s t e m ~ l ~ ~ ~ ~ .
Fig 25 shows that the potential losses in maximum lift are very large and
demonstrates why it is so important to conduct high-lift development at the
highest-possible Reynolds number.

3. 4 Comparison of Various Leading- and Trailing-edge

Devices

Fig 26 shows a set of leading-edge slats and Krugers, whose maximum

lift performance as the deployment angle is varied, is shown in Fig 2 7 .
This demonstrates that the maximum lift perf.ormance is not very sensitive
to the angle of the leading-edge device - so long as it is at its optimum
setting for that angle. The increments in maximum lift and profile drag
arising from these devices, are shown in Fig 28

Figs 29 and 30 show the lift curves for a double- and a triple-
slotted flap each set at three angles. It is noteworthy that the maximum
lift for both flaps varies very little with angle, but the incidence at
which a given proportion of the maximum is achieved, varies quite
considerably. This feature can be of great value on military aircraft
where pilot vision, undercarriage length and position, and weapon carriage,
can all combine to impose significant restrictions on the maximum incidence
usable in the approach.

3.5 Project Level Analysis of Results

The wealth of two-dimensional and swept wing data available from this
Programme raised the possibility of a major enhancement to the prediction
of maximum lift within the RAE Multivariate Optimisation Programmez1. A
major strength of the dataset was that it was possible to carry out
internal consistency checks, since maximum lift increments could be derived
in several different ways; for instance the increment in maximum lift due
to fitting a leading-edge slat (L1) could be derived from
 1-11

(a) Ll/TO - LO/TO

or (b) Ll/T2 - LO/T2

similarly, the increment due to the deployment of a trailing-edge flap (T2)

could be derived from

(a) LO/T2 - LO/TO

or (b) L1/T2 - Ll/TO.

It was soon clear that the increments so derived differed consider-

ably one from the other, and this was because:

(a) a straightforward subtraction fails to take into account

properly the differing area extensions;
(b) there is, in some cases, an interaction between the leading-
and trailing-edge devices, which leads to considerable uncertainty in
the evaluation of the increments.

To examine the first of these factors, let us consider the idealised

situation sketched in Fig 31, where:

(a) the C, is based, in the usual way, on the reference area of the
wing (ie with the high-lift devices undeployed); the deployment of
the high-lift devices then leads to an increase in lift-curve slope,
(b) it is assumed that deployment of the flap leads to a constant
change in zero lift angle,
(c) it is assumed that the deployment of the slat leads to a
constant increment in stalling angle - whether the flap is deployed
or not.

From the relations set out on the lower part of Fig 31, it is clear that
each of the values Kl, K2, K3, has to be based on the a c t u a l area involved,
and then the addition of all' the increments is brought back to the
reference area at the end. This approach is confirmed in Fig 32a where,
after the change in zero-lift angle has been taken out by subtracting the
value of CL at 5 deg incidence, the lift curves for the various combina-
tions of high-lift devices collapse very well.

To look at the second factor, it can be seen from Fig 32a that, in
common with most other sets of data, there is a significant decrement in
stalling angle when the flap is deployed alone - and this violates the
assumptions of Fig 31. This situation is examined more closely in
Fig 32b. Here the two curves involving the slat represent the optimum
angle in each case - and these demonstrate very similar stalling angles; if
the curve for the flap deployed alone is extrapolated to the same stalling
angle as the "wing alone", it can then be seen that there is complete
consistency in the maximum lift increments, no matter which way they are
evaluated. In this way, it can be seen that the ACL = 0.25 "slat/flap
interaction" correctly belongs in the flap increment.

To produce a full project-level maximum lift prediction method,

clearly requires a way of moving from the prediction of increments in two-
dimensional, to the prediction of increments on a swept wing of finite
Aspect Ratio. This was accomplished by assuming that

(a) the two-dimensional increments varied as cos" (sweep) - where

In' was allowed to have different values for
1-12

(i) leading-edge devices

(ii) the main wing
(iii) trailing-edge devices

and "sweep" was taken as the l/rl-chord sweep of the element when
deployed ie taking note of the dependence of the pressure distri-
bution on "local sweep" identified in section 3.1;
<

(b) the ratio between the "two-dimensional value modified by sweep"

defined above and the "overall max.lift" was purely a planform effect
ie initial stall occurred at the highest local CL point across the
span

The values of 'n' which resulted from this analysis showed a fair
degree of scatter, but nevertheless ~onsistently~~e 23 showed the type of
wide variation across the chord indicated by the .values given below in
Table 4 . It is difficult to devise an acceptable explanation for this
behaviour, but the pressure distributions genuinely indicate that near
maximum lift with increasing sweep, that the proportion of the total load
carried by the slat increases, and that carried by the flap decreases.

DEVICE In'

(a) Leading-edge devices 1.10 +/- 0.50

(b) Main Wing 2.10 with flap deployed
1.20 with flap retracted
(c) Trailing-edge devices 3.60 for flap angles up to 20 deg

Finally, the effect of part-span flaps was derived empirically by

analysis of the changes in zero-lift angle and maximum lift found on
Model 477 as the part-span extent was varied.

The final expression for maximum lift on a finite swept wing was
written

CLrnax = l[[A] ( 1 + (%)

CL/CL
C
SLAT + (%)
C FLAP (s = l ))
s n, ]'

total area of wing

+ A(CLmax)FLAP cosnf@
+ part-span flap
dx Sn=l

total area of wing + (notional) full-span flap

chord extension of the slat
chord extension of the flap
highest value of local CL across the span
f(t/c) = empirical function of thickness:chord ratio.
 1-13

The results from this expression are compared with the achieved maximum
lift values on Model 477 over a full range of sweeps and various leading-
edge devices in Fig 3 3 .

4 RESULTS FROM THE STRUCTURAL/MECHANICAL DESIGN PROGRAMME

The structural/mechanical designs were done in great detail and

produced a wealth of detailed weight and cost data for each device. For a
Variable Geometry fighter, the weight of each leading-edge device24 is
compared with that of the 12.5% chord leading-edge slat (Ll) in Fig 34, and
corresponding comparisons are made for trailing-edge devices25 based on the
3 3 % chord single-slotted flap (T2), in Fig 3 5 . Similar corn par is on^^^-^^
for a civil aircraft are given in Figs 36 and 3 7 . All the weights of the
high-lift devices lie between +20% of the weight of the datum devices -
with the exception of the Flexible Kruger which is almost 80% heavier than
L1. The costs vary much more, and complicated triple-slotted flaps may
cost nearly 2.5 times the cost of a simple single-slotted flap.

Despite this wealth of data, it is difficult to use it to draw

guidelines for the selection and design of high-lift systems for complete
aircraft because the weights have to be interpreted in relation to the
aerodynamic capability, and both have to be related to exchange rates for
the design of the whole aircraft. The next section describes one possible
approach to this problem for civil aircraft, and reviews the results from
several applications of the method.

5 ANALYSIS OF RESEARCH RESULTS IN THE CONTEXT OF A COMPLETE

AIRCRAFT

5.1 Aerodynamic Model

For a civil aircraft, the take-off is usually the most demanding

design case for the selection of the high-lift system - especially for a
twin-engined aircraft - because the high-lift system has to supply not only
high-lift (to allow a short take-off run) but also low drag so as to allow
the aircraft to comply with the statutory minimum climb gradient after
take-off .

Perry30 has shown that the take-off wing loading can be expressed in
a form which, after some manipulation, can be written as:

where W = Aircraft Take-off weight

T = Take-off Thrust
S = Aircraft wing area
W/OSd = Parametric Take-off wing loading
T/OSd = Parametric Take-off thrust loading
d = take-off distance
O = density ratio
Kd = Empirical relation incorporating the effects of variation of
thrust during the take-off run, and the airborne transition
distance, etc.
1-14

The c l i m b g r a d i e n t (y) a f t e r t a k e - o f f may be w r i t t e n

where (y) = Climb G r a d i e n t

(y) r e q = minimum s t a t u t o r y climb g r a d i e n t a f t e r t a k e - o f f = f ( n )
n = number of e n g i n e s
cL/cd = A i r c r a f t 1 i f t : d r a g r a t i o d u r i n g climb-out a t V2 .

I n p h y s i c a l t e r m s , t h e l i f t - d e p e n d e n t d r a g , c,/cd n o r m a l l y d e c r e a s e s
w i t h i n c r e a s i n g CL i n t h e r a n g e of i n t e r e s t f o r t a k e - o f f , and hence f o r any
g i v e n v a l u e of T/W t h e r e i s an upper l i m i t t o t h e CL a t which a i r c r a f t can
m a i n t a i n a n a d e q u a t e climb g r a d i e n t a f t e r t a k e - o f f .

However, i f w e c o n f i n e o u r s e l v e s t o c o n d i t i o n s w h e r e t h e climb
g r a d i e n t r e q u i r e m e n t i s j u s t m e t , t h e n ( y ) becomes a c o n s t a n t i n
e q u a t i o n ( 2 ) and e q u a t i o n s (1) and ( 2 ) may b e t r e a t e d s i m u l t a n e o u s l y t o
yield solutions

These two e q u a t i o n s d e f i n e t h e r e q u i r e d aerodynamic c o n d i t i o n s f o r a

g i v e n a i r c r a f t s u c h t h a t t h e l i m i t a t i o n s s e t by t h e a v a i l a b l e t a k e - o f f
d i s t a n c e and by t h e minimum c l i m b g r a d i e n t a r e b o t h s a t i s f i e d .

Alternatively, equation ( 3 ) may b e r e - w r i t t e n i n t e r m s of T/aSd

from e q u a t i o n (41, t o y i e l d ,

For s a f e t y reasons, t h e take-off speed i s r e q u i r e d t o be a t l e a s t 20%

g r e a t e r t h a n t h e s t a l l s p e e d - . t h u s t h e t a k e - o f f C, i s 0.7CL,,,. Thus from
wind t u n n e l d a t a t h e v a l u e s of

c a n b e i n s e r t e d i n t o e q u a t i o n s ( 3 ) and ( 4 ) [ o r (S)] t o y i e l d v a l u e s of

"W/aSd" a s a f u n c t i o n of "TIaSd' .
I f t h e wind t u n n e l d a t a i s c o n v e r t e d t o t r i m m e d c o n d i t i o n s , t h i s , i n
p r i n c i p l e , e n a b l e s t h e aerodynamic d a t a t o b e i n c o r p o r a t e d w i t h t h e w e i g h t
d a t a from t h e S t r u c t u r a l / M e c h a n i c a l Design programme, t o y i e l d c o n c l u s i o n s
on t h e r e l a t i v e m e r i t s of d i f f e r e n t h i g h - l i f t d e v i c e s i n a c o m p l e t e
aircraft context.
 1-15

Two methods have been used to convert the simple lift and drag values
from wind tunnel tests to trimmed conditions. The first of these simply
calculates a lift reduction corresponding to the down load on the tail to
trim the aircraft with the CG assumed to lie at the 1/4-chord point of the
Mean Aerodynamic chord. Thus, if we make the simple assumption that the
effective tail arm is 3.5 mean aerodynamic chords in length, then

This method gives trim corrections that are, in general, too large; this is
because of the neglect of the downwash at the tail arising from the
trailing vorticity of the wing and flap system. However this method was
used for the investigations described in sections 5.2 and 5.3; for the
exercise reported in section 5.4 a more rigorous method was used which
properly took into account the downwash at the tail.

5.2 Aerodynamic Optintisation of Part-span Flap Extent

Five sets of aerodynamic data were obtained from Model 477, as set
out in Table 4 below. To examine the optimum part-span extent, the
analysis method above was used to compare sets 1 and 2 (to demonstrate the
effect of change of sweep) and sets 2 and 3 (to demonstrate the effect of
changing the leading-edge device). Because the Parametric Wing Loading
(W/OSd) varies very rapidly with the Parametric Thrust Loading (T/OSd), the
data resulting from feeding wind tunnel values of CLtrimand CLtrim/CDtrim
into equations ( 3 ) and (4), has been ratioed with corresponding data
calculated for a hypothetical aircraft having an analytical variation of
CJCD with CL:

Hypothetical aircraft CLtrim/CDtrim= 14 - 3CLtrim .

The resulting "Wing Loading Ratio" (W/aSd)4 7 7 / (W/oSD)datum is designated by
the va'riable ' x l .

Table 5
Sets of Part-span Data from Model 477

Set Sweep and Leading- Flap Gaps Shroud

aspect ratio edge position

20 deg/8.1 PLAIN 3 at each angle 90%

30 deg/7.0 PLAIN 3 at each angle 90%
30 deg/7.0 SLATTED 3 at each angle 90%
30 deg/7.0 PLAIN 3 at 20, 14, 8 deg 84%
30 degl7.0 PLAIN 3 at 20, 14, 8 deg 96%

Flap Spans = 56%, 75%, 100%

Flap angles = 20, 17, 14, 11, 8 , 5 degrees
1-16

C a r p e t p l o t s of t h e v a r i a t i o n of CLtrim and CLtrim/CDtrim f o r data

set 2 are g i v e n i n F i g s 38 and 39. When t h e s e v a l u e s a r e i n s e r t e d i n t o
e q u a t i o n s (3) and ( 4 ) , t h e t y p i c a l v a r i a t i o n of 'xf w i t h v a r i a t i o n of t h e
P a r a m e t r i c T h r u s t l o a d i n g ( T / O S d ) shown i n F i g 4 0 i s o b t a i n e d . From
F i g 4 0 , it c a n be s e e n t h a t t h e h i g h e s t v a l u e of 'x' a t a c o n s t a n t v a l u e of
T/OSd , w i l l o c c u r f o r a f l a p span t h a t i s less t h a n 1 0 0 % . By a p r o c e s s of
f u r t h e r i n t e r p o l a t i o n , t h e r e s u l t s shown i n F i g 4 1 a r e o b t a i n e d which show
a n optimum f l a p s p a n of a b o u t 80%, b u t performance i s w i t h i n 0 . 1 % f o r a l l
f l a p s p a n s between 70% and 9 0 % .

The f i g u r e s f o r d a t a sets 1 and 3, c o r r e s p o n d i n g t o F i g 4 1 , a r e shown

i n F i g s 42 and 43. Comparing t h e s e t h r e e f i g u r e s , i t c a n be s e e n t h a t :

(i) With a P l a i n Leading-edge a t 20 deg sweep, a n d Aspect R a t i o 8,

t h e optimum f l a p s p a n i s 1 0 0 % f o r a l l t h r u s t l e v e l s a b o v e t h a t
n e c e s s a r y t o a c h i e v e "matched" t a k e - o f f c o n d i t i o n s a t 7-8 deg f l a p .
For a l l t h r u s t l e v e l s below t h i s , i t i s b e t t e r t o r e d u c e t h e s p a n of
t h e f l a p t h a n t o r e d u c e t h e a n g l e of t h e f l a p s t i l l f u r t h e r . Data
from t h e E n d p l a t e Model a t low f l a p a n g l e s , shows t h a t t h e p r o f i l e
d r a g i s a minimum a t around 8 deg of f l a p - c o n s e q u e n t l y t h i s r e s u l t
might be e x p e c t e d .

(ii) I n c r e a s i n g t h e sweep of t h i s h i g h - l i f t c o n f i g u r a t i o n t o 30 deg

( a n d c o n s e q u e n t l y r e d u c i n g t h e Aspect R a t i o t o 7 ) p r o d u c e s a n optimum
s p a n of f l a p which i s a b o u t 80%, and t h e t a k e - o f f CL c a n be matched
t o t h e t h r u s t i n an optimum way by changing f l a p a n g l e between 20 deg
a n d 5 deg. I t s e e m s probable t h a t t h i s change i n behaviour i s a
r e s u l t o f t h e d i f f e r i n g r e l a t i v e r a t e s of change of drag a s e i t h e r
t h e f l a p s p a n o r f l a p a n g l e i s reduced. A t 2 0 deg sweep, w i t h i t s
h i g h e r Aspect R a t i o , t h e i n c r e a s e i n i n d u c e d d r a g a s t h e f l a p s p a n i s
reduced i s l e s s t h a n t h e i n c r e a s e i n p r o f i l e d r a g a s t h e f l a p a n g l e
i s r e d u c e d below 7-8 deg. A t 30 deg sweep, w i t h i t s l o w e r Aspect
R a t i o , t h e o p p o s i t e happens - t h e i n c r e a s e i n i n d u c e d d r a g a s t h e
f l a p s p a n i s r e d u c e d i s now greater t h a n t h e i n c r e a s e i n t h e p r o f i l e
drag a s s o c i a t e d with reducing f l a p angle.

(iii)P u t t i n g a l e a d i n g - e d g e s l a t on t o t h e wing a t 30 deg sweep,

i n c r e a s e s somewhat t h e v a r i a t i o n of optimum f l a p a n g l e w i t h f l a p
s p a n , a n d r e d u c e s t h e s p a n by a b o u t 6 % . T h i s b e h a v i o u r i s t o b e
expected because, i n t h e s l a t t e d c a s e , t h e p r o p o r t i o n of t h e t o t a l
s p a n l o a d i n g a r i s i n g from i n c i d e n c e ( a n d h e n c e s p r e a d a c r o s s t h e
whole s p a n i r r e s p e c t i v e of t h e f l a p s p a n ) i s much g r e a t e r . It i s
worth p o i n t i n g o u t t h a t t h e maximum l i f t f o r t h e s l a t t e d c a s e i s o v e r
ACL = 1 h i g h e r t h a n f o r t h e p l a i n l e a d i n g - e d g e .

5.3 Aerodynamic/Structural Investigation of Leading-edge

Devices

The above method was u s e d on wind t u n n e l d a t a f r o m Model 4 7 7 t o

compare t h e p e r f o r m a n c e o f s e v e r a l l e a d i n g - e d g e d e v i c e s 3 3, u s i n g w e i g h t
d a t a o b t a i n e d from t h e r e l e v a n t s t r u c t u r a l d e s i g n s t u d i e s . The most
i n t e r e s t i n g of t h e r e s u l t s o b t a i n e d i s t h e comparison between

(a) t h e 1 2 . 5 % c h o r d l e a d i n g - e d g e s l a t L1 and
(b) t h e 1 7 % c h o r d f l e x i b l e Kruger L7

The Kruger h a d a s l i g h t l y h i g h e r maximum l i f t a n d l o w e r drag, g i v i n g an

a d v a n t a g e o f a b o u t 1 . 5 % i n t a k e - o f f wing l o a d i n g ; however t h e g r e a t e r
 1-17

weight of the Kruger reduces this to about 0.75% in terms of greater

fuel/payload. Further optimisation of the slat setting across the span,
succeeded in reducing the drag to the point where there was no difference
in the payload performance of the two devices34.

5.4 Aerodynamic/Structural Investigation of Optimum

Shroud Length

From the previous parts of the National High-Lift Programme (and from
standard aerodynamic theory) it was known that moving the position of the
flap shroud rearwards

(i) increased the maximum lift and the sectional value of C,/C,;
(ii) increased the weight of the high-lift system, mainly because of
the increase in length of the cantilever tracks that carry the flap
at full extension:
(iii) increased the tail load to trim because of the further aft
centre of lift;
(iv) increased the induced drag of part-span configurations because
of the larger discontinuity in chord distribution.

With one advantage and three disadvantages, it was unclear where the
optimum shroud length would lie; in addition the evidence from different
aircraft manufacturers was conflicting.

The aerodynamic data for this exercise came from sets 2, 4 , and 5 -
but, in this case, in addition to the analysis performed above, it was
necessary to:

(a) deduce a landing flap design to go with each combination of

flap angle and shroud length -
so that the weight of the trailing-
edge high-lift system could be estimated.
(b) conduct a tail-sizing exercise - so that a proper estimate could
be made of the trim drag taking into account the downwash at the tail
from the wing trailing vorticity

To do the first of these, landing and take-off speeds and weights for
a large range of transport aircraft were reviewed and revealed that the
ratio

CL (on approach)
CL (at take-off)
= 1.12 .

It was further assumed that

C, (at take-off) = 0.7CLm,, (for take-off)

C, (on approach) = 0.6CLm,, (for landing)

so that

CL (for landing)
CL (for take-off)
= 1.31 .

From Endplate Model data, the following values of the ratio of maximum
lifts with different flaps were obtained.
1-18

Table 6
Values of t h e Ratio of Maximum L i f t with Various Landing
Flaps t o t h a t of a Flap a t Take-off S e t t i n g

SINGLE DOUBLE TRIPLE

SLOTTED SLOTTED SLOTTED

CLmax at landing flap (40 deg)

1.19 1.30 1.46
CLmax at landing flap
-
Thus from the results at 20 deg flap in each of sets 2, 4, and 5, the I
I
max lift for each type of landing flap could be estimated, and compared
with the demanded max lift from equation (6) in order to select the
landing flap required. With the landing flap decided, the resulting
Pitching Moment could be derived from Endplate Model data.

To work out the size of the tail, it was assumed that it was
necessary to provide a usable CG range of 35% Standard Mean Chord, and the
tail volume coefficient was plotted against CG position for the two
conditions

(a) to trim the landing flap

(b) to provide an adequate stick-fixed static margin for the cruise
configuration

leading to a typical graph shown in Fig 44. The resulting tail size,
expressed as a ratio with the wing area, for a tail arm of 3.5 chords, is
shown in Fig 45. Having decided the tail size, the Trim Drag can be
evaluated, and this is shown in Fig 46; it will be noted that, as explained
above, these trim drag increments are negative due to the downwash at the
tail. The aerodynamic take-off performance can now be calculated, as
before, by inserting

into equations ( 3 ) and (41, and evaluating (W/aSd) and (T/aSd).

The weight of the high-lift system can then be estimated using the
method due to Brookes8. Unfortunately for the generality of this research
exercise, Brookes' method produces weights/unit area that depend on
aircraft size. For relevance at the time the work was done, an aircraft
approximating to a BAC 1-11, with a range of 3500 km, and a Maximum Take-
off Weight (TOW) of 45350 kg, was chosen to continue the exercise. A
typical weight breakdown of the high-lift system for 90% shroud position
and 75% span flaps, is shown in Fig 47.
 1-19

To complete the exercise, the Maximum Take-off Weight (TOW) of the

aircraft was written:

TOW = Wstructure + 'fuel + 'payload + 'high-lift system *

K ~ c h e m a n nshows
~~ that for this range

-
Wstructure - Wengines + Wu/c+services + Wwing+controls + Wfuselage = 0 * 525.TOW

and therefore

W f u e l -t 'payload = 0.475ToW - W h i g h - l i f t systems .

The resulting values of "fuel + payload" as the take-off flap angle
and shroud length are varied, are given for 75% and 100% span flaps in
Figs 48 and 49. For each shroud length, the "Fuel + Payload" weight rises
as the thrust is increased and the take-off flap angle is increased to
maintain matched take-off conditions; the variation is smooth until the
single-slotted flap is no longer capable of supplying an adequate landing.
CLmax - at this point the landing flap has to become double-slotted.
Because this is heavier, the "Fuel + Payload" drops discontinuously - and
the same happens at higher thrusts and take-off flap angles where the land-
ing flap becomes triple-slotted. The interesting part is that the graphs
show a clear advantage to increasing the shroud length in comparison to
retaining the same shroud length and increasing the complexity of the flap.
However, the differences are quite small (of the order of 3% of "Fuel +
Payload") and this presumably explains why different manufacturers have
selected different designs of flap and all have been convinced that they
have selected the optimum design - because other factors (such as cruise
drag) could have effects which are larger than the differences shown here.

6 HIGH LIFT WORK IN THE UK SINCE NATIONAL HIGH LIFT PROGRAMME

Since the end of the National High Lift Programme, high lift research
within RAE (now DRA) has centred around the exploration of Scale and Mach
number effects on maximum lift using the capability of the DRA 5 metre wind
tunnel. This has resulted in a number of papers describing the nature, and
occurrence, of various identified types of Scale Effect13g19f36n
37.

7 CONCLUSIONS

1 A major research programme mounted in the UK in the 19703, and known

as the National High-Lift Programme has been described. In particular,
(a) the philosophy
(b) the experimental techniques
(c) the results obtained
(d) the implications for aircraft design,
have all been addressed. The programme was mounted in response to a
perceived shortfall in capability in comparison with other countries - the
USA in particular -
and has played a major part in restoring the balance.
It was one of four major research programmes mounted by the RAE in
collaboration with BAe which contributed to the design of the A 3 2 0
aircraft, and for which RAE and BAe were jointly awarded a Queen's Award
for Technology in 1988.
1-20

Table 1
Designations for Trailing-edge Devices
-
TO Plain trailing-edge
T1*t 33% chord single-slotted flap, 8 4 % ~shroud TE
T2*t 33% chord single-slotted flap, 9 0 % ~shroud TE
T2S*t As T2, but slot sealed and flap faired into wing
T3 33% chord single-slotted flap, 9 6 % ~shroud TE
T4 33% chord double-slotted flap, 8 4 % ~shroud TE
T5*t 33% chord double-slotted flap, 9 0 % ~shroud TE
T6t 3 3 % chord double-slotted flap, 9 6 % ~shroud TE
T7*t 3 3 % chord double-slotted flap, 9 0 % ~shroud TE, 8 . 5 % ~extending vane
T8*t 33% chord triple-slotted flap, 9 0 % ~shroud TE, 8 . 5 % ~extending
vane, 1 1 . 5 % ~extending tab
T8T As T8, but vane retracted into fairing
T9* 33% chord double-slotted flap, 9 0 % ~ shroud TE, 11.5%~extending tab
T10* As T5, but deployed on an external hinge
T11* As T7, but deployed on an external hinge

* Denotes devices on which structural and mechanical studies have been made
in the context of civil transport applications. These configurations
differed in some respects from their representations tested
aerodynamically.
Denotes devices on which structural and mechanical studies have been made
in the context of military applications.

Table 2
Designations €or Leading-edge Devices

LO*t Plain leading-edge

Ll*t 12.5% chord slat
L2*t 18% chord slat
L3t Plain droop, from 1 5 % ~station
L4*t Extending droop, from 12.5%~station, circular arc extension
L5* Extending droop, from 1 8 % ~station, circular arc extension
L6*t 18% chord RAEVAM, flexible upper skin 5 % to~ 18%~
L7 Kruger devices, having an upper-surface profile giving an optimum
pressure distribution when functioning as a flap with sealed flap/
wing gap
Flexible flaps:
L7A Operating in slat mode, optimised configuration
L7B* Operating in flap mode, optimised configuration
L7C Unsealed gap, non-optimum configuration, similar to
Boeing 747
L8* Flexible Kruger slat, optimised for angle, lap and gap.
Upper-surface profile designed to give optimum pressure
distribution when functioning as a slat with optimised slot

Denotes devices on which structural and mechanical studies have been made
in the context of civil transport applications. These configurations
differed in some respects from their representations tested
aerodynamically.
Denotes devices on which structural and mechanical studies have been made
in the context of military applications.
 1-21

REFERENCES

1 Barnes, C.H. Handley-Page Aircraft since 1907,

pp 210-216, Putnam (1988)

2 Smith, A.M.O. Aerodynamics of High-Lift Airfoil Systems.

AGARD CP 102, April 1972

3 Kuethe, A.M. Foundations of Aerodynamics.

Schetzer, J.D. John Wiley & Sons Inc (1959)

4 Thwaites, B. Incompressible Aerodynamics.

Oxford Clarendon Press (1960)

5 Houghton, E.L. Further Aerodynamics for Engineering Students.

Boswell, R.P. Edward Arnold, London (1969)

6 Duncan, W. J. The Principles of the Control and Stability of

Aircraft.
Cambridge University Press (1959)

7 Foster, D.N. The Nature, Development and Effect of the

Ashill, P.R. Viscous flow around an aerofoil with High-Lift
Williams, B.R. devices.
RAE Technical Report 72227 (1972)

8 Brookes, W. National High-Lift Programme Phase 5 -

Parametric Weight Study.
Report No.BAe-MFP-R-NHL-0079 (1981)

9 Woodward, D. S. Further work on the integration of closed loops

specified only as discrete data points.
RAE Technical Report 73177 (1973)

10 Moir, I.R.M. The Measurement and Analysis of the Profile Drag

Foster, D.N. of a Wing with a Slotted Flap.
Holt, D.R. RAE Technical Report 71158 (1971)

11 Foster, D.N. The two-dimensional flow around a Slotted Flap.

Irwin, H.P.A.H. ARC R&M 3681 (1971)
Williams, B.R.

12 Cairns, I.C.D. Analysis of Slat (Ll), 20 deg single-slotted

Wedderspoon, J.R. flap (T2), and 45 deg double-slotted flap (T7),
tested on a two-dimensional model.
BAC (Weybridge) Aero/FM/Report 041 (1973)

13 Fiddes, S.P. Investigations into the effects of Scale and

Kirby, D.A. Compressibility on lift and drag in the RAE
Woodward, D.S. 5m Pressurised Low-Speed Wind Tunnel.
Peckham, D. H. Journal of the Royal Aero SOC, pp 93-108,
March 1985

14 Cairns, I.C.D. Analysis of slat (Ll) and 20 deg flap (T2)

Wedderspoon, J.R. optimisation on a quasi-two-dimensional Endplate
Model.
BAC (Weybridge) Aero/FM/Report 039 (1971)
1-22

15 King, D.A. Development in Computational methods for High-

Williams, B.R. Lift Aerodynamics.
Journal of the Royal Aero Soc, pp 265-288,
August 1988

16 Finch, B. S. P . Investigation into the performance of a 40 deg

Wedderspoon, J.R. single-slotted flap on a quasi-two-dimensional
Endplate Model.
BAC (Weybridge) Aero/FM/Report 046 (19 1

17 Finch, B.S.P. Evaluation of a double-slotted landing flap (T7)

on a quasi-two-dimensional Endplate Model.
BAC (Weybridge) Aero/FM/Report 052 (1978)

18 Finch, B.S. P . Evaluation of a triple-slotted landing flap (T8)

on a quasi-two-dimensional Endplate Model.
BAC (Weybridge) Aero/FM/Report 061 (1978)

19 Woodward, D.S. Some Types of Scale Effect on high-lift Wings.

Ashill, P.R. ICAS Paper 43, ICAS (1984)
Hardy, B.C.

20 Garner, P .L. Areas for future CFD development as illustrated

Meredith, P.T. by Transport Aircraft Applications.
Stoner, R.C. AIAA-91-1527-CPr Honolulu, Hawaii (1991)

21 Collingbourne, J. Multivariate Optimisation applied to the initial

design of transport aircraft.
RAE Technical Report 84044 (1984)

22 Harris, K.D. A Review of the Investigations into the effect

of sweep on the Sectional High-Lift
characteristics using the Panel Model Technique.
BAe (Hatfield) Report HRS-R-RES-FM2362 (1978)

23 Woodward, D.S. Steps towards an improved project level

prediction of maximum lift.
Unpublished RAE NHLP Review Paper 5/1978

24 Lee, N. National High-Lift Programme Phase 1 -

Comparative Report on Wing Leading-edge Devices.
BAC SOR(P) 54, S&T Memo 8-72, DRIC-BR-32245
(1972)

25 Lee, N. National High-Lift Programme Phase 1 -

Comparative Report on Wing Trailing-edge
Devices.
BAC SOR(P) 55, S&T Memo 4-73, DRIC-BR-32940
(1972)

26 Scott, D. National High-Lift Programme - Comparative

Taylor, J. Study of Leading-edge Devices.
HSA (Woodford) Proj/NHLP/M42 (1975)

27 Scott, D. Comparative Study of single-slot Flap with 84%,

Taylor, J. 90%, 96% Chord Shroud (Tl, T2 and T3).
HSA (Woodford) Proj/NHLP/M37 Part I1 (1974)
 1-23

28 Scott, D. National High Lift Programme - Design Study of

Taylor, J. Externally-Hinged Double-slotted Flaps with 78%
Chord Shroud (T10 and T11).
HSA (Woodford) Proj/NHLP/M70 (1975)

29 Scott, D. National High Lift Programme Phase 5 - Weight

Teagle, P.J. Update of single-slotted flap with 90% Chord
Shroud (T2), and triple-slotted flap with 90%
Chord Shroud (T8).
BAe-MFP-R-NHL-0 102 ( 1979)

30 Perry, D. Exchange rates between some design variables for

an aircraft just satisfying take-off distance
and climb requirements.
RAE Technical Report 69167 (1969)

31 Woodward, D.S. An Investigation of the optimum take-off

Lean, D.E. configuration of a single-slotted flap on a
swept wing.
RAE Technical Report 82074 (1982)

32 Appleyard, G.M . An investigation of the effects of part-span

extent on the optimum settings for take-off
of a 33%-chord single-slotted flap with a 90%
shroud.
BAe (Brough) Technical Note YAD 3313 (1978)

33 Woodward, D.S. National High Lift Programme - A Preliminary

Assessment of the high-lift performance of three
flexible leading-edge devices.
RAE Technical Memorandum Aero 1617 (1975)

34 Woodward, D. S. A preliminary investigation of the performance

Cherrington, K.D gains obtainable by variation of the slat
setting across the span of a swept wing.
RAE Technical Memorandum Aero 1699 (1976)

35 Kuchemann, D. Performance aspects of various types of Aircraft

Weber, J . Progress in Aeronautical Sciences, Vo1.9,
pp 329-456, Fig 3.5 (1968)

36 Hardy, B.C. Experimental Investigation of Attachment-line

Transition in Low-Speed High-Lift Wind Tunnel
testing.
Paper 2 AGARD CP438 (1988)

37 Moir, I.R.M. An experimental investigation of the Optimum

Slat setting on a Strike Fighter Model.
RAE Technical Report 84023 (1984)
1-24

Viscous l a y e r p r o f i l e 0.20

Flow over two-dimensional

wing with slat and slotted flap

See also Tables 1 and 2

: 12.5% c slat
U :18% c stat fixed vane

90% shroud

U:15% plain droop

fixed vane

17:90% shroud,
Q: 12.5% c extending droop extending vane
e:I8% c exlending drwp

5%

s:lev. c RAEVAM
E:90% shroud,
extending vane and tab
\

Fig 2 Basic NHLP high-lift devices (schematic)

%
17:18% llexible Kruger llap go./. shroud,
U:18% llexible Kruger slat aux l l a p l i e tab)

Tunnel, A i Tunnel

NHLP standard
floor
(bled1 >
\

\ I
5 roof (bled1
/
/
transport s \ I
1 Table 71 \
\
\ II ;.5:
\
\
I
I /a.-
1
/
0
\ I /
\
\ /
\ A /

r ;Endplate
I
I
CE I
.- w I
82 I
I
I
I Fig 3&4 BAe (Weybridge) 2-D and
 a a 1-25

Boundary layei
diverter gap

Tunnel png g
* m N

Fig 5 Constent chord swept panel model (BAe)

$?

Sweep pivot on quarter-chord line

Fig Sa Tapered swept-wing model (RAE M477)

(0.01

II I'
r __----
flap nose

ZF (gap)
Wing chordline
(retracted flap)

a ) Single-slotted lrailing-edge flap

II Slal trailing-edge t zslgap)

Fig 6b Model 477 in the DRA Bedford 13fl x MI]

wind tunnel

Wing c h o r i i k
(retracted slat)

bl Simple slatted Leading-edge Islat)

Fig 7 Positional coordinate systems - basic devices

1-26
0
LE reference
/points 6" flap

Lower stub
shroud

External hinge - f ixes

nonoptirnurn laps and gaps

a1 Standard configuration ( T 7 ) -
lap and gap conventions
bl
configuration (Tdfe
External-hinge t

Fig 8 Double-slotted trailing-edge flaps (l7 & TlO) - nomencalture

a1 Flap elements and their

b l Lap and gap conventions-

leading flap ( T 8 )

c ) Lap and gap conventions -

take-off flap ( T O T )

Fig 9 Trlpleslottedtrallingsdge flap (T8 & T8T) - nomenclature

 1-27

- 18.0
CP
-16.0
-x- 2-D model
-14.0 -0- E-Pmodel

-12.0

-10.0

-8.0

-6.0

-4.0

-2 .o

0
1 .o

Fig 10 Comparison of wall-to-wall and end plate model pressure distributions

COPE = profile drag estimated from
IZ;ZI~ ng
Ce
balance measurements
extended chord

'OW'
'"c EP model - L7B/T2

0
r I I
1
I I
AX/Ce 2
I I I I I

- . - XDwot
a) Streamwise variation of measured CDW
0.05 I U I
ce E / P model
CDPE ax

b) C, dependence of streamwisegradient of CDw

E / P model
0.6

1.28 0
1.29 0 Fig 11 Typical wake momentum defect
1.30 0
1.32 V coefficient (CDW) measurements

0.2 I I I 1 I I I I
0 2 CL L
c ) CLdependence of %,,,measured at Ax/ci= 1.3
1-28

0.04 0.08(x/d

Fig 12 Comparison of wall-to-wall and normalised panel model pressure

distributions

-14

SLAT PRESSORB DISTRIBUTIONS

-12

8
B
Lo
z -10

!i
!

9
P
-*

3 -6

E
d b Datum -

4-
-4
P

t Datum -
-2

8 Ik
0 0

Fig 13a Comparison of swept panel model results at various sweeps for slat L1
at 25" and flap T2 at 20"
 1-29

-12

8
B -10
m

g -e
F
0

PP
s
-I

1
rn
w -I

E!
P
0

::
3 -
P

Fig 13b Comparison of swept panel model results at various sweeps for slat L1
at 25" and flap T2 at 20"

2
-3.0

-2.5
I
W

2 -2.0
0

P -1.5
:
v)

8
ci

8 -0.5
et

0.0
0

Fig 13c Comparison of swept panel model results at various sweeps for slat L1
at 25" and flap T2 at 20"
1-30

Fig 14a Pressure distributions on two-dimensional and model 477

Fig 14b Pressure distributions on two-dimensional and model 477

 1-31

-18.0
(Cp)N
-16.0

-1h.O

-12.0

-10.0

-8.0

- 6.0

-4.0

-2 .o

0
1 .o
76 0.84 0.92 (X/C)N

Fig 15a Comparison of normalised sheared wing pressure distributions based

on local sweep angles

-
-0- 30"sweep
20"sweep

0.1 02 0.3 0.4 0.5 0.6 0.7 0.8(x1

Fig 15b Comparison of normalised sheared wing pressure distributions based

on local sweep angles
I
1-32

-8

-1

-6

-5

8 -4

-3

-2

-1

0
24 -0.10 -0.16 -0.12 .-0.08 -0.04 0.00 0.04 1
( XS -5.28 I

Fig 16a (slat TE) v (Xs 5*Zs) - Fig 16b Slat peak v (Xs 5*Zs) -

( Xa - 5.18 1
1.24 -0.20 -0.16 -0.12 -0.08 -0.04 0.00 0.04 0.08
( X8 - 5.28 1

-
Fig 17 (slat peak slat TE) v (Xs 5*Zs) - -
Fig 18 (wing peak wing TE) v (Xs 5*Zs) -
 0 0
1-33

0.05 0.05

I x. I 0 . m I. - a.o* I , Y - 0.m. I. - z.o* ,

0.04 0.04

0.03 0.03

0.02 0.02

2 s
-
U)
N E
0
8 0
0
0.01 0.01

0.00 0.00

-0.01 -0.01

-0.02 -0.02
05 -0.04 -0.03 -0.02 -0.01 0.00 0.01 -0.01 -0.04 -0.03 -0.02 -0.01 0.00 0.01
O+./C) (XSIC)

Fig 19a Prediction of optimum slat location Fig 19b Prediction of optimum slat location
Slat angle = 25" Slat angle = 20"

Fig 20 Optimisation of 12'/2% chord slat

L1 for maximum lift
W
1-34

Fig 21 Boundary layer shapes on single slotted flap (T2) at various chordwise
positions on a 2-D model

20.00-
Slat = L1 at 25O
Flap = T7 at 45O
16.67 -
a =12O C,= 4.4
U

s
$13.33 -
m
zalo.oo -
L

2i
n
m
E 6.67 -
.-m
al
I
3.33 -

0-

Fig 22a Boundary layer shapes on double slotted flap (l7)

at various chordwise
positions on a 2-D model
 1-35

2o.oor

16.67 -
U

s
813.33 -
m
c

2$10.00 -
0
n
m

.-Q
6.67 -
I

7
8 5
Sn I Overlap
-n
m --
Em 0 -
tI ATLl
2 :
.-
c

0"
Underlap
t , I , I l
-5 - A
.'

5
s - LO
3 - h
v
L7B Ll
,-.n
v

0
3 ) -
-
=3 0 -
- Fig 24 Effect of leading-edge device on the flap slot setting
EIP model

2 :
.-a
4-4
T2 flap at 20"
I I I I I I
required for optimum C L, performance
1-36

cH a
Grid region below
A Conventional lift curve
B Flap separations improving
at high incidence
C Excessive separations,
lift curve collapse

Fig 25a NHLP 90 in x 30 in E/P model with

single slotted flap (T2) Flap angle
6, = 40" C L,, contours derived
from carpet plots

Flap flow development indicated by- -boundaries

TE g
0.009in t r a n s i t i o n t r i p
( a l l configs)
3.5 0.3 F
TE f l a p const cn
C
6, L 20' l0,2.5) c, .-
v-0
3.0 0.2 1p
n u
ZLn
- c
2.5 0.1 E 5
2
c
e
In
2.0 0 6

Fig 25b NHLP 90 in x 30 in E/P model

with single slotted flap (T2)
Flap angle dF = 40" Variation of
flap flow separation with incidence

Fig 26 Configurationsof LE slats L1, L2, L8 and K13,

tested on the end plate model
 1-37

All results obtained in conjunction with single slotted flap (T2) at kp20" (0,2.3)and L 2 2 0 FPS. except where noted

,&;\a\
\o x---
_ _x- - - - --- - - x - - - -L8- $ (sealed v:-in -conjunction
_ _ _ sl_w_i t h T8,
ZOOfps

(Nominal slat d e f l e c t i o n a n g l e 6,or 6K)'

Fig 27 Summary of measured CL max performance of 6 LE slats/Krugers on the

90 in x 30 in end-plate model
x
j 0.02,

Sealed '

. I

1
Sealed Krugers
2
Slat chordlwing chord *CL rnax
a) bl

Fig 28 Summary of slat performance comparisons on the end-plate model

+'
dZ
J

Fig 29 Lift-incidence curves for a double slotted flap

at 3 angles I I I I I I
c 5 10 15 20 25 L. 30 35
 e
5.0 -
CL

L.8

L.6
-

-
K‘7
L.1 -

L.2 -
kv.45 degree flap

[ P L O degree flap

4.0 -
1-1 L 3 5 degree flap

I
25 30
- 35

Fig 30 Lift-incidence curves for a triple Fig 31 Method of determinationof CL,,= increments
slotted flap at 3 angles

C, = j C T D from the endplate model is a close approximation to the CLon a 2D - wing

Fig 32 a Demonstration that lift curve slope is closely proportional to

deployed chord
 0 0 1-39

C,= ?J;:C +C from the endplate model is a close approximation to the CLon a 2D - wing

c, JvD
=

Based on
deployed
area

cIo

Fig 32b Derivation of flap and slat increments from typical end-plate model data

L .O

3.0

1.0

1.0 2.0 3 .O L .o

Fig 33 Correlation of C L
,,
 - 34a Weights of a range of leading-edge
Fig
devkes on a variable-geometry fighter
-
Fia 34b Costs of a range of leading-edge
devices on a variable-geometryfighter

Fig 35a Weights of a range of trailing-edge Fig 35b Costs of a range of trailing-edge devices
devices on a variable-geometry fighter on a variable-geometryfighter
 1-41

Fig 36a Weights of a range of leading-edge Fig 36b Costs of a range of leading-edge
devices on a civil transport aircraft devices on a civil transport aircraft

_."." .. ., . .._._ .._..._....I_.___.__ ... . . .

Fig 37a Weights of a range of trailing-edge Fig 37b Costs of a range of trailing-edge
devices on a civil transport aircraft devices on a civil transport aircraft
1-42

.-E
A-
V

1.1

i n
60'1. 70'1. 80% 90.1. 1 00'1. 11VI. 120%
56'1. s p a n f l a p Span t 10 7759(Zf - 1 96'1.1 +0.8333(20*-ef)

Fig 38 Variation of trimmed lift coefficient, at 0.7 C max, with flap gap,
deflection and span; plain leading edge; sweep = 30"
12

11

.
U"
,E
2
0

10

9
60'1. 70'1. 80'1. 90'1. 100'1. 110'1. 120'1.
-
Span + 10 7759(Zf 1.96%1-0.8333~20*- 8 , )

Fig 39 Variation of trimmed lift-to-drag ratio, at 0.7 C max, with flap gap,
deflection and span; plain leading-edge; sweep 30"

1.03

1.01

n
Y

.-
e
0

m
L

m
._
3 0.99
-
0

m
.-
3

0.97

.
0.95
0.22 0.26 0.26 0.28 0.30 0.32 0.34
Parametric thrust loading (Nlm?

Fig 40 Derivation of optimum take-off performance at optimum flap gap;

plain leading-edge; sweep = 30"
 1-43

Within Within Optimum

0.5'1. optimum 0.1'1. optimum
20' -

.-
e
15' -

--
U

0
U

:
a
10' -

Fig 41 Variation of optimum flap deflection for

- ---- take-off with flap span; plain leading edge;
5
' sweep = 30"

60'1. 70'10 80'1. 90'1. 100'1. Within Within

0.5'1. optimum 0.1% optimum
20' -

s
.- 15. -

-
L

-U
U

D
W

n
10' -
Variation of optimum flap deflection
1

for take-off with flap-span; plain

leading-edge; sweep 20" -
'5

Within .Within Optimum I , 1 I

6 0'1. 7Vl. 80'1.
20' - 0.5V.optimum O,l*l. optimum
Flap span
90'1. 100'1.

.f" 15' -
-
W
U

U
V

n
;lo' -

'5 - Fig 43 Variation of optimum flap deflection for

take-off with flap span; sweep = 30"
60'1. 70'1. 80'1. 90'1. 100'1.
Flap span

CG position x q / ~

Fig 44 Graphical determination of tail volume coefficient

1-44

0.30-

0.29.

0.28 -
* 0.27 -
e
:0.26.
.-c
2 0.25 -
e 0.24 -
f
-n
I 0.23 -

--
N

-
' E 0.22 ~

m
0.21 -
0.20 ~

0.19 -

Fig 45 Tail size as a function of flap angle, span and shroud length

Fig 46 Trim drag coefficient as a function of a flap angle,

span and shroud length
1300C

1200c

11ooc

10ooc

soot

8OOC

:o 1001

8
600C I I I
+
500c

Sudaces
400C -

300C

ZOO(

100C

I
60 65 10 11 I
maus1 r kn

Fig 47 Typical weight breakdown of high lift

system
 1-45

195

190

18s

180

I
c
115

0E no

E!
+ 165
8
P
8
2
lSO

155

150

145

140
0

Fig 48 Variation of take-off performance with thrust for 75% flap span

Fig 49 Variation of take-off performance with thrust for 100% flap span
 2-1

EXPERIMENTAL INVESTIGATION OF FLOW

AROUND A MULTIELEMENT AIRFOIL

Nafu Alemdaroglu
Aeronautical Engineering Department
Middle East Technical University
Inonu, Bulvari
Ankara 06531
nrkey

1 Summary not exist and are still in development. The complex

nature of the multielement geometry and the flow field
makes the problem more difficult if the real physics of
This paper presents the results of experimental in-
vestigations performed around a multielement air- the shear flow is to be modelled. The effort under-
foil and gives detailed information about the flow taken by Cebeci et al. [6],tries to compute the flow-
in and around the flap-well and slat cavity regions. field around multielement airfoils and uses the present
data and those of Nakayama [2], for code validation.
Measurements are made using pitot probes, hot-
wire anemometer and laser Doppler velocimeter a t a Among the available data on multielement airfoils few
Reynolds number of 0.5 and 0 . 8 ~ 1 0 ~The. results were involved with the flap well and slat cavity flows.
obtained show the complex nature of the shear flows In particular, references [7] and [8] were concerned
investigated and put into evidence the necessity of ac- with wing-flap configurations. The work of Adair
curate modelling of these flows by numerical methods. and Horne [9], deals with a wing-flap configuration
and contains turbulence measurements in addition to
mean flow quantities. The work reported in references
[lo] and [ll]involves three element airfoil with a lead-
2 Introduction ing edge slat, but is limited only to pressure and mean
velocity measurements inside the boundary layers.
The present set of experiments is an extension of the
experiments carried out by Nakayama et al. [l],[2] The recent work of Savory et al. [12], is of particu-
at the Langley Research Center low turbulence pres- lar importance since it deals with slat and flap cavity
surized wind tunnel (LTPT). The model used in the measurements and in particular gives emphasis to tur-
present experiments is a scaled down version of the bulence as well as to mean quantity measurements. In
model used in the LTPT tests. Although the mea- these respects it is very much similar to the present
surements of Nakayama were performed at realistic experiments.
Reynolds and Mach numbers and were complete in
terms of the overall external flowfield description, they
lacked detailed information about the flowfield inside
the flap and slat cavity regions. Hence, a series of 3 Experimental Set-Up
experimental investigations were undertaken at Cali-
fornia State University, Long Beach to complete the 3.1 Wind Tunnel
definition of the total flowfield around this multiele-
ment airfoil [3], [4].
The experiments are conducted a t the low speed wind
The measurements inside the flap and slat cavity re- tunnel of the Aerospace Engineering Department of
gions are of crucial importance for the computational California State University at Long Beach. The tun-
aerodynamicist since the calculation of the separated nel has a test section of 22 in. x 28 in. and a stream-
flow in these regions are still not accurate enough and wise length of 44 in. with a free stream turbulence
needs to be validated against reliable experimental level of less than 0.5%. The tunnel has no facilities
data. While powerful1 computational tools are avail- for side wall boundary layer control nor any provisions
able to calculate flows around single element airfoils for high-lift tests. However it can be considered to be
with separation and at high angles of attack [5],ac- sufficient for measurements at relatively low angles of
curate and reliable methods that compute the flow attack where flow separation is small and is limited to
around high-lift systems of multielement airfoils do the flap-well and 'slat cavity regions.
2-2

3.2 Model 3.3.3 Hot wire Anemometer

The multielement airfoil model used in the present Two channels of TSI 1050 Constant Temperature
work is the two dimensional LB-572 high lift model Anemometers (CTA) are used with standard TSI-
as shown in figure 1. It is a three element airfoil with 1243 x-wire boundary layer type probes to measure
a leading edge slat and a single-segmented flap. The the mean velocity and the turbulence quantities. The
size of the model is such that the chord of the config- microcomputer based data acquisition and reduction
uration with the elements stowed is 12 in. system is fully verified and well documented in report
The slat and the flap are attached to the main wing ~31.
by means of a set of brackets at two spanwise loca-
tions. The brackets are designed in such a way as to 3.3.4 Laser Doppler Anemometer
allow optical access to the flap-well and slat cavity
regions by laser velocimeter. The brackets are placed The laser velocimeter system used in the present ex-
sufficiently away from the centerline so that the two- periments is described fully in a previous report [14].
dimensionality of the flow is not altered a t the mea- It consists of a 3W Argon ion laser source and a
surement location. These brackets can be moved both two color, dual beam back scatter optics using the
along the chord in order to adjust the overhang, OH, 488 nm and 1514.5 n m lines and a two component
and normal to the chord to adjust the gap, G. counter processor. In later experiments, the counters
The model is equipped with a total of 125 sur- are replaced by a Burst Spectrum Analyser (BSA),
face pressure tabs on all three elements. For two- which provided significant improvements in LDA sig-
dimensionality checks, there are two spanwise arrays nal analysis. The optical components are all mounted
of pressure tabs, one along the spoiler trailing edge on a large optical bench which in turn is mounted on
(main element) and the other along the trailing edge a milling machine bed and is traversed along three
of the flap. Seven pressure orifices are placed along axis. The data acquisition and reduction were ac-
the flap well ceiling surface and three orifices on the complished by means of a PDP-11/44 minicomputer
rearward facing step. which w a s replaced with a personal computer when
BSA w a s introduced.
The model is installed inside the tunnel horizontally
between circular end plates on the side walls. These
end plates can be rotated to change the angle of at-
tack. An optical quality glass is inserted in part of the 4 Flap and Slat Cavity Flows
end plate so that the optical paths of the laser beams
and the scattered light are left clean. The geometry and the definitions of the coordinates
used in a typical flap-well flow are indicated in fig-
ure 2. In general a flap-well flow has the characteris-
3.3 Instrumentation tic features of both internal and external flows and is
composed of partly irrotational, partly turbulent and
3.3.1 Pressure Measurements reversed flows. The two important streamlines shown
in figure 2 are 1) The dividing streamline which leaves
All pressure data including the tunnel static and to- the main airfoil lower surface at the step of the flap-
tal pressures are measured by electronic manometers well and divides the flow circulation in the separation
connected to two Scanivalve pneumatic scanners. The bubble from the flow that passes through the gap be-
analog output is digitized with a computer controlled tween the main airfoil trailing edge and the flap upper
12 bit A/D converter. Sufficient care w a s given to surface. 2) The flap stagnation streamline that ter-
check and validate the response time of the pressure minates at the forward stagnation point of the flap.
measurement system. If this stagnation streamline is known, then the flow
above this streamline can be thought of as an internal
channel flow with sudden expansion on one side and
3.3.2 Flow Visualization
a curved wall on the other. However, if one consid-
ers the pressure distribution, it is governed basically
Flow is visualized by means of smoke wire tech-
by the external flow and is mainly influenced by the
nique. The technique uses a thin (0.1 m m diameter)
entire geometry.
nichrome wire stretched across the flow and connected
to a voltage regulated AC power supply with a timing As far as the variables are concerned, the reattach-
and delay control circuit. Smoke is generated by coat- ment point, X R ,of the dividing streamline is the most
ing toy train smoke oil on the wire and then burning important aerodynamic parameter since it determines
it by resistive heating. a number of important quantities. This point strongly
 2-3

depends to the shape of the separation bubble and to of 0.8 x lo6 and 0.5 x lo6 with corresponding Mach
the pressure distribution on the ceiling surface of the numbers of 0.13 and 0.08 are used during these ex-
flap-well, and in particular to the location, X,, , where periments. Reynolds number is based on the clean
the pressure peaks. In general we use geometric pa- wing chord length of 12 in. One of the major in-
rameters such as the gap, G I and the overhang, OH, conveniences of the small test section is definitely the
both expressed as percentages of the mean chord, the blockage effect. However, it does not invalidate the
flap deflection angle, 6 ~ and, the overall angle of at- trends of the flow characteristics with respect to the
tack, CY, as the parameters influencing these flow quan- parameters changed.
tities. Detailed measurements of the flap-well flow The test conditions are summarized in tables (1) and
were made during these experiments and the effects (2) for flap-well and slat cavity flow measurements re-
of the flap location on the quantities indicated in fig- spectively. Due to space limitations, for hot wire mea-
ure 2 were determined at various flow conditions. The surements only 3, in some cases up to 4, streamwise
mechanism of the separation and the reattachment of
stations can be traversed, whereas for LDV measure-
the shear layer w a s investigated. The coordinates 2 ments more stations could be probed giving a better
and y designate the distances along the chord (which resolution of the separation bubble and the recircu-
is parallel to the flap-well ceiling surface) and nor-
lating regions.
mal to the chord (ceiling) and downwards respectively.
One can now define the displacement thickness, 6', The two dimensionality of the flow is demonstrated
using its conventional definition. in figures 4a and 4b which shows the spanwise vari-
ation of the pressure along the flap trailing edge at
6' = l6 (1 - g) dy two different flap deflection angles. It is observed on
these figures that the flow uniformity is maintained
where U, is the external velocity measured a t the edge over a spanwise extent of about a chord length near
of the boundary layer, y = 6, at separation. The the centerline for angles of attack up to 12'. At higher
displacement thickness, y = 6' , will be a smooth curve angles of attack, this distribution becomes worse near
extending from the step edge down through the gap. the tunnel side walls due to the thickened and proba-
Due to a large normal to surface pressure gradient, bly separated side wall boundary layers.
the shear layer edge, 6, and the edge velocity, U,, can To give qualitative pictures of the flowfield, figures 5,6
not be defined in the usual way and hence the edge is and 7 show the results of smoke wire flow visualiza-
taken as the position where the turbulence intensity tions. Figure 5a to d show the flow visualization in
is twice the free stream value. and around the slat cavity region at various slat set-
.As far as the slat cavity flow is concerned, figure 3 tings and Reynolds numbers. Figure 7.a is the flow
shows the geometry and the probe traverse positions. visualization a t an angle of attack of 4 O and at a flap
Similar to flap-well flow some very interesting features deflection angle. of 10'. It is observed that the bound-
of the flowfield such as the large curvature of the di- ary layers are thin and attached up to the flap trailing
viding streamline, the stagnation streamline and the edge. Figure 7.b shows the same picture with a flap
position of the stagnation point on the lower surface deflection angle of 25'. Figure 7.c shows the result
of the main airfoil, the extend of the circulation re- when the angle of attack is increased to 10' with a
gion in the slat.cavity and the total mass flow rate flap deflection angle of 25'.
through the gap between the slat and the main ele-
ment are investigated. In addition, the effects of slat
angle, 6,, slat overhang, S O H ,and slat gap, SG,on 6 Results
the above mentioned flow features are studied.
6.1 Surface Pressure Distribution
5 Test Conditions Since it is impossible to show here all the surface pres-
sure data, only those representative cases will be pre-
The experiments are conducted to obtain detailed in- sented. Data presented in figures 8.a to 8.c are all
formation of the flow in the flap-well and slat cavity taken at Re = 0.5 x lo6, and can be reproduced at a
regions with emphasis on the effects of parameters higher Reynolds number of Re = 0.8 x lo6. Figures
such as; slat gap, slat overhang, slat deflection an- 8.a and 8.b show the effects of flap overhang and gap
gle, flap gap, flap overhang, flap deflection angles as ratios on the pressure distribution at a fixed flap angle
~
well as the overall angle of attack on the separated of 10'. The effect of angle of attack on the pressure
and reattaching flows in the slat and flap cavity re- distribution for a fixed geometry setting of 2 % gap
I gions. Since flight Mach number and Reynolds num- and 4 % overhang with a flap deflection angle of 10'
ber simulation w a s not possible, Reynolds numbers is shown in figure 8.c. It is observed in all of the cases
I
2-4

that the pressure distribution near the step edge and where the first integral reduces to yd with the use of
in the forward portion of the well is fairly constant, equation (3). The quantities Y d , 6 and 6' are very
which is followed by a small pressure peak and a sharp important since they can be used in the process of
negative pressure gradient to join the much lower up- fairing (smoothing) the original geometry of the bod-
per surface pressure at the trailing edge of the main ies and to include the effects of displacement thickness
element. Similar figures can be presented for a flap in interactive viscous/inviscid calculation procedures.
deflection angle of 25'. The mean velocity vectors in and around the slat cav-
ity region are presented in figures 1l.a through 1l.h.
6.2 Hotwire Data These plots show clearly the orientation of the veloc-
ity vectors and indicate clearly the strong curvature of
The mean velocity vectors obtained by the x-wire the streamlines. By using the same procedure as for
probe are shown in figures 9.a through 9.f for a flap flap-well flows, one can easily determine the locations
deflection angle of 10' and in figures 10.a through 10.f of the dividing and the stagnation streamlines as well
for a flap deflection angle of 25'. Also shown on the as the stagnation point on the main airfoil.
same figures are the mean velocity vector distribution The results of LDV runs for flap-well flow measure-
in the gap region between the main element trailing ments are presented in figures 12.a, b and c for the
edge and the flap upper surface. These measurements three different test conditions specified in table 1. Fig-
are realized by means of a pitot tube and is used to ures 12.a and b show the effect of different overhang
determine the total flow rate through this gap. If this ratios with other parameters kept constant. Figures
flow rate is known, it can be used in the determina- 12.b and c compare the effect of two different flap an-
tion of the flap stagnation streamline. The dividing gle settings. In all of these cases it is observed that
streamline, the flap stagnation streamline, the edge of the measurements are in good agreement with the hot
the separated shear layer and the displacement thick- wire data with the exception of the back flow velocity
ness distributions are also shown in these figures. The which can not be determined with hot wire.
symbols and definitions used are the same as in figure
2. In order to obtain the dividing streamline smooth
curves are drawn that are tangent to the step lower 6.2.1 Turbulence Data
surface edge. and to the local velocity vectors. The -
position of the flap stagnation streamline, ys, is ob- - -
The turbulent shear stress --utut, and the total nor-
tained by using the continuity equation and equating mal stress U ' ~+ u t 2 , where U' and U' are the fluctuat-
it to the total flow rate Q calculated through the gap ing velocity components along and normal to the flap
between the main airfoil and the flap. well ceiling surface respectively, are shown plotted in
figures 13 and 14. These quantities give ideas about

1: Udy = Q (2)
the level of turbulence intensities and the degree of
their correlation. Both of these turbulent stresses are
significant in the momentum equations for separated
Since the directional ambiguity of the flow can not and reattaching flows. The positions of the dividing
be resolved by hot wire measurements, the near wall streamline, yd, and the shear layer edge, 6, as obtained
measurements where the flow is known to reverse are from the mean velocity data are also shown in these
inaccurate. However, an estimate of this backflow ve- plots. It is observed that the position of the sepa-
locity can be given by using the continuity equation rated shear layer, as indicated by the region of non-
and the position of the dividing streamline; yd
-
zero shear stress, --u"vt, is not perfectly aligned with

I"" Udy = 0 (3)

the location of the dividing streamline. This inaccor-
dance which may seem contradictory to our intuitive
guesses may be explained in terms of the influence
where U is the velocity component parallel to the sur- of the stability condition of the curving flow and the
face. These crude estimates of the back flow velocities position appears to shift lateraly much more than is
are also shown in these figures. Calculation of 6' is implied by the mean streamline.
performed using equation (1) which can also be writ-
ten as;
7 Discussion
The discussion of the results for flap-well and slat cav-
ity measurements will be concentrated on the effects
of two major parameters, namely the overhang ratio,
OH, and the gap ratio, G, on the separated flow.
 2-5

7.1 Effects of Overhang overhangs. This behavior can also be observed in the
C, distributions on the slat and on the main element
A close observation of the pressure distribution curve, while the C, distribution on the flap remains unef-
figure 8.a indicates that with increasing overhang rib fected by the slat overhangs. The suction peak on
tio from 4% to 5% there is an abrupt change in the the main element which has a value of -3.8 a t 4% slat
pressure distribution on the upper surface of the main overhang reaches a value of -5.5 when the slat over-
element. A possible explanation of this sudden change hang is reduced to 3%. With further reduction in slat
can be attributed to a shift of the transition location overhang to 2%, the suction peak on the main element
on the upper surface of the main element. attains a value of -6.0. The behavior of the slat pres-
sure distribution is also as expected. With decreasing
More detailed flow analysis is presented in figure 15 slat overhang the difference in pressure between the
for flap deflection angles of 6~ = l o o . This figure lower (inner) and upper (outer) surfaces of the slat
shows, the position of the dividing streamline, the gets smaller and in fact with 1% slat overhang the
pressure distribution along the flap-well ceiling sur- lower surface C, becomes equal to that of the upper
face, the position of the flap-stagnation streamline cal- surface, figure 17. As a consequence, the lift of the
culated from gap flow rates, the static pressure along slat shows a continuous increase with increasing slat
the edge of the separated shear layer calculated from overhang.
external velocity and the trailing edge pressures of the
main element airfoil, for different overhang ratios a t
fixed flap deflection angle and gap ratios. Figure 15 7.2 Gap Effects
indicates that X,, and the dividing streamline move
in the same direction, resulting in shorter separation Figure 18 summarizes the effects of gap on the sep-
bubbles for larger overhangs and vice versa. The LDV arated flap-well flow. The most noticeable effect is
data of figures 12.a, b and c for three test cases also the opposite changes in the directions in which the
confirm that the position X,, of the peak pressure on dividing streamline and the flap move. As the flap
the flap-well ceiling surface is very closely related to is lowered, (increasing gap), the dividing streamline
the reattachment point X R . moves up. This is due to the fact that as the gap is
Figure 16 indicates more clearly the influence of over- increased more fluid from below the main airfoil moves
hang on the X,, and CPP,the peak pressure coeffi- into the flap-well and passes through the gap, thereby
cient, for three different gaps. X,, is inversely pro- the dividing streamline is pushed farther away from
portional and varies linearly with OH, indicating that the flap leading edge to allow for this increased mass
X,, moves almost the same distance as the displace- flow rate. It is observed on these figures that the di-
ment of the flap leading edge. The peak pressure viding streamline and the flap stagnation streamlines
coefficient, C,, , however, remains uneffected and is remain nearly parallel with changing G. If the flap
practically constant with changing OH. closes in, (approaching the main airfoil), the divid-
ing streamline will be lowered more and will even be
It is also important that the main airfoil’s trailing lower than the flap and instead of going through the
edge pressure decreases considerably as the overhang gap will hit the flap surface. Moving the flap further
is increased but in such a way that the streamwise apart from the main element, causes the gap to widen,
pressure gradient between the peak pressure point and moves the dividing streamline upwards and shortens
the gap exit is maintained nearly constant. the separation bubble even more. These conditions
The surface pressure within the separated region is are verified with further experiments where the flap
fairly constant whereas the pressure at the edge of is moved farther away from the main airfoil with new
bracket’s. If the flap is removed altogether, however,
the shear layer, y = 6, is seen to be consistently higher
the separation bubble becomes longer than any other
than the surface value. The difference between these
case with the flap.
pressures is related to the streamline curvature of the
separated flow. The difference between the surface The flap stagnation streamline moves with the flap.
and the shear layer pressures is larger for larger over- With decreasing gap, (flap moving up), the shear layer
hangs for which the separation bubble is shorter and edge will move against the flap which will reduce the
hence the streamline curves up more sharply. potential core going through the gap.
The effect of slat overhang is observed when cases CI, From the foregoing discussions of the isolated effects
Cz and A are compared in figure 11. It is found that of gap and overhang, their combined effect for a t r a n s
the slat overhang effect is more important than the lational motion of the flap such that the gap is in-
slat gap effect. One can conclude that more fluid is creased and the overhang is reduced at the same time,
injected through the slat gap for smaller overhangs may lead to a separation bubble and a reattachment
due to smaller recirculation bubbles formed a t small length X A practically unaltered.
2-6

For the smallest gap, the dividing streamline is the large turbulence intensities. Nevertheless it exhibits
lowest and the flap stagnation point is the highest. similar characteristics to backward facing step flow
This leads to a very little potential flow through the or to sudden expansion channel flow. In all of these
gap. For 6~ = 25O case, there is no potential core left flows the reattachment length X R seems to be the
and the totaly turbulent shear flow impinges on the major parameter. For an internal flow the basic pa-
flap. rameter that determines the reattachment length is
The effect of slat gap is observed when cases D1,0 2 the expansion ratio. In the case of a flap-well flow the
equivalence of the expansion ratio is simply the ratio
and A are compared in figure 11. With increasing
gap distance more fluid is pushed into the slat cavity of the widths of the flow that goes through the gap
and through the slat/main element gap. This results at the trailing edge and at the upstream edge of the
in a reduced size circulation bubble in the slat cavity flap-well. The incompressible continuity equation im-
which in turn causes the the main element stagnation plies that this ratio of widths is inversely proportional
point to move closer to its leading edge. With increas- to the ratio of the velocities. For cases in which there
ing slat gap, the pressure on the inner surface of the is a potential core, the velocity ratio is related to the
slat decreases further due to increased and accelerated pressure ratio. Hence the ratio of the pressures up-
flow passing through the gap, figure 19. stream the flap-well and at the trailing edge seems to
be the key parameter. This pressure ratio can further
be related to the pressure drop from the lower surface
7.3 Flap Deflection Angle Effect to the trailing edge of the main airfoil.
Therefore, the characteristics of the flow is tried to
Although only two flap deflection angles, l o o and 25O
be explained in terms of the gap exit velocity. It is
are tested and there is not sufficient amount of data
observed already that the most important parame-
available to give quantitative information about the
ter that effects the gap velocity is the gap ratio, G,
effect of flap deflection angle on the flow, the results
and the overhang, OH, has practically very little or
presented here give some information about the gen-
no significant effect. With increasing gap increasingly
eral trend of the effects. With increasing flap deflec-
higher velocity flow is added near the flap upper sur-
tion angle, the angle of attack of the flap with respect
face. This increase in total flow rate is larger than
to the local flow is increased which will lead to lower
the proportional increase of the gap itself. This is an
suction pressures near the flap leading edge and hence
amplified change in the total flow rate due to a small
an increase in the flow rate through the gap. An in-
change in the gap width. Increasing the flap deflec-
crease in flow rate, on the other hand, means shorter
tion angle, 6~~ has a similar effect as increasing the
separation bubbles in the flap well.
gap width.
The effect of flap angle on the slat cavity flow is ob-
served when cases A and G are compared. Larger In order to analyse the relationship between the gap
flap angles cause the flow to change direction much flow rate Q and the separation bubble, the parameter
earlier. Increasing flap angle modifies the C, distri- AX,, = X F L E- Xpp,(where X F L E is the chordwise
butions on the slat and on the main element. The C, distance of the flap leading edge from the leading edge
values on the upper surface of the main element are of the clean airfoil), or the C,, is plotted against the
moved towards more negative values resulting in an gap flow rate. The plot of the nondimensional A X , , / c
increased lift coefficient for the main element. Same against Q normalized with respect to the edge velocity
observations are also true for the slat. at the flap-well step, A, is shown in figure 21. In
this way the correlatioi between the location of the
The effect of slat angle is observed when cases E and pressure peak distance relative to the flap leading edge
A of figure 11 with respective slat angles of 20° and and the gap flow rate is put forward.
30° are compared. The velocity vectors become more
horizontal with smaller slat angles. While the C, dis- Another important feature of the flap-well flow is con-
tribution for the main element does not change with cerned with the turbulent stress and its modelling.
increasing slat angle, that on the slat shows a drastic Gooray et al. [15] and Autret et al. [16] have tried
change. The suction peak on the slat is reduced with to model the turbulence structure in a backward fac-
increasing slat angle, see figure 20. ing step using the K - c turbulence model and found
that the model required some modifications due to
the large local streamline curvature effects. The mean
7.4 Characteristics of the Separated velocity w a s over and underpredicted before and af-
Shear Flow ter the reattachment point respectively, across which
the sense of streamline curvature reverses. This local
The separated and reattaching flap-well flow is ex- streamline curvatures are much more pronounced in
tremely complex with large pressure gradients and flap-well flows as compared to the unobstructed back-
 2-7

ward facing step flows. As a result the lengths of the Multielement Airfoil”, AIAA J. Vol28, Jan 1990,
separation bubbles are much smaller than the corre- pp 14-21.
sponding back step flows due to the blockage effects
of the flap. As far as the turbulence structure is con- Nakayama A., “Flow Field Survey Around High-
cerned, it is observed that upstream of reattachment Lift Airfoil Model LB546”, McDonnell Douglas
the sense of the streamline curvature tends to supress Report, MDCJ4827, Feb. 1987.
the turbulence and consequently the turbulent shear
stress and the intensities are much lower when the Alemdaroglu H.N., “Experimental Investiga-
separation bubble is shorter or when the curvature of tion of the Flow About Multielement Airfoil
the dividing streamline is larger. Model”, California State University, Long Beach,
Aerospace Engineering Department, Report No.
In order to show this more explicitly, in figure 22, tur- AS-23996-C, 1988.
bulent shear stress distributions are superimposed for
various flow conditions. This figure not only shows Alemdaroglu H.N., “Experimental Investigation
the trend of the shear stress levels, but also reveals of the Flow About the Slat of a Multi-Element
that the position of the separated shear layer shifts Airfoil”, California State University, Long Beach,
significantly. It is seen that as the dividing stream- Aerospace Engineering Deparment, Contract Re-
line’s curvature increases, the position of the shear port No. AS-24477-C, June, 1989.
layer moves up more as indicated by the region of
significant shear stress. A close examination of these Cebeci, T., Jau, J., Domenico, V. and Chang,
profiles will yield the fact that this shift of the shear K.C., “Prediction of Post-Stall Flows on Air-
layer location is due to the increased damping of the foils”, in “Numerical and Physical Aspects of
shear layer at the lower edge of the shear layer as the Aerodynamic Flows”, ed. Cebeci, T., Springer-
streamline curvature is increased. This implies that Verlag, 1990. (ISBN 0-387-52259-X) pp. 97-109.
an intuitively plausible assumption that the separated
shear layer follows the dividing streamline is a poor Jau, J., Vitiello, D. and Cebeci, T., “A Cooper-
one. ative Program With Douglas Aircraft Company
and Aeritalia on the Development of Multiele-
ment Aerodynamics Code, Final Report”, Re-
port MDC K4377, August 1989, Long Beach.
8 Conclusion
Seetharam, H.C. and Wentz, W.H.J., “A Low-
Detailed measurements of the flowfield in and around Speed Two-Dimensional Study of Flow Separa-
the slat cavity and the flap-well regions of a three com- tion on the GA(W)-1 Airfoil with 30-Percent
ponent multielement airfoil are performed using pres- Chord Fowler Flap”, NASA CR-2844, May 1977.
sure probes, hot-wire anemometer and laser Doppler
velocimeter. The effects of flap angle, flap overhang Van den Berg, B., “Boundary Layer Measure-
and gap, slat angle, slat overhang and gap as well as ments on a Two-Dimensional Wing with Flap”,
the angle of attack and the Reynolds number on the Rept. NLR TR79009U, National Aerospace Lab-
mean flow and on the turbulence quantities are inves- oratory, Amsterdam, Jan. 1979.
tigated. The data serves as test cases for computa-
tional methods. An immediate use of the experimen- Adair, D. and Horne, W. C., “Turbulent Sep-
tal results in computational work will be the fairing arated Flow in the Vicinity of a Single-Slotted
of the separation bubbles in the slat and flap cav- Airfoil Flap”, AIAA paper 88-0613, Jan 1988.
ity regions by smooth curves corresponding to the di-
[lo] Ljungstrom B., “Boundary Layer Studies on a
viding streamlines. Furthermore, detailed mean-flow
Two Dimensional High Lift Wing”, Report AU-
and turbulence measurements provide information for
862, FFA, The Aeronautical Research Institute
better understanding of the flap-well and slat cavity
of Sweeden, Stockholm 1972.
flows. The rate of fluid flow through the slat and flap
gaps is the key parameter in the determination of the [ll] Bertelrud, A. and Ljungstrom, B., “The Vis-
slat cavity and flap-well flows and is very closely re- cous Flow Around a Two Dimensional High Lift
lated to the lift coefficient of the main airfoil. Wing, Analysis of Boundary Layer Measure-
ments”, Technical Note AU-115, FFA, The Aero-
nautical Research Institute. of Sweeden, Stock-
References holm 1974.

[l] Nakayama A., Kreplin H.P., Morgan H.L., “Ex- [12] Savory E., Toy N., Tahouri B. and Dalley S.,
perimental Investigation of Flow Field About a “The Flow Regimes in the Cove Regions Between
2-8

a Slat and Wing and Wing and Flap of a Multi-

Element Aerofoil”, Proceedings of the Interna-
tional Symposium on Engineering and Turbu-
lence Modelling and Experiments” editors; Rodi,
W.and Ganic, E.N., Elsevier, 1990.
[13] Akdag, V., Nakayama, A., Liu, B., Kilik, E. and
Unt, H . , “Automated Hot wire Measurements
Using a Microcomputer”, California State Uni-
versity, Long Beach, Mechanical Engineering De-
partment, Report No. ME-84-5, 1984.
[14] Alemdaroglu, H . N., Nakayama, A. and Egan, D.
“ T w o Component Laser Doppler Velocimeter for
Measurements of Separated Turbulent Flows”,
California State University, Long Beach, Center Figure 3: Slat cavity flow
For Aerodynamics Research, Report No. CAR-
88-1.
[15] Gooray, A.M., Watkins, C.B. and Aung, W.,
“Improvements to the k-e Model for Calcula-
tions for Turbulent Recirculating Flow.” Proc.
4th Symposium on Turbulent Shear Flows, Univ. 0 a-4‘
0 8’
of Karlsruhe, Karlsruhe, FRG, 1983, pp. 18-26. b 12’
v 14’
[16] Autret, A., Grandotto, M. and Dekeyser, I., “Fi-
nite Element Computation of a Turbulent Flow
Over a T w o Dimensional Backward Facing Step”
Int. J . for Num. Meth. in Fluids, Vol. 7, 1987, pp.
89-102.

Figure 1: LB-572 3-Element high-lift model

1 I I
-0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8
( b ) gF * 25’ ZIC

Figure 4: Spanwise pressure distribution along the

flap trailing edge
HAIN-AIRFOIL

Figure 2: Flap-well flow geometry

 2-9

X
2

Figure 5 Smoke-wire flow visualization within the alat cavity.

2-10

Table 1: Tent conditions for flap well flow measurements

Table 2: Test conditions for slat cavity flow measurements

...

Figure 6: Smoke wire flow visualisation

within the flapwell
 2-11

-
F i e 7 a: Smoke wire 6
overall flow field, 6 p
nualiration,
P,U = 4'.

pigure 7 b Smoke wire flow visualisation,

overallflo~field, 6~ = 25', U = 4'.

Figurr7 c: Smoke wire flow visualiration,

mallflor &Id, 6, = 25O, (I = .'OI
2-12

OHIC
-
._._____
32
41
51
............. 61

-*D

CP
-3.0
1.1 * 10.. 6F . 10.. O/E = 2s. RI .
0.5 x lo6.

-zo

-11

1.0
0.0 0.t 0.2 0-3 0.4 U 0.a 0.7 OJ 0.. 10 I, U U .I
x/c
Figure 8.8: Surface pressure distribution,
effect of flap overhang

-I.,

.,.I

-*I

-*I

CP
-2.c

4.c

0.0

,.a

.-.
4 0 t

XIC

Figure 8.c: Surface pressure distribution,

effect of angle of attack
 e 2-13

( a ) OH/c * 1%. G/c - 1%

(b) OH/c - 3%. G/c = 2% (b) OH/c = 1%. G/c = 2%

/.-
( c ) OH/c = 1%. G/C 3%
( c ) OH/c * 3%. G/C = 3%

-
(d) OH/c - 4%. G/c * 1%
(d) Otl/c = 2%. G/c 1%

(f)
---
OH/c = 41, G/c = 3%
(f) OH/c - 2%. G/c * 3%
-z

I
Figure 9: Mean velocity vectors in flap-
I well obtained by hot-wire and pitot Figure 10: Mean velocity vectors in flap-
I measurements, 6 p = IO', a = 10'. well obtained by hot-wire and pitot
measurements, 6 p = 2 5 O , a = loo.
2-14
e
2-15 I
 2-16

( a ) OH/c = 3%. G/c = 1%

C

(b) OH/c - 3%. G/c = 2%

I

4
( b ) bF = 10 deg, G/c . 1%. OH/c * 5X.

(d) OH/c - 43. G/c = 1X

(e) OH/c = 4%. G/c - 2%

-
(C) 25 deg. G/c - l X , OH/c - 3%.

Figure 12: Mean velocity vectors obtained

by LDV measurements
(f) Oli/c - 42. G/c - 32 \
Figure 13: Turbulent shear stress and to-
tal normal stress distributions, 6 p =
100.
 2-17

OH/c Cp AT THE PAIN-AIRFOIL TRAILIIIG EDGE

- 3%
-0.67 -0.51
42

9 I 1
0.74 0.78 XIC 0.86

(a) OH/c = 1%. G/c = 1% PAIN AIRFOIL

0 0.1 I

(b) OH/c - 1%. G/C = 2%

Figure 15: Effects of overhang on the di-

viding streamline and pressure inside
the flap-well, 6 p = lo', G/c = 2%.

xPP cPP
0 0 1: lo.7o
0 2% I

(d) OHIc 2%. G/C 1%

Figure 16: The positions and the values of

I (e) OH/c = 2%. G l c = 2%
the peak pressure on the flap-well
ceiling surface as functions of O H ,
ap = 10'.

I
Figure 14: Turbulent shear stress and to-
tal normal stress distributions, 6 p =
I 25'.
 2-18

9
0
9
I
9
I
0
0
9
I
-
9
I
9 -
9

a
0
9 9 3 0
N -
9
x .9
Y

I d

9
, 9 3
I
9 9
I
-9, 9
-
9 9
I
9
8
9
I
9
,
9
, -
9 9

,
9
1
n ,

9
1
3 ,
9
I
9I
, ,

9
,
,

9
1
-
9
,& 9
-
9

I
9 9 9, 9 9 9
, 4
9
I , I
*

- 3
0
o tad
U
1 I
9 9 9 90 9 9 9 9
, *
 2-19 I

0.4 -

AX /C
PP
0.2 -

0 1 I I I I I
0 0.02 0.04 0.06

QG''stepc

Figure 21: AX,, and C,, as function of

flap gap flow rate Q.

( a ) Gtc - 1%.
OH = 4%&& 1%

( b ) G/c 2%.

MAIN AIRFOIL \

Figure 22: Effect of streamline curvature

on turbulent stress in the flapwell.
e e
 4- 1

A D E S I G N PROCEDURE FOR SLOTTED F L A P S

by
Sergio De Ponte* - Alessandro Cella - Mario Marcazzan
* Politecnico di Milano - Dipartimento di Ingegneria Aerospaziale
Via Golgi 4 0 1-20133 Milano Italy-

If it could be done, large lift increment

could be reached, as shown in the figure.
ABSTRACT
In the design of slotted flaps it is
attempted to avoid a reacceleration between
the trailing edge of an upstream element of "t
the system and the peak velocity of the Potrntial circulation
downstraem element, to reach the maximum
lift. increase
It is proved that it is possible by means
of a nemerical procedure based on a vortex
distribution.
The resulting shapes are then discussed
with reference to the application to real
design.
Fig. 1
Lift increment
INTRODUCTION The first approach to the problem was to
see the amount of lift increment for a
The design of slotted flaps might given flap chord that could be reached
either be done in an optimizarion way or it increasing the dumping velocity. In the
is in any case an attempt to have a good same way it was explored the possibility of
compromise between cruise and reducing the flap chord for a given lift
take-offllanding performances. coefficient. All this was done without any
consideration about the feasibility of the
The most naif way to try this compromise is airfoil.
to choose the airfoil shape for cruise
conditions and the wing loading for Some very rough study has shown that the
take-off, as done in many aircraft designs use of a flap as mean of increasing the
in a far past. dumping velocity could lead to very short
flap chord for a given lift increment. For
In any attempt to have a compromise between a double Liebeck pressure distribution some
two conflicting conditions, it is useful to possible result is shown in the following
know which are the opposite extremes: in figure, from ref. ( 3 )
this case they are the plain airfoil and
the unconstrained best lift flap. The
former is well known since a long time,
while the latter is the subject of the
present research. Aim of the present work
is to answer to some fundamental questions
which arise on the design of maximum lift c 1 Incre;,,
/+p
slotted airfoils. It is not therefore a
method of designing usful airfoils but a
method for knowing the other extreme in a
design process, to establish correctly the
penalties imposed by the compromise.
The overall scheme is similar to the
procedure of Liebeck and Omsbee (1),, with
the main considerations of A.M.O. Smith (2)
about flaps, and is conducted on a single
slotted flap as an example. this will not
prohibit to extend the procedure to more
element airfoils.
In this way maximum deceleration is 'sought
on the suGtion side of the airfoil and the
highest dumping velocity at the trailing
edge of the main airfoil is attempted.
Observing usual flaps, it could be noted
that, after the trailing edge of the main Fig. 2 preliminary lift evaluation
airfoil, the flow is more or less
reaccelerating on the flap. Then the But if the answer to this main question
following question naturally arises: could be affirmative, a second arises
immediately and it is about the shape of
Is it possible to reach at the such an airfoil system, mainly with respect
trailing edge of the main element the same to possibility of obtaining some cruise
velocity as the maximum on the flap, in condition by rejoining the elements in a
order to get the maximum dumpinq velocity? reasonable contour.
4-2

Of course, although the first question has But while the source strength leads to a
larger theoretical importance, the latter linear system of algebric equations, their
is quite practical in engineering sciences. locations lead to a non linear problem: it
In order to answer to both questions it was is therefore common practice to prescribe
thaught that the numerical choice coud be the source locations in some reasonable
the only feasible one, due to the way. It means that we have some degree of
complexity of the problem for a pure freedom in locating the singularities.
analytical approach.
Putting N sources on the chord line, each
THE NUMERICAL PROCEDURE of intensity qi, the closure is related to
the total outcoming flow, which should be
The numerical procedure is based on a panel zero. This leads to the relation:
method, being the conformal mapping Cq1 =o
unsuitable for multiple airfoil shapes. In l = l , N
fact, Timman method ( 4 ) allows to treat To satisfy this condition, only N-1
double aiorfoils, but it seems too
difficult to extend it to a larger number velocities are assigned, i.e. the
of contours. velocities between the points where sources
are placed, being the last (zero
On the basis of many experiences, it seems outflow) the Nth condition.
that a method which does not allow any The non crossing condition is similar, and
contour modification in the chordwise says that, at any station along the chord,
direction is rather difficult to use, the total flow produced upstream by the
therefore a method that has no restriction sources should not be negative, resulting
on the airfoil modification was chosen. in the N-2 conditions:
The basis is then an iterative vortex panel q1>0 1=2,N-1
technique, starting from a first l=l, I
approximation shape. being the first source always positive in
our procedure and the total intensity equal
THE FIRST APPROACH AIRFOIL. to zero as said.
A first approximation is obtained by It is possible to see that the conditions
linearized small perturbation theory, are redundant if one does allow to
superimposing mean line and thickness prescribe velocities in the points where a
distribution. naif idea could suggest.
The mean line is obtained by a simple
standard design, while for multi component Furthermore there is a set of inequalities
airfoils it is simply broken and shifted in to be satisfied for the non crossing
the normal direction by a prescribed slot conditions.
width. This shift is a still open problem. In our procedure the inequalities are
Thickness distribution is first obtained by simply verified at each iteration and the
a discrete point source distribution computation is stopped whenever the
prescribing tangential velocity along the inequalities are not satisfied. In
chord. practice, it has been noted that, beyond a
certain number of sources, the scheme does
At this point it is possible to iterate the not converge.
computation of the thickness, placing the
condition of given tangential velocity In order to spend the freedom of choosing
distribution along the airfoil contour the chordwise location of the
instead of the chord line; this might be a singularities, the first source is placed
good improvement in case of rather thick at the point along the chord, where it
airfoils. produces a Rankine ogee with the same
velocity derivative at the stagnation point
In this procedure particular care is taken as the desired airfoil velocity
to modify the velocity distribution in distribution,thus avoiding an extra
order to obtain a closed and non crossing stagnation point condition, but adding a
airfoil. constraint.
These are properties of the thickness
distribution and not of the camber line. In this way the places where sources could
The relevant conditions ought' to be be located is limited and it this avoids
investigated before going to the detailed problems near the nose .
In fact, any
design, in order to avoid convergence attempt to place a source upstream to this
problems during further calculations. first one, produces a sink and meaningless
The problem of closure is very well posed contours.
for conformal mapping and also the non
crossing of the contour requires only an THE INVERSE PANEL METHOD.
univocal mapping of the airfoil into the
cercle. The inverse panel mathod is the inversion
of a classical Martensen procedure ,in an
Also in the singularity method the question iteration process. The Maretesen method (
is rather well defined for the small 5) represents the airfoil by a vortex
perturbation problem. distributiion according to the Chaucy
In principle, the representation of a formula :
thickness distribution by means of a
discrete number N of point sources is a 2 N
unknown problem, i.e. the strength and
location of each source. which . i s transformed in a second kind
 4-3

Fredholm integral equation by putting the then computed as:

boundary condition of no internal
Ani-i+2Ani+Ani+i
tangential velocity.
Ani = -
- .-
The airfoil contour from the former 4
iteration ( the first approximation airfoil
for the starting step) is divided into GEOMETRICAL PROBLEMS
panels and at the midpoint of them the
velocity is prescribed.This velocity gives Having displaced the panels, a new set of
the vortex strength, which is kept constant segments represents the airfoil. Of course
on each panel and, if the shape would be it is not closed and has not exactly the
correct, the inside tangential velocity prescribed chordlength, which is unity in
should be zero, according to the Martensen dimensionless form. This is due to the fact
condition and is therefore computed. Being, that the panels near the leading edge might
in general, non zero, a modification move in the chordwise direction, as
procedure is then started. formerly stated.

The first attempt was to compute the It is therefore necessary to reshape again
derivative of the tangential velocity in the geometry to obtain a closed airfoil and
the direction normal to the panel and try then scale all the lengths by a constant
to use this as guide to correct the related to the increment in airfoil length.
airfoil. But it resulted almost impossible The reshaping is done reconnecting panels
to state a feasible modification criterion joining them at midpoint of the
on this basis. discontinuity each side point, as shown in
the figure:
Therefore a more complex procedure was
attempted. It consists in the computation
of the efffect of perturbing the airfoil
shape displacing each panel in the normal
direction by a certain small amount, (a
given fraction of its length ) and
computing the perturbed inside velocity.
Only the two neighbour panels at each side
are displaced in order to save aorfoil
closure, as shown in the following figure.

/-- _/---
Fig. 4 Panel reconnection
We have therefore a tvo step contour
modification: the first is the shifting and
the second a rotation and a change in
length and,of course. this last changes the
slope of each panel and thus the normal and
tangential velocity components. For this
reason it is rather difficult to asses
, theoretically the convergence of the
scheme.
But , more important, this closes anytime
the contour, without modifying the velocity
Fig 3 Panel displacement distribution. The question will be
The procedure is then repeated for each discussed later on.
panel. For rescaling the lengths the most upstream
At this stage, a complete matrix [A] of point is then found giving the new chord
influence of contour modification is length and all the geometry is normalized
obtained, in the sense that we have the with respect to this latter.
effect of displacing each panel on the The new center of panels are found and, by
overall velocity distribution. interpolation, at each center point a new
velocity is given, changing accordingly the
In this way it is possible to shift the vortex strength.This is a third step in the
contour in the normal direction according airfoil correction
to the influence matrix and the difference
between the computed and the target Incidentally, testing the program on
velocites by a simple relaxation scheme. cylinders approximated by regular polygons
Being j the known vortex strengthland An in the first stages of the research, it was
the shift in normal direction used in the observed that, without rescaling, all the
matrix calculation, the new velocity is v geometrical quantities are reduced at each
and the old v. iteratioh, tending to a zero chord shape.
All this procedure is programmed on a
The new velocity v is computed as: personal computer
v = [A! j + V m
The shift A required is therefore: FIRST TESTS.
The first tests were made on known shapes
for relaxing the scheme the real value is and on analytical velocity distributions,
first of all the circular cylinder and
ellypses of changing aspect ratio.
4-4

AIRFOIL DESIGN.
Having in mind that the contour was
enforced to be closed, only velocity Having acquired a certain confiance in the
distributions for closed contours were use of the method, the design of flapped
attempted at this stage. airfoil was attempted, with the aim of the
least reaccelereting flow beyond the
Test of giving the pressure distribution of velocity of the trailing edge of the main
one airfoil and the first approximation airfoil.
shape of another gave significant results
even in exteme case as shown in the The pressure distribution was prescribed on
following figure. the basis of "equilibrium" turbulent
boundary layers and similar solutions for
laminar ones.
The velocity distributions along the chord
From i n n e r t o o u t e r a i r f b i l are therefore:
j v = A X" formthe laminar part and
V = B (X-xl)
for the turbulent part and the exponent m
has the minimum value of 0.25 assumed as
limit for attached flows.
No closure condition was formally imposed
on the velocity distribution. This is
different from Liebeck approach and gives
-0.11 slightly different shapes but is thaught to
be more conservative and easier to use. The
velocity distribution is given in the
following figure:

y'c4
0.1
From o u t e r t o i n n e r a i r f o i l

0.0

4 . 1 j

Fig.5 Examples of iterative procedure

It has been observed that the obtained
contour is different from the target one,
as might be seen from the figures. 1 L.E. A; xi T.E
In general, the angle of attack is larger
and the mean line camber is smaller.
Suspecting that this could be significant,
direct computation of pressure distribution
€or both target and obtained airfoil were
performed. Fig. 6 Velocity distribution
To avoid to use computational schemes with
similar kind of approximations in direct
and inverse calculations, a Hess-Smith
direct method was used instead of using a
Martensen code.
The pressure distributions were
surprisingly similar for target and
obtained airfoils suggesting that the
design problem could have, at least from
the pratical point, a nearly non unique
solution, in the approximate sense that
small changes in the pressure distribution,
mainly near the stagnation point, may
produce large geometrical changes. Similar
effects were observed by Volpe in trensonic
design ( 6 )
Further tests were made on twice connected
contours of known flapped airfoils.
Fig. 7 Target - ..
and obtained., .
pressure distributions and airfoil shape
 4-5

Again, as in the first test stages, the of the previous figure. Further designs,
velocity distribution on the final shape reducing the spacing between components,
was computed by a different code to be gave first ideas about a possible slot.
compared to the target one. This was obtained also reducing the
The results are interesting although some acceleration on the pressure side upstream
problem is still open. the slot, as shown in the figure:
First of all it is possible to prevent But it is interesting to notice that the
reacceleration without any unfeasible slot should be a non converging channel in
conditionong of the shape, as shown in the contrast to the usual practice.
former figure. Furthermore, the flap curvature immediately
A comparison to an airfoil of Ormsbee and downstream the slot is rather small,
Chen ( 7 ) gives an idea of the strong incidentally explaining the success of some
reacceleration on other kind of designs. Fowler flap.
The first attempts gave almost independant
airfoil shapes, without any indication In this last case the pressure on the lower
about the possible slot shape. The results side was kept as high as possible, giving a
are similar to the one of Omsbee and Chen very high camber and a design lift
coefficient of 4.43.
Out of the design condition the velocity
-1
distribution is rather regular and possibly
at 13O angle of attack there is no nose
deceleration, suggesting an increase in
lift coefficent up to 4.75 as a limit for a
rooftop pressure distribution, but no
boundary layer calculation was attempted up
to now, because of some wiggle in the
pressure distribution.
CLOSURE PROBLEMS
AS said no formal closure condition was
posed in the design, but,is well known that
a small modification of the stagnation
point location could change very much the
airfoil thickness.
In fact, the computed velocities show a
typical peak near the leading edge and it
is suspected that it is due to the lack of
closure condition.
In an attempt to improve the method, it was
Fig. 8 Ormsbee and Chen airfoil decided to allow simply a shift in the
prescribed stagnation point location. When
v / d

2.5-

2.0-

1.5-

1 .o-

0 5-

:I
0.0
-0 1

Fig. 9 A better airfoil

4-6

the airfoil is modified, the leading edge good velocity distribution around the slot
changes chordwise location with respect to from the basic point of view of boundary
the prescribed stagnation point. This new layer and interference, before going on in
distance is then taken as new stagnation airfoil design.
point location in the rescaling of the
airfoil, giving a velocity that is References
prescribed in all points in the same way as
in the previous iteration, except near the 1) Liebeck - Ormsbee Optimization of
leading edge, where the stagnation point Airfoils for maximum lift Journal Of
location is revised according to the new Aircraft Vol 7 N 5- 1970
iteration of the airfoil shape.
This procedure allows a better agreement of 2 ) Smith A.M.O. -Aerodynamics of High-Lift
the computed velocity distribution to the Airfoil Systems AGARD CP 102- 1972
target one, everywhere except near the
stagnation point.
This is a third step in the modification.
3) Martellini -
Verso l’ottimizzazione di
ipersostentatori Thesis, Milano, 1988
OPEN PROBLEMS 4) Timman - The Direct and Inverse Problem
of Airfoil Theory. A Method to Obtain
The first open problem is related to the
wiggles in the contour near the leading and
Numerical Solutions NLR Report F 16 -1951
trailing edges. Probably something is 5) Martensen Berechnung der Druckverteilung
related too the small number of panel used, an Git terprofi 1en in ebener
but there is the suspect of the need of Potentialstroemung mit einer Fredholmschen
some smoothing of the contour at each Integralgleichung Arch. Rat. Mech and
iteration. Lacking of any convergence Analysis Vol 3 -1959
criterion, it is difficult to state that
there is no dendence of some local 6) Volpe -
Inverse Design of Airfoil
divergence even in an overall convergence Contours: Constraints, Numerical Methods
and critical points are always leading and and Applications. AGARD CP 463
trailing edges in panel methods.
-1989

7) Ormsbee -
Chen Multiple Element Airfoils
A second problem is related to the width of Optimlized for Maximum Lift Coefficients
the slot, which is now almost prescribed AIAA Journal Vol.10 n. 12 , 1972
because the method does not modify it
significantly. It was stated that the slot 8) Blom - Comment on Paper N 13 during the
width could be related to the airfoil drag discussion of AGARD Symposium V/STOL
( 8 1. Aerodynamics, Agard CP 143 pag A-9
Probably the approximation of the velocity
distribution used in these first examples
is too crude for a reasonable insight into
the slot problems and should be modified * This research vas partially supported by
accordingly. CNR ( Italian Research Council)

CONCLUSIONS
The first conclusion is that the aim of
designing slotted flaps with a dumping
velocity very close to the maximum velocity
on the flap is almost obtained, showing, in
principle, the feasibility of such a
design.
Second is the result that the shapes are
not so unusual for a complete design
procedure.
Then it might be observed that a flap could
be designed having in mind that it might be
mainly a system of increasing the dumping
velocity, instead of a system of producing
direct lift. In this sense a rough approach
suggests that a good design could reduce
the flap chord for the same lift increment,
saving simplicity and room for structure,
fuel and equipements.
All this should, of course, verified in the
overall deesign procedure.
A lot of problems are still open, both from
the theoretical and from the practical
point of view. Among them, the smoothing of
the shape and a better insight in the
closure condition, but in any case the way
seems to be promising for further
investigation.
Also the problem of a correct slot design
requires first an investigation about e
 5-1

Calculation of Maximum and High Lift Characteristics of

Multi Element Airfoils

Wiliy Fritz, Dornier Lufifkhrt GntbN, Deutsche Aerospace

Posqach 1303, D-7990 Friedrichshafen, Wist-Germany

Introduction

The flow field around multi-element aerofoil sections possesses a high degree of complexity. Due to the strong
interactions between wakes from the upstream elements and and the upper surface boundary layers devcloping
on the downstream elements, there are thick viscous layers present over the upper surface of the trailing edge
flap. Large regions of separated flow can be present, even for conditions well below maximum lift. Finally
the flow around the leading edge slat can become locally supersonic, even for low frecstream Mach numbers,
due to the large suction Ievcls induced in this region.
'I'he viscous-inviscid interaction methods, which are most widely in use for the prediction of multi-element
aerofoil flows, are cornputationally very efticient, but unable to describe many of the complex flow features
present.
A more complete description of the physical phenomena can be achieved only by methods based on a solution
of the Reynolds-averaged Navier-Stokes equations, in conjunction with a suitablc turbulence model. The
generation of a suitable computational grid then becomes a major problem. Within the present work, a me-
thod to generate suitable block structured grids around multi-element aerofoils has been developed. The
Dornier 2-D block structured Navier-Stokes solver has been extended for grids with arbitrary block structure.
After very encouraging results for a two-element high lift system at low angles of attack [ 3 3, in this work the
method was applied for realistic two- and three-element high lift systems at high angles of attack.
Grid Generation
As the configuration can become very complex, a generation of structured grids without any singular points
is completely impossible. On the other side, discontinuities in the computational grid can produce large nu-
merical errors in the solution. In order to minimize those effects, a grid generation technique, which avoids
singular points and/or discontinuities in the grid spacing at least along the surfaces, has been developed. Sui-
table C-mesh type grids around each component are patched up to an inner grid. This inner grid is ctnbedded
into an outer C-type mesh (Figure 1). To guarantee a smooth point distribution, the elementary grids are
normally generated numerically using elliptical partial differential equations for the point distribution:

where X = (xa) being the physical and (g, q+) the computational coordinates. The cocficients A, B and C are
fixed by the transformation relations between the physical and the computational spacc, only the control
functions QI and QJ can be used for grid control and/or adaptation of thc grid lincs [ 3 1. I'he discretization
of the limiting boundaries is performed according to the condition
I
WoAsi = const. (2)
I
with s being the arc length of the boundary line. W is any weighting futiction such as geometric or arithmetric
stretching. 'This elliptical grid generation technique has normally the best smoothing qualities, but it fails for
large concave slopes and sharp corners which are usually present at multi-elemcnt aerofoils. 'I'herforc a hy-
I perbolic grid generation technique is used in such regions:

wed difficulty in generating grids about surfaces with concave curvature or slope discontinuities. Therefore a
more general marching algorithm suggested by Pulliam in [ 2 3 has been applied in the present work. The
original algorithm of Steger and Chaussee is reformulated such that numerical dissipation can be added and
5-2

Governing Equations
Navier-Stokes Equations
The Reynolds averaged Navier-Stokes equations describing two-dimensional, unsteady and compressible
flows in conservation form are given by
au aF ac
-+-+-=o (4)
at ax ay
where:

r PV 1

and
au
uxx= - I [ - -ax + -1aya~ - 2 [ p + p ' 1 au3 7
uyy = - A[ -+ -3 - 2 [ p + p ] -,av
au av
ax ay ay
au av
oxy = - CCL + P,IC ay 1 9

Q = - k - aT Q v = - k - aT
X
ax ay

with density p and mcan total energy per unit volume E:

E = pe -t0.5p(u2+ v2) (7)

'The coefficient p is the laminar viscosity and p, is the eddy viscosity which takes into account the effects of
turbulence. This eddy viscosity has to be estimated by the turbulence modell.'The thcrrnal conductivity k is
given by

The perfect gas equation of state is used to define the mean static pressure p via the internal energy e:
P = CY - Ilpe (8)
The bulk viscosity A is defined as A = -213 p, the turbulent Prandtl number is fixed at 0.90 and the ratio of
specific heat, y , is maintained constant at 1.4.

Turbulence Model.
The Reynolds-averaging of the Navier Stokes eqautions introduces a further set of unknowns, the so called
Reynolds stresses. Those Reynolds stress components are normally related to the mean flow quantities
through a set of additional equations that represent the turbulence model. In the present method either an
algebraic turbulence model, based on the well known Baldwin-Ixmax turbulence model with some extensions
for free turbulence, or the Lam-Bremhorst two equation Lmw Reynolds Number k-6 turbulence model can
be used.
The algebraic turbulence model computes the eddy viscosity by algebraic relations from the mean velocity field
taking no account of the transport of turbulence. This model is only used during the calculations in the coarse
meshes within the multi-level grid technique. In the Lam-Bremhorst k-E model, thc eddy viscosity is com-
puted using two transport variables, which are the turbulent kinetic energy k and the turbulent dissipation rate
t.
 5-3

(9)

with the production term

Pt
Pk=7[7=-+7
aU
ax xy
au
(-+-
ay
aV
ax +
I - - & ( - +23 - )
aV
-ay . au
ax
av
ay . .

pkoS5y Pk2
R,, = -
CLl
Rt= -
. PlE

c f , = 1.44 P = PI + jL1

4 = = [1 - exp( - O.0165Ry)]( I + - ) 20.5

R,

f i = l + ( - - )005
4
= I - exp( - ~ ; j

By introducing the additional damping functions J; ,fi and f,,the Lam-Bremhorst k-c model maintains the
High Reynolds-Number formulation and takes into account for the influence of molecular viscosity that is
I not negligible in wall regions, the normal velocity fluctuation damping exhered by a solid boundary and the
presence of a nonisotropic contribution to the dissipation rate of turbulence that becomes dominant in the
viscous layer. As bot11 turbulencc models don’t use any wall functions, both require a very fine grid resolution
of all viscous layers.

I Finite Volume Method.

The finite volume approach in combination with a Runge- Kutta type multi-stage time-stepping scheme de-
veloped by Jameson et al. [4] is used for the numerical solution of the above equations. Applying the integral
form of equation (4)

to each cell of the computational domain separately where all physical properties are defined to be constant,
the resulting system of ordinary differential equations in time are solved by the following multi-stage Runge-
Kutta type time-stepping method:

I
5-4

with the coefficients

“1 = 4 2 = 0.6, “3 = 1.0
for the three stage scheme and
1 1 a3 = -
3 1 a5 = 1.0
a4=-
aI=4, a2=- 6, 3,
8 ’
for the five stage scheme. n denotes the previous time-level and P represents a spatial (central and therefore
second order) difference operator. The viscous terms are treated using central differences throughout the do-
main and one-sided formulas in the wall normal direction at solid surfaces.
FiItehg Technique.
To prevent an odd-even decoupling, blended second and fourth order artificial dissipation [6] is used. If the
filtering technique is applied only once, stability analysis indicates the best damping property as well as the
largest extension of the stability region to the left of the real axis giving latitude in the introduction of dissi-
pative terms. In practice, the fourth order fdter is active throughout the computational domain except in areas
with larger pressure gradients where the second order fdter takes over. Filter fluxes through walls and/or
symmetry lines are avoided. In order to minimize the numerical dissipation especially near walls, the dissipa-
tive fluxes in those regions are normalized by the ratio of the local Mach number to local isentropic Mach
number.
Convergence Acceleration

Introducing the residual averaging approach [6] i.e. collecting the information from residuals implicitly, per-
mits stable calculations beyond the ordinary Courant number limit of the explicit scheme. Furthermore, as
long as the steady state is of interest, a variable timestep approach has been used accelerating-convergence.
The most effective way of accelerating the convergence rate is furnished by the multigrid technique ([SI and
[6]). The general idea behind any multigrid time stepping scheme is to transfere part of the task of tracking
the evolution of the original system onto coarser grids by introducing a sequence of grid levels. During a
multigrid cycle first the flow properties of the finer grids (index h) are collected onto the coarser grids (index
2h):

4;) =p5
4

1
v2h

Then a forcing function

4

P2h = c R h ( u h ) - RZh(4?)
1

is defined for each coarser grid level and this forcing function is included into the titnc stepping scheme for the
coarser grids.
3 (1)h - 3,
(0)- a,dt(R$) ps +
...

The corrections of the coarser subgrids are then interpolated back to the finer grids by bilinear interpolation.
Besides of the faster convergence rate of the coarser subgrids the computational effort per time step is drama-
tically reduced on the coarser grids. Duriig the multigrid cycle, the coarser meshes arc generated by e l i i n a -
tion of alternate points in each direction. Therefore each cell on a given grid corresponds to a group of four
cells on the next finer grid.
Additional reduction of computing time can be achhieved by a multilevel grid technique. The solution process
begins in a coarse mesh. This solution then is interpolated to the next finer mesh up to the finest grid. In each
of the different grids the multigrid strategy can be applied. At each switch to the ncxt finer grid this grid can
be adapted by the results of the previous coarser grid.
 5-5

Block Structure
The computational space is devided into multiple blocks. During the solution process the flow field is stored
on an external storage device and will be updated block wise. For each block a complete time step including
the complete multigrid cycle is performed, then the updated data are stored back. At the block interfaces the
blocks are overlapping by two cell rows so that the physical fluxes and the fourth order dissipative terms can
be constructed correctly across the block boundaries. Across the block boundaries switches between coarse
and fine grid resolutions and between the Euler and the Navier Stokes equations are possible. Each block

,
always gets the actual boundary values of the neighbouring blocks.
Boundary Conditions.
I

The following boundary conditions are valid for all calculations:

At the solid wall boundary no-slip conditions are implemented and the flow is assumed to be adiabatic.

At the boundaries of the computational domain fixed and extrapolated Riemann invariants are introduced as
farfieldconditions. At an outflow situation the tangential velocity component and the entropy are extrapo-
lated from the interior, while at an inflow boundary they are chosen to free stream values.

I Results
I
The first test case was a laminar airfoil with a 30" deflected thick trailing edge flap. Figure 1 shows the com-
putational grid. More details of the grid generation technique can be seen in Figure 2. It can be seen, that the
singular point is placed in a region, where the flow can be expected inviscid, so that total pressure losses, which
might be produced by the discontinuities in the curvature of the grid lines across this point, will not influence
the viscous region along the flap. Figure 3 shows the predicted flow field for Ma = 0.15, Re = 1.7 x lo6 at an
angle of attack of 12 ". This angle of attack has been found in the experiment to be shortly before the lift break
down. As the airfoil is a laminar type airfoil, it tends to a leading edge separation. In the veloctiy field there
seems to be a small separation bubble at the airfoil leading edge. The gradients of the pressure contours in
Figure 4 also indicate a small separation bubble at the leading edge of the main airfoil. The development of
the boundary layer is demonstrated by the total pressure loss contours in Figure 5. As it follows from this
picture, the flow along the main airfoil seems to be shortly before separation, whereas the flow along the upper
side of the flap is fully attached. Above a thin core flow along the upper side of the flap, there is a very large
viscous region produced by the wake of the main airfoil. Figure 6 shows a comparison of flow field and
pressure distribution for two different angles of attack. The angle of attack a = 14" was observed in the expe-
riment to be completely separated. The same can be seen in the calculation: the flow along the upper side of
the airfoil separates at the leading edge and by this there is a complete lift breakdown. At a = 12" the agree-
ment between calculation and experiment is very good. Altough the pressure distribution for a = 14" looks
very smooth, it is not a stable solution. The convergence behaviour, which is illustrated in Figure 7 for those
two cases, shows that for the separated flow field the lift and the drag are oscillating, whereas for attached
flows, they converge towards a constant value. The breakdown of the lift also can be seen in in the convergence
history plot.
As the experimental results of the above test case are not very reliable, it is not a very suitable test case for a
validation of the separation prediction of the method. Therefore the NLR 7301 generic airfoil with flap, of
which the experimental results are documented in [ 7 3, was used as next test case. Figure 8 shows the com-
putational grid. Compared with the frrst test case, the geometry is very smooth. The rather thin flap is de-
flected 20 O. Figure 9 gives a comparison of calculated and measured surface pressure distribution for two
different angles of attack. In both cases the agreement is very good. In Figure 10 the mean velocity profiles
at 4 different positions are depicted. The frrst one is exactly at the trailing edge position of the mean airfoil and
shows the velocity profile of a strongly retarded flow. The other three plots show the mean velocity profiles
along the flap surface at about 33%, 66% and 100% of the local flap chord. Due to the increase of pressure
in that region, the wake of the main airfoil is further retarded and spreads up. Figure 11 finally shows the
comparison of a computed and a measured c/ versus a curve for this configuration. The position of the l i t
breakdown as well as the maximum lift are predicted quite good. Of coarse there is a large disagreement bet-
ween computed and measured lift in the separated region, as the computed lift coefficients are mean values
of fluctuating values.
Figure 12 shows the computational grid around a realistic 3-element high lift system. Very fine C-type meshes
are imbedded into coarser outer grids. Due to the hyperbolic grid generation technique in the inner regions,
the grid lines are orthogonal along the surfaces, even in the concave regions of the slat and the main airfoil.
The results of the flow computation for the Reynolds number Re = 1 . 9 0 ~ 1 0and ~ 12.2" angle of attack are
presented in the Figures 13, 14 and 15. The pressure contours in Figure 13 indicate constant pressure at the
trailing edge cove of the main airfoil, but considerable pressure variations in the slat cove. The total pressure
loss contours in Figure 11 give an impression of the viscously dominated flow regions. Separation occurs in
the coves of the slat and the main airfoil. Along the upper surface of the main airfoil there is a confluence of
5-6

the wake of the slat and the boundary layer developing along the upper side of the airfoil. Above the flap, the
flow field consists of the very thin boundary layer of the flap, followed by a region of an inviscid core flow,
which is limited by the viscous region produced by the common wakes of the two preceeding components.
At this configuration, it was found out, that an accurate discretization of the slat region is of great importance
in order to obtain a proper solution. Total pressure losses which are produced by numerical effects (such as
improper grid spacing) in that region, influence the further development of the slat wake and by this also the
flow field along the upper side of the main airfoil and that of the flap. Finally Figure 15 presents the surface
pressure distributions compared with the experimental data of ref. [ 8 3 . Thc agrcement is very good, only
at the slat and the main airfoil, the suction levels are not matched exactly by the calculation.

Conclusions
The presented grid generation technique renders the generation of block structured grids around multi-element
aerofoils. The technique of embedded C-type meshes can be applied to high lift systems consisting of any
number of elements. For a certain configuration, the method can be highly automated and the regeneration
of a grid for modified flap or slat settings can be done very quickly. The results of the presented work are quite
encouraging. They show, that even for realistic high lift codigurations without any simplifications on the
geometry, the prediction of the l i up to the maximum Lift is possible. The correct prediction of the drag is
much more difficult than that of the lift. As the drag is estimated by an integration of the surface skin friction
and the surface pressure distribution, it is very sensitive to pressure fluctuations in regions with large "drag
areas". Such regions are the slat and the main airfoil coves.

Refevences

1. J. L. Steger, D. Chaussee
Generation of Body-Fitted Coordinates Using Hyperbiic Partial Differential Equations
SIAM J. Sci. Stat. Comput. 1 (1980), 431-437
2. T. J. Garth, T. H. Pulliam, P. G . Buning
Navier-Stokes Computations for Exotic Airfoils
A I M Paper AIM-85-0109 1985
3. W. Fritz
Numerical Simulation of 2-D Turbulent Flow Field$ with Strong Separation
ICAS Paper ICAS-88-4.6.4 1988 1988 Miami Beach, Florida, USA
4. A. Jameson, W. Schmidt, E. Turkel
Numerical Solution of the Euler Equations by Finite Volume Methods Using Runge-Kutta Time Step-
ping Schemes.
A I M Paper 8 1- 1259 1981
5. A. Jameson
Solution of the Euier Equations by a Multigrid Method.
Appl. Math. Comput., 13 1983, pp327 - 356 1983
6. L. Martinelli, A. James on
Validation of a Multigrid Method for the Reynolds Averaged Equations.
A I M Paper 88-0414 1988
7. B. van den Berg
Boundary Layer Measurements on a Two-Dimensional Wing with Flap
NLR Bericht NLR TR 79009 U
I
8. M. van Lent, L. R. Buning, P. B. Rohne
High Lift Experiments on the GARTEUR AD (AG08) Two-Dimensional Model M3 in the NLR LST and
HST Wind Tunnels
NLR Report NLR TR 89124 C 1989
Figures
e 5-7

-1
.................
..::::;::.........
.,..
;:........::::::::::::::::
~ ~~~~

Fig. 1
5-8

--
e--
$
.a-

-
DO AL 3 FLAP = 30 DEG. GRID C RE 2.51 *lO**b M-0.200 AL. -12.
PRESSURE CONTOURS ( CP I

Fig. 4

- -
Airfoil h i t h 30 Degree Troiling Edge Flop M 0.6 Re 1.7*10..6
Toto1 Pressure Loss Contours I - PI/PtO I

Fig. 5
 5-9

Fig. 6

Residual

con~wgmccBehaviour
L
Airfoil with 30 ’Trailing Fdge nap a = 12’

I M=0.15 RI I
6
1.7x10

1,

--
Fig. 7
a ---e---

a = 14’
5-10

Fig. a

-
___
0
FLRP
M A I N AIRFOIL
EXPERIYENI
I EWERIMEN1

Fig. 9
 5-11

NLR 73 Generic Ai! Vlap

0 e

Fig. 11
F i g . 10

F i g . 12
5-12

3 -Element Airfoil
Pressure Contours
-
RE 1.9 * I 0 4 M-0.220 AL. -12.2
ICPI
v /
F i g . 13

3- Element Airfoil RE
Total Pressure Loss Contours
- 1.9 *IO**b M-0.220 AL. -12.2

F i g . 14

>Element Airfoil:
Surface Pressure Distribution ill I = 12.2'
Si81
M = 0.22 Re = 1.901IW

Main Ai6011

F i g . 15
 4
6-1

Navier-Stokes Computations of ”hrbulent Flow

Around High-Lift Configurations
P.Bartsch, W.Nitsche, M.Britsch
htitut Nr Ld-U d R a m
TecmUNdWBerlin
Marchstr. 12-14
D-loo0 Berlin 10
0-Y

1 SUMMARY work is to imestigatethe capabilitiesand restrictions of

lhispapa presentsNavis-Stokeacalculationsof the tur- this appmwh when pndcthg the flow Edd m d high-
bulent flow around two diffennt high-lift configurations, lift configurations.
for which eJrpaimcntal data exist. In the dculations, The following sections outline the. mthemhd ‘ modcl
the. flow Edd M considend of steadyatate and two- aswellasthendcalsolutionprccedmusedinthe
dimensional. Because of the low Mach numbers, the pnsent calculations. Ihc imestigated high-lift wnEgu-
fluid is heated as inwmpfedble. The solution p r o c ~ rations are introduced then, while results are presented in
dure u8e8 a Enite volume method in order to solve the the next section. Thc paper is concluded with an outlook
Reynolds-averaged Navis-Stokeaequations. The effects on to future.w o k
of turbulence on the mean Bow Eeld are describedby the
k-e hdmknce m d . The computational mesh is sys-
twnatidy refined in order to asses numerical solution 3 MATHEMATXCAL MODEL
emns.
3.1 Mean Flow Equatiolls
The results presented in this papa include surface pns-
The mean flow Eeld is consideredas of skdy-state. two-
sure. distributions as well as mean velocities and turbu- dimensionaland incompressible. The timeamaged con-
lence quantities. If possible. the calculations are com-
paredtoexperimentaldata tinuity and momentum equation read then

2 INTRODUCTION
At present, the development of efEcient high-lift config-
urations is mainly bawd on wind tunnel testing. How-
ever,as m m and monPowaful computers have become where Uj denotes the mean velocities and uj the cone-
available, the ‘numerical experiment’ evolves as an in- sponding fluctuating velocities in the zj-direetions. P
aeasingly helpful design tool.
stands fM Static pressure.and p and p fOr VkCOSity
The flow Eeld amund high-lift configurationshas a vey and density of theBuid,respectively. An overbarindicatea
mmplex nature, e.g. Bow spafation may occur and vis- time averaging.
cous&ectshwetobetalrenintoaccounL Hence,@
resolve the steep variations of the dependent variables
in a numerical calculation, many grid nodes are neces- 33 ’hrbuleneeModel
sary, making the. flow simulation &er expensive. In The standard k-e model [11 is used in order to relate
addition. due to the often complex geometry, numeri- the Reynolds stresses - p Z in Eq. (2) to mean Bow
cal grid generation may also pose a problem. Because quantities:
ofthese r e q h e n t s , calculationsof the complete three
dimensionalflowfieldarebeyondthescopeofthepresent (3)
work. Hence., the two-dimensional Reynolds-averaged
Navier-Stokea equations am.solved in order to calculate The eddy viscosity pz is given by:
the turbulent flow around multiclement airfoils. The
standard k-e turbulence model [11 is used to describe tur- Y
bulence effects on the mean flow field. The aim of this Irt = P”’T (4)

\‘
0
6-2

The local distributions of the turbulent kinetic energy k

and the dissipationrate &are derived from semi-empirical
1 TCl I TCZ I
msport equations: 0.12 -
slat cs

chord length [m] wing c 0.60 0.90

flapcr 0.21 036

I angle of attack a I 0°-16' I 82' I

with Pk being the production rate of k:
flap defleetion
(7) 21.80

The model constants appearing in Eq. (4-6)have their 1 RcynoldsnumberRc 1 1 . 6 1~ 6 I 1 . 8 106

~ I
standard values as proposed by Launder and Spalding 111:
Table 1: Characteristicsof the conEgurations
{ C , , , C ~ , , C ~ ~ , U ~=, ~{0.09,1.44,1.92,1.0,1.3}
~ }

equations over a control volume. The present procedure

3.3 Boundary Conditions employs a general non-orthogonal&d and Cartesian ve
Because of the elliptic character of the transportequations, locity components. All dependent variables are stored
boundary conditions are necessary along all boundaries in the centre of the control volumes ('colocated arrange-
of the flow domain. ment'). Central differences are used to discretize the
diffusive fluxes, w h e m the convective fluxes are ap-
proximated by a combination of upwind and central dif-
Inlet At the inlet plane, the values for the mean stream- ferences. The coefficients are arranged in the 'deferred
wise velocity U and the turbulent kinetic energy k are corntion' scheme as proposed by Khosla and Rubm
prescribed using experimental data. In cases where k is [21. The resulting system of linea equations is solved
not available from experiments, the following relation is by means of the strongly implicit method of Stone [3].
Used: based on an incomplete LU factorisation. When calcu-
2
k = -Tu%' (8) lating incompsible fluids. there, is no explicit equation
3 for pressure. Instead, pressure and velocities are linked
with Tu being the turbulence level. The mean cross- via the SIMPLE algorithm of F'atankar and Spalding 141.
stream velocity V is set to zem. The dissipation rate E Due to the colocated variable arrangement, unphysical
along the inlet plane is derived from: oscillatorypressure fields have to be suppressed by using
a special interpolationpractice for the cell face velocities,
(9) see [51.
Underrelaxation has to be employed for achieving con-
where L represents a typical turbulence length scale of
vergence of the iterative solution procedure. l).picaIly,
the energy-containingeddies.
the relaxation factors are set to 0.7 for velocities and
0.1.. .0.3forpressuredependingonthegridsize. Relax-
Outlet The streamwise gradients of all dependent vari-
ation factorsfor kand&aresettoO.land0.7,respectively.
ables are set to zero along the exit plane. Convergence is declared when the maximum normalized
residual drops below a given limit, which is set to 0.001
in the present calculations.
Walls Instead of resolving the steep gradients in the
vicinity of a wall, a wall function approach [l] is used
to describe the fluxes through the wall-adjacent control 5 TEST CONFIGURATIONS
volumes. In the present calculations, the logarithmiclaw The high-lift configurationsinvestigated in this work are
is employed. shown in Fig. 1. Throughout this paper, they will be
referred to as TCl and TC2. respectively. TCl comprises
a leading edge flap ('slat'), the main airfoil and a trailing
4 NUMERICAL SOLUTION PROCEDURE edge flap ('fowler'). TC2 consists of a NACA 4412 main
The numerical solution procedure is based on a conserva- airfoil and aNACA 4415 flap airfoil. Somecharacteristics
tive finite volume method. In this approach. it is not the of the configurations (such as chord length, Bap angles
transport equation itself which is approximated but a flux etc.) are summarized in Table 1. For both configurations,
balance derived from a formal integration of the transport experimental data are available [6,7].However. TCl. 89
 6-3

used in the present calculations, differs slightly from the 7 3 Results lor TCZ
wind tunnel model. In the available geomehy data. slat
Despite the apparentlysimpler geometry. the second con-
and main airfoil are 'smoothed out' in order to allow a
figuration turned out to be a computationally more dif-
parabolic flow solver to be applied to this configuration.
ficult test case. According to experiments [71, the Bow
The Reynolds number based on chord length and inlet
remains attxhed on the main airfoil, whaeas a small
velocityis Re = 1.6 x 106 inthecaseofTC1. The angle
separated region exists near the trailing edge of the
of am& is varied from a = O0 to 1 6 O in increments
flap. The flow oyer the flap is charactaid by strong
of 2'. Both, free sheam and wind tunnel conditions are
wakehoundaylayer inteJactions.
simulated numerically. In case of W2.the Reynolds
number is Re = 1.8 x 106. 'Ihe flap deflection angle is Fig. 7 shows pressure coetlieient distributions for this
set to 6~ = 21.8O. as in the experiments. configuration. Results obtained with an upwind (UDS,
dashed line) and a central differencing scheme (CDS.
solid line) are compared to experimental data. The v a y
6 GridGeneration strong suction peak at the leading edge of the wing is
In order to obtain satisfactory numerical grids, the flow not caphued by the calculations As a conseqmce, the
Eeldisdividedintowed subdomains.asshowninFig.2 pressure remains too high throughout the suction side up
for TC1. For each subdomain, an €I-type grid is created to the trailing edge of the main airfoil. The numdcal
and then all subgrids are connected. Because of this result compares quite favourably to experiments on the
approach, it is not possible to assure smooth transition pressure side. More severe deviations are found on the
from one subgrid.to an other with respect to mesh size flap airfoil. .Again. the suction peak is computed too
and aspect ratio in all cases. Fig. 3 shows the resulting small. No separation is predicted at the trailing edge,
numerical grids for both configurations. which is in conIradiction to experimental data. As for the
main airfoil, better agreement between calculation and
experimental data is found on the lower side of the flap.
7 RESULTS Funheron, the influence of the approximationscheme on
the numerical results is evident from this Egure. Only
7.1 Results for TC1 a small improvement of the computed c,-distrihtion is
Fig. 4 shows streamlinesof the turbulent flow around the found on the main airfoil. However,on the flap the CDS
conEguration at different angles of attack (a = 4O, So, results agree much better with experiments than theUDS
1 2 O and 16"). The qaration bubble behind the slat be results, except near the flap trailing edge. An o v d view
comes smaller with increasing angle and finally vanishes of the flow domain is presented in Fig. 8. It shows cal-
completely. No separation is found on the upper side of culated proEles of the U velwity. The influence of the
the fowler flap. The flow field on the rear of the slat and conEguration on the upstream flow Eeld can be deduced
through the gap between slat and main airfoil is shown from this Egure. In addition, the jet emanating from the
in Fig. 5. It depicts velocity vectors (Fig. 5a) as well as gap between main airfoil and flap as well as the develop-
prosles of the streamwise velocity U Fig. 5b). The angle ment of the wake can bee seen. A more detailed view of
of attack is a = 16' for this case. The reverse flow near the flow Eeld shortly behind the flap is given in Fig. 9. It
the leading edge of the main airfoil indicates the location shows distributions of U at two locations (z/c = 1.323
of the stagnation point on the lower side of the wing. and 1.558) downstream of the flap in comparison with
experiments. Again, results obtained with UDS and CDS
Both, free stream and wind tunnel conditions were in- w plotted. As is to be expted. the gradients became
vestigated when calculating the flow around this con- steeper when using the central differencing scheme b e
figuration. The influence of the walls can be seen in cause it does not introduce 'numerical diffusion' as is the
Fig. 6, which shows pressure coefficient distributions wse with the upwind differencing scheme. As a c o n s
cp = (P - P,)/(O.SpU&) in comparison with exper- quence, thecomputedspnxdhgrateofthewalnbecomes
iments for both test cases. In Fig. 6a.free stream con- smaller, which leads to a much better agreement of the
ditions are used as boundary condition, while Fig. 6b distributions with measurements. see Fig. 9a Nevathe
shows the distribution derived from simulating wind tun- less, the 'gapjet' is still predicted too weak and the u p p
nel conditions. In general, a better agreement between mixing layer profile which stems from the main airfoil
calculation and experiment is found in the latter wse. boundary layer is not captured wrrectly. see Fig. 9a Cal-
The c,-distributionsaround the main airfoil as well as the culated as well as measured distributionsof theReynolds
fowler flap comparc quite favourably with experimental shear stress --W are given in Fig. 10. In general the
data. The deviations at the upper side of the slat and near computed profiles agree reasonably well with the exper-
the leading edge of the main airfoil are mainly due to imental data. The minima and maxima locations of the
an insufficientgrid resolution in these regions. Neverths calculated distributionscorrespond with the p i t h s of
less, measurements and computationscornparequite well. the points of inflection in the mean veldity proEles, see
However, it most be noted that the shape of the lower side Fig. 9, as to be expected because of the production rate
of the slat, as used in the calculations, differs from the term of the turbulent kinetic energy in the k-r turbulence
original geometry as already mentioned in Section 5. model.
6-4

73 Emestimation References
Inordertoassessthenumerical solutionerror,allcalcula-
tions are performed on different grids, which are refined [l] h d e r . BE.. Spalding, D.B.. ‘The Numerical
inaconsisientmannu. Eachcoarscgridcontvolumeis Computation of Turbulent Flow’, Comp. Me&
dividedintofourhe gridcontrolvolumes,thusdoubling Appl. Mech. and fig., Vol. 3, pp. 269-289.1974
the number of dements in esch direction. Ihe applied [21 Khosla P.K.,Rubm. S.G.. ‘A Diagonally Dominant
enorestimationmethodisbasedonRichardsonextrapo- Second-OrderAccurate Implicit Scheme’, Contpur-
Mon. see Caruso et al. [8]. Assuming that the numerical ers and Fluids, Vol. 2, pp. 207-209,1972
error can be cxpnssed as a Taylor series, the error ch on
a mesh with size h becomes for a discretisation scheme [31 Stone, H.L., ‘Itaative Solution of Implicit Approx-
withfirstordertruncationerror imations of Multi-Dimensional Mal DifIaential
Equations’, SIAMJ. Num. Anol.. Vol. 5, pp. 530-
558.1968

where 0 denotes the exact solution and @h the numerical [41 FWuku,S.V. Spalding, D.B., ‘A Calculation Ro-
cedure for Heat, Mass and Momentum Transfer in
solution. On the coarser grid (with mesh size Zh). the
errorreads: lluee-Dimensional Parabolic Flows’, Int I. Heat
Mars Tmfw, Vol. 15, pp. 1787-1806.1972
a@
= @zzn - @ = a h - + O(h2)
€221 (11) [51 M.F’eriC. R. Kessler and G. Scheuerer, ‘Compari-
ah
son of Finite-VolumeNumerid Methods with Stag-
Subtraction of Eq. (11) from Q. (10) gives an approxi- gend and Colocated Grids’, Computers & Fluids.
mation of the numerical error on the Ene grid Vol. 16, No. 4, pp. 389403,1988
a@
@zh - @A =Ah-
ah
Ez: f h + O(hz) (12)
[61 Szodruch. J., Kotschote. J.. ‘Grenzschichfmessun-
gen an einem Transsonilrprosl im MBB-LSWT’,
Fig. 11 shows as an example the solution error derived ErgebnisberichtNr. 36. MBB GrnbH., 1983
from Eq. (12) for the streamwise velocity U and static [71 Adair. D.. Home, C.. ‘Turbulentseparated flow over
pressure P. In the case of the streamwise velocity and downs- of a two-element airfoil’. Experi-
(Rg. lla), the error is well below 5% except for a small ments in Fluids. Vol. 7, pp. 531-541.1989
region near the slat In the case of the static pressure
(Fig. llb) the error amounts to about 10%near the slat [81 Caruso, S.C., Feniger, J.H., Oliger. J., ‘Adaptive
with a maximum of up to 25% in a very small region. Grid Techniques for Elliptic Flow Problems’, Rept.
Therefore, with an even Ener grid, better resolution of the NO. TF-23, Thennosc. Div.. Stanford University,
strong suction peaks in these regions can be expected. 1985

8 CONCLUSION AND OUTLOOK

The results presented in this paper clearly show the capa-
bilitiesas well as some restrictionsof Navier-Stokes com-
putations using k-E turbulence modelling to predict flow
around high-lift conEgurations. In general, good agree
ment with experimentaldata is found for the two different
2D-conEgurations investigated in this work. However,
when consideringthe details of the flow Eeld, some devi-
ationsbetween calculationsand measurements are found.
When discussingresults such as these, numerical solution
errors have to be separated carefully from deEcienciesof
the applied turbulence model.
In order to improve the reliability of Navier-Stokes codes
for calculating the flow around high-lift conEgurations.
Erst of all. more detailed expeximental data must be avail-
able. 'Ibis includes mean flow as well as turbulence
quantities. such as fluctuating velocitiesand pressure and
distributions of turbulent kinetic energy and Reynolds
sh’esses. Based on these. data, further numerical inves-
tigations have to be performed in order to facilitate the
selection of an appropriate turbulence model for numeri-
cal simulationsof high-lift flows.
 6-5

Figure 1: Test Configurations: a) TCI b) TC2

12

2I
Figure 2:
11

Subdomains for grid generation (XI)

6

Figure 3: Numerical grids (grid lines ommed for clarity) a) TCI b) TC2
 Fwre 4 Computed streamlines ( X I )

Figure 5: Computed flow field between slat and main airfoil ( E l )

a) Velocity vectors
b) Profiles of the streamwise velocity U
//,
 6-7

Slat Wing
. . . . a
.....
. ......;...;...
.. ..
, . e .

. . . . . .;....:.....;....;...- . . ....;...;
I...J.-..C..-*...,

cP , . I .

....:....;-...;....:.-..;....
.. .. .. .. .. .. ... ... ... ......................... .. .. ..
..............................................
. . . . . . . . . ,
. . . . . . e

....:...: ....:....:...: ....:.... .....................

. . . . . . . . .
...A...: _._.:
* . . I

...... ... ......-*.....

, I . .

. e . .

>-
. * . .
8..

....;-..;....;-..;..-
0
-C
. .

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

....................................
,

.....................
, . . *
,
.
.
.
.
.
a
I
e
.
.

. . . . . . . .....................
. . . .
.
,
e
. .
.
e
.

. .. ........ ..
...I-..,
, . e
C...*...l
.

.....;-..;
., ......;...;
.* .....
......................
. . a .

0.0 0.2 0.4 0.6 0.8 1.0 0.0 1.0

XJCS XIC %
Slat Wing Fowler
b
C .. .. .... .....
...a.. .a... .C

P .. .. ...,...,...
...,...,.
.
.. .. . a .

.......................... ..
, . . *
. . . .
....................
, . a .
. a . *

.....-.,.-..
, * . I
, . e .

. ..-..,..
-..,..-
,
,

.
. .,...
. .

*
~
.
.

a
C...
I
a
..s

*
.

, . e .

....................... .
,
,
.
.
e
I
.
.

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

,. .. .. .*
,
. a
*
.
o

..-,...,...-
.** ,**.. ..-,
.. . .
, . e .
C.-. ...L
, . a .

.......
. . . ..... ,....

0.0 0.2 0.4 0.6 0.8 1.0 0.0 1.0

XIC

Figure 6: Surface pressure distributions (TC1) 0 Experiments(61, -present calculations

a) free stream conditions
b) wind tunnel conditions
6-8

Wing Flap
-12 -3
CP CP
-10
-2
-8

-6 -1

-4 0
-2
1
0
2 2
0.0 0.2 0.4 0.6 0.8 1.0 0.00 0.10 0.20 0.30 0.40
x/c [-I x/c [-I
Figure 7: Pressure coefficient distribution (TC2)
0 Experiments[7j,- - - - - (UDS) - (CDS) present calculations

U velocity H= 60.3 m / s

Figure 8: Computed profiles of mean streamwise velocity U (TC2)

 6-9

x/c=1.322 X/C= 1.558

-10 0 10 20 30 40 -10 0 10 20 30 40
U [m/sl U [m/sl
Figure 9: Mean streamwise velocity profiles at two locations downstream of the flap (TC2)
-
0 Experiments 171, - - - - (UDS) -(CDS) present calculations

x/c= 1.322 x/c=1.558

-20 -10 0 10 -20 -10 0 10

- -
-uv [m2/s2] -uv [m2/s2]
Figure IO: Reynolds shear stress profiles at two locations downstream of the flap (TC2)
0 Experiments [q,-present calculations
6-10

.:...........
.:.r....:.:.::::::w
::::=:$$$::

.m<a
.....
...........................
.........:.........
... >.:*.,:.,
...................
1.........
:
..
.

.:.:.:.:.:;.: +.&,
*.*. ::.::::.
, w

Figure 11: Numerical solution error (TCI)

a) Streamwise velocity U
b) Static pressure P
 7-1

EFFICIENT SIMULATION OF INCOMPRESSIBLE VISCOUS FLOW

OVER MULTI-ELEMENT AIRFOILS

Stuart E. Rogem
N. Lyn Wiltberger
Dochan Kwnk
NASA Ames h u c h Center
~.il stop 2 5 8 1
Moflett Field, CA 94035-1000
USA

1. SUMMARY of challenging problem to the numerical inratigatom.

The i n c o m p d b l e , Visco~,turbulent flow over dn& These indude problem involving turbdent boundary
and multi-element airfoils is numerically simulated in an layer separation, confluent boundary layern and w h ,
dhcient manner by solving the incompressible Navier Reynold. number &e&, three-dimensional &e&, com-
S t o h eqnatiolu The solution algorithm employ. the p d b i l i t y &&, transition, and complex geometries.
method ofpnr.ndocompreadbility and utilirea UI upwind- Although the problem ue inherently three-dimenaional,
diiTerencing scheme for the convective fluxes, and an im- there is still much to be learned about the flowphysia
p&t line-relaxation scbeme. The motivation for tbm by studying two-dimensional models.
work indndes interedt in studying high-lift t a k e 4 and
landing configorations of various aircraft. In particular. The computational toob available range from the more
accurate computation of lift and drag at &M &ea dhcieut and simpler inVircid/vhns coupled method..
of attack up to .tall in desled. Two different turbu- to a Reynoldbaveraged Navier-Stoh (RANS) analysis.
knce models ue tested in computing the flow over an An example of the former method h given by KMUUWG
NACA U12 UI accurate prediction of stall is o b et al.I They uae f d potential method coupled with UI
tained. T h e approach nsed for multCelement airfoils in- integral boundary-layer method. These method. have
volra the II(K ofmnltiple sones of structnnd grids fitted been found to be s u d d in accurately computing the
to each element. Two different approaches are compued, prrnare dmtributiou for multi-element airfoils. indudmg
a patched system of grids, and an a a l a i d Chimera s y b c w up to maximum lift, zome of which involve =par*
k m of grids. Computational nsdt. ue presenkd for tion. The coupled method h u been p r o m to be n8ef11I
two-elemat, t h h e n t , and four-element airfoil con- an dtective engineerins design tool. Thin method L
f i g ~ r a t i o Ex&nt
~. agreement with experimental sur- limited by ita inability to compute beyond maximum lift
f.ec p m u r e d c i e n t . is san. The code convergea in ditio io^, and may have problems with certain featom
lean than 200 iterations, reqniring on tbe order of one of nome airfoil system such as flap wells, thick trailing
minuk of CPU time on a CRAY YMP per element in edges, or unsteady dleetr.
the airfoil configuration.
Navier-Stokes calcdations for high-lift syskms have
been investigated by a number of anthorn."' Schnskr
1. INTRODUCTION and Biickelbaw' computed the flow over a two-dement
An increlad nndemtanding of high-lift system8 d airfoil using a structured, compressible, RANS solver.
play an important role in designing the next generation The grid system nsed was a pointwise patched oyskm
of trmwport aircraft. Current designs for such aircraft with t h m zones, with Cgrids around both the main
typically involve multiple elements. such M leading edge element and flap, and another outer Cgrid surround-
data and multiple-slotted flaps. The current trend is b ing those. Good resdt. were obtained for low Reynolds
ward a more &dent, yet simpler deign which d kad number turbulent flow. The next two anthorn, Barth:
to reduced manufacturing and maintenance coat.. At the and Mavriplis' both nsed an unstructured grid a p p r o d
ume time, increves in lift coefficient. for a given angle of to handle the difficnlty of dmretising multi-element ge-
attack and increves in maximum lift d c i e n t will lead ometries. They were e d able to produce accurate pres-
to a luger payload capability. Improad d+ d rlso sure caffiaent information on the airfoil surfaces. The
allow for rednced noise in areas snrrounding airport.. accuracy of the unstructured grid a p p d , however, U
Uuderstmding of high-lift flow physics harbors the p limited because of the very large aspect ratio of the tri-
tential to i m p m e airport capacity through a reduction angdu cells required to reaolve high Reynolds number
of .a airplane's wake vortices, allowing dmer spacing boundary layer flows. Abo, thin approach is not well de-
between subvquent airplanes taking off and landw. veloped for three-dimensional problems. Large compu-
tational resources ue required, eapecially CPU memory,
Increased knowledge of the flow p h p i a involved with to make these methoda work for V ~ flora. M Unstmc-
high-lift a y s t e m in thereton of greakr inkmthan ever turd methods ue currently generating a lot of intereat
before M the need to improve over current daigns be- in the reaearcb community; improvements to these limi-
comes ~ t e Study . of these configurations d r e q h tations are to be expated in the near futon. Until mch
both compntatioual and experimental efforts. Computa- a time, the current anthorn believe that astmctured grid
tional fluid dynunicd (CFD)is playing a large role in thin approach ia the most suitable for solving &M multi-
work. Mdti-element configurations plaeut a number element problems in two and three dimensions.
1-2

The current work use8 an incompnnible RANS flow over a &gle airfoil. One of the studies for thin problem
solver to compute the flow over multi-element airfoib. includa a compuimn of the Bddwin-Buth turbulence
Two different grid approaches are used; the first a p modd with the Baldwin-Lomax" algebraic turbulence
p m h employs the patched grids utilised in Rd. 2, modd.
and the second use8 an overlaid grid approuch known M
the Chimera scheme.' The current work examines sev-
eral airfoil flow problems in two dimensions in an effort 4. COMPUTED RESULTS
to chuackrise current capability to numerically study Thin d o n describa results of the current compnta-
such problems. Grid topology, computational efficiency, tions for T OM M e r e n t airfoil configuratioos with one.
and resulting accuracy are issues to be examined in the two, three, and four elemenb.
current work. An incompressible flow solver is being nti-
bed becawe the flow conditions for take-off and landing
rill generally be less than a Mach number of 0.2. In the 4.1. NACA 4412 Airfoil
actual flow, comprewibility effects will generally be con- Calculations were performed for the d o n over an NACA
fined to a small lochliaed region, such M near the m a 4412 .irfoil at a Reynolds number of 1.52 million. A
of a leading-edge slat. Since the incomprasible Navier- Ggrid with dimensions of 241x63 WM nsed, with wall
Stokes system hu, one leea equation than its compreaaible spacing on the order of lo,-' which corresponds to y+
counterpart, less computing resources u e required. values on the order of one. The grid WM computed using
a hyperbolic grid generator.'' A dose view of this grid
is shown in fig. 1.1. In order to compuk flow quanti-
3. ALGORITHM ties for the points on the computational boundary in the
The current computations .%re performed nsing the 'wake cut" line of the e-mesh, two lines of dummy points
INSZD computer code which solving the incompreeaible are added such that t h a e dummy points coincide with
Navier-Stokea equations for steady-state flows' and un- points on the other side of the w&e line. The first line of
steady computations.' This algorithm has al.0 been a p t h a e dummy points is updated by injecting values from
plied to problems in three dimensions using the INS3D- the coincident inkrior points on which they lie. Using
UP code! The code is based on the method of utifi- thin overlap produces smooth solutions to the eqnations
cial comprewibility a8 developed by Chorin' in which a across this computational boundary. Thin p d u r e llso
pseud*time derivative of preeaure is added to the con- add. dummy points inside the airfoil. T h a e poinh ue
tinuity equation. Thus the convective part of the qua- merely blanked ont and never used in the solution p m
tions form a hyperbolic system, which can be iterated in cedun. All of the Ggrid. in this work use this overlap.
peudc+time until a steady-state solution is fonnd. For
unsteady problems, subiterations in pseudwtime are per-
formed for each physical time step. Since the convective
terms of the resulting equations are hyperbolic, upwind
differencing can be applied to these terms. The cur-
rent code nses dux-differencing splitting modeled .iter
the scheme of Roe." The upwind differencing leads to
a more dihgonally dominant syskm than does central
differencing and does not reqnire the additional use of
utifidal diedipation. The system of eqnations is solved
nsing a GauskSeidel type line-relaxation scheme. The
line-relaxation scheme is very useful for computing multi-
ronal grids because it makes it passible to iteratively p u s
AQ (which is the change in the dependent variables for Fig. 1.1. 241x63 grid used for flow over an NACA 4412
one time step) information between the ronal boundaries
airfoil.
M the line-relaxation sweeping takes place. The result is
a semi-implicit p d g of boundary conditions between
zoned, which further enhances the code stability." The This flow wa8 computed using two different tnrbulence
molting code ia very robust and stable. It is capable of models, the Baldwin-Buth model" and the Baldwin-
producing steady-state solutions to fine-mesh problems Lomu model." Figure 1.2 shows a compuimn between
in 100 to 200 iterations. More detail about the computer these computations and the experimental r a u l t s of C o b
code can be found in Refs. 6-8. and W.dcock'6 at hn angle of attack of 13.81 degrees,
which U very neuly muimum lift conditions. In the ex-
Moet of the present calcnlations used the turbulence periment the flow separated at approximately 85 percent
model developed by Baldwin and Buth,"~" where the of chord. Tripstrips were employed in the experiment
specific formulatiou found in Ref. 12 waa used. Thin is a on the suction and prewnre surfaces at chord locations
one-equation turbulence model that avoids the need for of +/e of 0.023 and 0.1, respectively. The cornput-
an algebraic length scale and in derived from a simpli- tions thus specify these M the transition points. For the
-
fied form of the standard k e model equations. In the Baldwin-Barth model this is implemented by setting the
current application, the equation is solved using a line- the production terms to iero upstream of t h a e locations;
relaxation procedure similar to that used for the mean- for the Baldwin-Lomax model the eddy viscosity is set to
flow equations. Thin model haa been found to be very zero upstream of the transition location. The agreement
robust and easy to implement for multiple-body configu- is fairly good, with the biggest discrepancy occurring at
rations. The next section indudes compntations of flow the trailing edge where the predicted preeaure is too h i h .
 7-3

However, the Bddwin-Barth model does give a flattening ala0 suffen from too small of a separation region. The
of the preaanre over the aft 15 percent of chord, indicat- c u e without transition ahowa the bat agrement with
ing flow separation, where the Baldwin-Lomu solution the experimental profi*l.
d o a not show thia kndency. This figure .bo shows that
the cornputations with the transition predict a leding-
edge laminar separation bubble. The experiment reporb
that there WM no laminar separation bubble at this an-
gle of attack, so an additional computation WM mu M-
ing the Badwin-Barth model with the production k r n u
turned on everywhere, thus the boundary layer WM fnUy
turbulent. The pressure coefficientfor this is rbo shown
in fig. 1.2. There is a slight improvement in the tr.iling
edge area for this solution. The Bddwin-Lomu modd
showed no differencein the prcmure or velocity solution
when it WM run without specifying transition, except

j
that it removed the laminar separation bubble.
I I I I I
4 0 S 10 1s 20

Fig. 1.4. Lift coefficient VGRMangle of attack for flow

-2
over an NACA U 1 2 airfoil.

1 '
0.m ais O i l 0.75 1.0d
XIC

Fig. 1.2. Pressure d a e n t on surface of an NACA

4412 airfoil at Reynolds number of 1.52 million com-
paring cdcnlations with Baldwin-Barth m d Baldwin-
e
L o m u turbulence models and experimental d a t r

0. 0. 0. 0. 0. 0. 0. 020.40.60.88.0
WC
I I
I
Fi.1.5. Pressure fodfiaent on the sorfux of the NACA
om'- Ly YY airfoil at 16 d e w angk of attack for 6rva Merent
o a , ! ~ o o J L _ o a ( ! ~ o o J I
0 0 5 1 o a ( 1 om1 timn during the unsteady periodic motion.
IJVd

Fi. 1.3. Velocity profiles on upper s d a c e of an NACA Computations were run for a range of anglea of attack
4412 airfoil at streamwise stations of z/c = 0.62,0.675,
from aero lift to maximum lift. The lift coeffident v e m
angle of attack is plotkd in Fi. 1.4. Thb shows that
0.731, 0.786, 0.842, 0.897,m d 0.953.
the Baldwin-Buth solution with transition giva very
good agrement in the lift, including the prediain of
Velocity profiles from the suction surfem boundary layer stall. For all casea, M the angle of maximum lift WM
u e plotted in fig. 1.3 at streamwise stations of 2 / C = approached the flow tended toward unsteadynm. That
0.62, 0.675, 0.731, 0.786, 0.842, 0.897. m d 0.953. The is, the Bteady-stak computations did not converge com-
profiles are shown using the stream& component of pletely, which, for the arrifichl compradbility f o r n u b
velocity in boundary-layer coordinates, that ia, the w tion m e w that the mdb do not satisfy the continuity
locity component tangential to the b e d airfoil surface. equation. In these casea the code WY then run in a t h e
This figure shows in greakr detail the problems of the accurak, unsteady mode. For the Baldwin-Barth model
Baldwin-Lomax soluti- in this region: the boundary with transition, at an angle of attack of 14 d e y a , the
layer profile is too fd and the solution shows only a Unskdinea dies out when the computations .IC 11111 in
tiny region of separation. The Baldwin-Buth solntion a timeaccurate mode. At 16 degrees, an nMkady ped-
is in closer agreement with the experimental re~nlt.,but odic behavior ensues; M shown in the figure, the mean
7-4

lift d r o p sharply below the vdues from smdler an& of much l u t e r than the Baldwin-Buth model. The com-
attack. Examination of the flow shows that the leading- puting time on a Cray YMP required for thin 241 x 63
edge laminar r p u a t i o n bnbbk is periodicdly shedding mesh is 100 second. for 200 iterations when using the
and traveling through the boundary layer on the top sur- Baldwin-Barth model, 90 seconds for 200 iterations with
f.ce of the airfoil,and past the tr.iling edge. F i g u n 1.5 the Bddwin-Lomu model. W-hen running the uuskady
shows the pressure coefficient on the ourface of the airfoil caaa, the algorithm requires subikrations at each p h y s
at .even &Rerent times through the period of this flow. ical time step to drive the divergence of velocity toward
The forming of the leading-edge vortex is evident, and zero. When running the unsteady 16 degree angle of at-
it can be . a n that it travels downstream and past the tack case with a non-dmensiond time step of 0.05, 40
trailing edge. physicd time s t e p resulted in one period of the flow.
T h i took about 10 minutea of computing time. Due to
For the Baldwin-Barth model without transition at an the diflicult nature of d v i n g the unsteady incompress
angle of attack of 16 degreea, dl oscillations damp out ible Navier-Stokes equations. it proably would not be
and it converges to a steady-state solution. At 18 de- computationally cheaper to use an incompresible formu-
green, the lift continues to oscillate periodically, yet there lation over a compresaible Navier-Stokes code to study
is only a slight drop in the lift, and there is a complete a b pt-stdl, unsteady airfoil flows.
r n c e of a leading-edge separation bubble. The Balwin-
L o m u computations do not have any type of periodic
unsteady behavior with or without tranaition. The re- 4.2. Tw*Element Airfoil
sults from thin model show that a drop in lift d o a not The geometry is made up of an NACA 4412 airfoil with
occur nntil an angle of attack of 20 degrees. an NACA 4415 flap deployed at 21.8 degrew, with the
entire configuration at 8.2 degrees angle of attack. Thin
In short, the Baldwin-Barth model shows promise for geometry was the subject of a wind tunnel experiment
une in predicting high-lift flows, and although nome defi- by Adsir and Horne." The chord Reynolds number was
ciencies are shown here, it is significantly better than the 1.8 million, and the Mach number in the experiment was
Baldwin-Lomax model. In addition, the Baldwin-Barth 0.09. The blockage in the wind-tunnel was r v e r e enough
model is much easier to uea than the Bddwin-Lomu that the wind-tunnel walls needed to be included in the
model, in that is d a s not require a length scale; it is calculations in order to get good agreement with the ex-
straightforward to implement for a multi-element airfoil perimental pressure coefficients.
computation. AU of the results in the later sections of
thin paper use the Baldwin-Barth model.

Fig. 2.1. T h r e e o n e patched grid used to compute flow

over an NACA 4412 airfoil with an NACA 4415 flap.

The airfoils were discretized using two different grid a p

..................... 4 ............. ........................ proaches. The first foUows the work of Schuskr and
Birckelbaw' and U- 3 zonw which arc patched together
!
using coincident points. This grid is shown in Fig. 2.1.
50 100 150 UM Each of the elements is surrounded by a Cgrid, and
these two grids are Surrounded by another C-yid which
Iteration Number extends out to the wind-tunnel w&. The dimension of
these grids are 374x44, 241x33, and 352x32, respectively,
Fig. 1.6. Convergence history showing Maximum resid- for a total of 35,000 points.
ual versus iteration number for flow over an NACA 4412
urfooil at 13.87 degrees angle of attack. The second type of grid U- a Chimera' approach, in
which Cgrids were generated about each of the elements.
To indode the effects of the wind-tunnel walls these grids
The convergence history is shown in Fig. 1.6 for the an- were inset into a third zone compoeed of an h-grid. A
gle of attack of 13.87 degrees for both turbulence mod- partial view of these three grids is shown at the top of
els with and without transition. In general, fast con- Fig. 2.2, with a close-up of the main-element grid in
vergence is seen, with converged solutions obtained in the vicinity of the flap shown in the bottom half of thin
100 to 200 iterations. Specifying the transition tends figurn. These grids had dimensions of 261x49. 203x35,
to produce an unsteady component into the flow field and 121x61, for a total of 27,500 points. To implc-
which somewhat delays the convergence. It can .Is0 ment the Chimera approach, these grids are given to the
be seen that the Baldwin-Lomax computations converge PEGSUS' code. This code first punches holes into grid.
 7-5

w h m they overhp a body (M shown in the bottom of and the experiment. Then WM m mbiiuity in the way
Fw. 2.2). It then computed the interpolation iten& in which the flap paition is defined.
wed to update the flow q u m t i t k at the fringe points
of t h a e bok., m d to update the flow qnmtitiea at the
outer bonnduia of grida which Lie inside mother grid
(like the onkr bounduia of the c-grids aeen in the top
of Fw. 2.2). For both the Chimera m d the patched grid
approaches, the apuinrr next to the ourfacm WM ut to
2 x lo-', which correspond to y+ d n a at the w d l m
the order of one.

...........

Fig. 2.4. Convergence history for flow over two-element

airfoil for patched grid m d overlaid grid schema.
Fq. 2.2. Overlaid Chimera grid wed to compute flow
over m NACA U 1 2 airfoil with m NACA 4415 gap.
Figure 2.4 shows the convergence history for these com-
putationa. They both converge m y well, madting in a
steady-date dntion in about 100 iterations. E& of
t h a e grid c u a t h about 100 rsondn of CPU time on
a Cray YMP for 100 ikrations. The eode NIU at about
a r a k of 80 MFLOPS, m d mains 36 X CPU SGC-
ond. per grid point per iteration. Since the Chimera a p
prcuh wen .bout 20% fewer grid points, it take a tittle
& kr computing time. The major diflerrnce between these
a p p r c u h a is the amount of time m d dort it t h to
generak the grids. The patched grid CMC t h on the
order of mverd honn of work; it involva generating in-
ner bound& which define the anrfacc with the proper
point diafribution to CMM that the prid. c.ll be patched
together. Then hsperbolic grids u e mucbed hdfway
.QOI the gap from each of the elemcnta. The rad&

X ing outer bonnduia of t h a e ue merged into a common

inkdace where they match. The inner grid8 ue recdm-
Fu. 2.3. P n r m Caffident on ourl.oe of two-dement Iakd to match this interface. Finally, the onkr Cgrid
airfoil comparing both patched grid m d overlaid grid is mucbed outward using a hyperbolic grid generator.
achema to U p a t n c n t d madts. The p- b talio~~and is not d y repeatable for
a different CMG ( n m flap placement, or flap angle), or
for a Merent geometry. On the other hand, the om-
The eompntationd madta compue w d with the exper- I d gid.u n be generated in only a matter of minukni
imcatd d t n of Adair m d Horne." A plot ofthe prrc one need only generate two independat b r r b d i e Pria,
a w e cocS&nt on the E& of the dementa ia dmwn about each of the dement., a d then feed t h a e into the
t Fig. 2.3. Rault. from both of the grid ap&a U PEGSUS code6 M dacribed above. Once t b h u ban
*nn.The b e a t dXarcnce between the computation ut up for one CMC it U very e q to reproduce it for
m d experiment I.. a n in the auction peak at the *ding mother cue or mother geometry. It in for t h a e I'CUOM
&a of the 841. The dXerence might be expLincd by that the overlaid grid apprcuh WM d o p k d for the mat
a dSacuce in the geometry between the com~~t.tioiU of the cua m d geometrk in thim work.
7-6

the cdculations indicate that the flow becomes unsteady

beyond that angle of attack. The skin friction dong the
surf- of the airfoil elements is shown in Fig. 2.7. It
can be ueen that the flow separates at the trailing edge
of the flap even at zero angle of attack, and that this
separation reduces in size with inclersing angle of attack.
The main element has trailing edge separation occurring
at angles of attack of 12 degrees and greater. It becomes
massively separated at an angle of attack of 16 degrea.
At thin angle of attack the flap shows evidence of a vortex
pYsiug over the top because of the large dip in the skin
friction on the surface of the flap.

0.031 I

Fig. 2.5. Velocity profiles from overlaid grid calculations

compared to experimental data.

0.00 0.25 0.50 0.75 I .00

x/c
G

E
Fig. 2.6. Coefficient oflift versus angle of attack as com-
puted by the overlaid grid approach for the twc-element
airfoil.

Figure 2.5 shows velocity profiles from the Chimera cal- 1.o 1.1 1.2 13
culatious at three locations on the top surface of the main x/c
element and flap. These are plotted with experimen-
tal measurements of the profiles by Adair and H o r ~ e . ~ ' Fig. 2.7. Skin friction on the surface of the main element
These plots show fairly good agreement with the experi- and flap for various angles of attack.
mental results. The biggest discrepancy is the difference
in the gap velocity off the surface of the flap's leading
edge. This is related to the difference seen in the pres-
sure coefficient plot in Fig. 2.3. The velocity profile from 4.3. Three-Element Airfoil
the trailing edge of the flap shows that there is a sepa- The three element computational configuration was
ration occurring over the top surface of the flap. This taken from an experimental geometry of a supercritical
profile shows that the compntational separation bubble airfoil which has been tested by Valarero et al.18 This
is not as thick as that seen in the experiment, but that airfoil consisted of a leading edge slat deployed at -30
the computations do an excellent job of capturing the degrees, a main element, and a trailing edge flap de-
wake from the main element in this region. ployed at 30 degrees. The experimental Mach number
was 0.2 and the chord Reynolds number was 9 million.
Further calculations were carried ont using free-stream The Chimera approach was used to discretize the g+
outer boundaries (neglecting wind-tunnel walls). These ometry and produce a computational grid. A Cgrid
calculations use the overlaid grids with two zones, where was placed around each element, with the main-element
the main element grid extends beyond ten chord lengths grid extending out to the far field. The grids for the
from the airfoil. These were run at various angles of slat-, main-, and flapelement had dimensions of 221x41,
attack to show the capability to compute maximum lift 401x75, and 221x47, respectively, for a total just under
conditions as well post-stdl conditions. The curve 51,000 points. The top of Fig. 3.1 shows every other
of lift coefficient versus angle of attack is shown in Fig. grid point in the first and third element grids, with the
2.6. The lift drops off sharply at alpha = 15 degrees, and resulting holes caused by the main element. The second
half of this figure shows the main element grid. The wake
cut boundary ofthis grid haa been digned just above the
top surface of the 0ap element in an attempt to put U
many points UI powible in the wake and boundary-layer
region found there.

Fig. 3.1. Grid and geometry for the the-element airfoil,

showing every other grid point around the slat and flap.

Pig. 3.2. Velocity magnitude contour8 at 20.4 degrees

angle of attack. s' I
0.0 0.5 I .o IS
XIC

Figure 3.2 shown veloaty magnitude mutours for the Fig. 3.3. Pressure coefficient comparing computation
three element configuration run at 20.4 degrees M.& of and experiment for angles of a t t d of 8.1, 20.4, and
attack. The wake of the slat is clearly seeu ecroea the 23.4 degrees.
top of the succeeding elements. The experimentd resulta
ofVdueso et al." and the compntatioud results of this
itudy are compared in Fig. 3.3. These figures show pres-
slue coefficients on the snrfacea of each element a t three 10 II i
diflerent angles of attack, 8.1, 20.4, and 23.4 degrees.
Very good agreement is seen except on the suction aide
of the &C. Also, there is a discrepancy ou the upper
surface of the flap trdling edge. The experimentd re
sults show a strong adverse pressnre gradient followed
by a flattening in the pressure c d c i e n t curve, which
is generally evidence of flow separation. The computr-
tiond results do not show this. This is probably due to
the generd trend of the turbulence model to underpre
did the amount of separation. The experiment allowed
free transition on the elements, and the computations a s
sumed a turbulent boundary layer everywhere. Further
work in this area could indude use of a transition model
for this calculation.

Convergence histories of these computations are shown 3

in Figure 3.4. These computations converge well, with
steady state wlutions being obtained after 200 iterations,
which corresponds to about 4 minotes of CPU time on
a Cray YMP. Fig. 3.4. Convergence for the thretelement airfoil.
4.4. Four-Element Airfoil The geometry wan discretiaed using the Chimera a p
The geometry is made np of a NASA 9.3 percent blunt- p m h . Cgrids were generated around each of the el-
hued, supercritical airfoil with a leading edge slat d e ements, with the main element grid being muebed ont
ployed at -47.2 degreen and two trailing edge flap at to the onter boundary. Thene grids were overhid and
30 degree and 49.7 degrees respectively. Thin configu- the PEGSUS’ code wan used to create an overlaid grid.
ration matches the geometry nsed in the experimental Approximately 55,000 points were used in the resulting
work done hy Omsr et al.” The Mach number in the composite grid in order i o d v e the flow p h y d a de-
experiment wan 0.201 and the chord Reynolds number qnakly in the boundary layers and waka. The grid spac-
WM 2.83 million. ing next to the s n b of the nirfoib wan which
ensurea y+ vdnes of one near the w d .

The compntationd nanh of this study were compared

with the the experimental results of O m u et al.”. Fig-
nre 4.1 shows velocity magnitude contonrs uonud the
four element confignration at 14.25 degrea angle of at-
tack. In thin figure the wake from the leading edge slat
is appuent over the main element. Snbeequent w h
from the main element and f l a p can .Is0 be observed.
Plot. of the pressure codficient on the surfaces of the el-
emento at angles of attack of 0.0, 8.13, and 14.25 degrees
Fig. 4.1. Velocity magnitude contours at 14.25 degreen U+ shorn in Fig. 4.2. Again, excellent agreement is
angle of attack. seen except there is once more evidence that the compn-
tatiou of the flow over the flap underpredicts the smonnt
of separation at the lower angles of attack.

-81 , I I I I

I I I I

- -
lo 0 100 200 33
Iteration Number
Fig. 4.3. Convergence for the fonr-element airfoil.

The convmgace hiaories for the four element configure

tiou at three &erent angles of attack are shown in Fig.
4.3. The computations converge weiI and a l t e d y state
solution L prodnad after abont 200 iterations. which
cornsponds to approximately fonr minntes of CPU time
on the Cray YMP.

0.0 I .o K. CONCLUSIONS
xx: An incompmacible flowd v e r h u been naed to compnk
Bow over several airfoil geometries for the pnrpme of
Fig. 4.2. Pressnn coefficient comparing compntniion developing a tool to stndy takeoffand Ian* config-
and experiment f o ~angles of attack of 0.0, 8.13, and urations. The code b robust and produma nnmerical
14.25 degreen. simnhtiona in a matter of minntea. Tbe Bow over an
 7-9

NACA 4412 airfoil was investigated, and the Baldwin- Stoka Equations.” AIAA Piper 88-2583, June 1988.
Barth and Baldwin-Lomax turbulence models were com-
pared. The Baldwin-Barth model gave signific.ntly bet-
ter result., and was much easier to use, particularly for
8.~~.
See .Im AIAA J., 242, February 1990, pp 253-262.
S. E., Kwak, D., and Kiris, C., ‘Numerical
Solution of the Incompressible Navier-Stoka Equk
multi-element flows. The use of the Chimera overlaid lions for Steady-State and TimrDependent
grid approach was found to be much easur &an using a Problems,” A I M Paper 89-0463,Jmuuy, 1989. See
patched grid scheme for solving multiple element airfoil .Im AIAA J., 29,4,April 1991, pp 603-610.
flows. Both approaches are capable of producing &xu- 9. Chorin, A. J., “A Numerical Method for Solving In-
rate solutions. A m r a t e pmrsure pndiction was shown c o m p d b l e V h n s Flow Problem.” J. Compnt.
for geometries with two, three, and four airfoil elementa. Phys., 2, 1967, pp. 12-26.
The common discrepancy between these calculations and 10. Ra,P. L., “Approximate Riemann Solvera, Param-
experimental results involve0 separated flow. The reuults eter Vectors, and DiRerence Schemes,” J. Compnt.
for the NACA 4412 airfoilindicate that deficiencies with Phya., 43,1981,pp 357-372.
the turbulence model are the most likely cause of t h a e 11. Rogers, S. E., “On The Use of Implicit Linc
in.ceolacie0. Work in prognss with different turbukncc Relaxation and Multi-Zonal Compuathns,” AIAA
models s h m promise in remedying this. Investigation Technical Note 91-1611-CP.AIAA CFD Conference,
of other turbulence models and theix implementation for Honolulu, HI, June 24-27, 1991.
a multi-element &il c d d a t i o n rill be the focru of 12. Baldwin, B. and Barth, T.. ‘A OncEqnation Tu?-
future work. In addition, future work will include the bnlence l h n ~ p o r tModel for High Reynold8 Num-
extension of the cnrrent work to three dimensions. ber Wall-Bounded Flows,” NASA TM 102847, Aug.
1990.
13. Baldwin, B. and Barth, T., “A One-Equation Tur-
6. REFERENCES bulence Transport Model for Hiih Reynolds Num-
1. Kusunwe, K., Wigton, L, and Meredith, P., “A her Wd-Bounded Flows,’ AIAA Paper 91-0610, Jan-
Rspidly Converging Vions/Invisdd Coupling Code uary, 1991.
for Multi-Element Airfoil Configurations.” AIAA Pa- 14. Baldwin, B. and Lomax, H., “Thin Layer Approxi-
pcr 91-0177,January, 1991. mation and Algebraic Model for Separated Turbnlent
2. Schuster, D. M. and Buckelbaw, 1. D., “Numerical Flows,’ AIAA Paper 78-257, January, 1978.
Computations of Viscous F l d e l d s about Multiple 15. Cordova, J. Q.and Barth, T. J., =Grid Generation for
Component AirIoils,’ AIAA Paper 85-0167, January, General ZD Regions Using Hyperbolic Equations,’
1985. AIAA Paper 88-0520, January. 1988.
3. Barth, T. J., ‘Numerical Aspects of Computing V i s 16. Coles, D. and Wadcock, A. J., ‘Flying-Hot- Wire
cons H i Reynolds Nnmber Flows on Unatrnctund Study of Flow Put an NACA 4412 Axoil at M d
Meahen,” AIAA Paper 91-0721,January, 1991. mum Lift,’ AIAA J., 17.4, 1979, pp 321-329.
4. Mavriplis. D., Qrbulent Flow Calculations Using 17. Adair. D. and Home, W. C., Tnrbulent Separated
Unstructured and Adaptive Mnhes,’ ICASE Report Flow Over and Downstream of a TmElement Air-
9041, September, 1990. foil,’ Experiments in Fluids. 7, 1989, pp 531-541.
5. Benek, J. A., Buning, P. G., and Steger, J. L., ‘A 3-D 18. Vhlamo, W. O., Dominik. C. J., McGhee, R. J.,
Chimera Grid Embedding Technique,* A I M Paper Goodman, W. L., and Puchal, K. B.,‘MuIti-Element
85-1523-CP,July 1985. Airfoil Optimization for Muimum Lift at High
6. Rogers, S. E. and Kwhk, D., “An Upwind Difler- Reynold8 Numben,” AIAA Paper 91-3332, Septem-
encing Scheme for the steady-atak Incompressible ber, 1991.
Navier-Stokes Equations,’ Journal of Applied Numer- 19. Omu E., Zierten, T.,Hahn, M., Szpuo, E., and Ma-
ical Mathematics, 8,1991, pp 43-64. hhl, *A., ‘Two-Dimensional Wind-Tonne1 Tests of a
I . h e r s , S. E. and Kwhk, D., aAn Upwind Differencing NASA Suptrcritical Airfoil with VMow Eigh-Lift
Scheme for the Time Accnrate Incompressible Navier- System,’ NACA CR-2215,September, 1973.
 8-1

NAVIER-STOKES CALcULAnoNS O N MULIT-ELEMENT AlRFOlIs USING A CHIMERA-WED SOLVER

Donald W.Jasper, Shreekant A p w d and Brim A. RobinSon

McDonnell Aircraft Company
McDonneU Douglas Corporation
PO.Box 516,St. Louis
Missouri, USA 63166
ABSTRACT
A sNdy of Navier-Stokes calculations of flows about multi- systems lead directly to greater operational flexibility
elemeat airfoils using a chimera grid approach is presented. through increased payloads, extended ranges. and d e m a g d
ThC chimera approach utilizes structured. overlappcd grids tale-off and landing distances. Traditional high-lift design
which allow great flexihility of grid arrangement and has relied on extensive wind NMel and flight testing
simplifies grid generation. Calculations are made for two-. which in expensive and difficult due to the extremely
&-. and four-elcment airfoils, and modeling of the effect complex flow interactions encountered. ?he complexity of
the flow interactions q u i r e s such extreme fidelity in
of gap distance betarcen elements is demoustmted for a twc-
element case. Solutions arc obtained using the thin-layer geometric representation and matching of test to actual
form of the Reynolds averaged Navier-Stokes equations operating conditions that the applicability of standard
with turbulence closure provided hy the Baldwin-Lomax scale model aerodynamic testing is limited. In order to
algebraic modcl or the Baldwin-Bprth one equation modcl. achieve optimum designs, uew tools for rapid and efficient
Tbe Baldwin-Bruth turhulence model is shown to provide analysis of high-lift configurations arc requimd
M e r a p u n e n t with experimental data and to dramatically Computational Fluid Dynamics (CPD)offens p t promise
improve convergence rates for some cases. Recently as a tool which provides valuahle insight into the flow
developd. improved farfield boundary conditions arc phenomena associated with high-lift system performance.
incorporated into the solver for greater efficiency. however to provide early design guidance rnpid and
Computed results show good comparison with efficient CFD techniques must be developed and validated.
experimental data which include aerodynamic forces.
d a c e pressures, and boundary layer velocity profiles. Traditionally (and for the foreseeable future) high-lift
systems incorporate multiclement geometries in which a
Symbols numbcr of highly loaded aerodynamic elements interact in
close proximity to each other. Figwe 1 shows an example
Section Lift Coefficient. llq-c cross section of a typical configuration incorporaring four
Section Lift elements: a leading edge slat. the main airfoil element. a
Dynamic Pressurs. 1/2pu* flap vane, and a trailiog edge flap. Such configurations
generate very complex flowfields containing regions of
Velocity
separated flow, vonical flow, and confluent boundary
Section Drag Coefficient. dq-c
layers. Laminar. turbulent. transitional. and re-
section D n g laminarizing boundary layers may exist. Although high
Skin Friction Coefficient. zw/q.. lift systems am typically deployed at low freestream Mach
M o i l chord numbers. they may still exhibit compressibility effects due
Shear Stress to the large pressure gradients generated Such complex
Mach Number, u/a flows are extremely difficult to analyze computationally
Sped of Sound and q u i r e solution of the Navier-Stokes equations to
Presswe Coefficient. @p,)/q, resolve all flowfield details. It should be uond that many
of the flowfield phenomena (e.g. scparation. transition.
Static Pressure
turbulence. etc.) arc areas of intenee research in the
Nm-dimensional Distance from Wall.
computational conmuoily md am not yet fully amenahk to
YlW(PWPW) computational analysis.
Re Reynolds Number, p,U,c/K. Confluent
Boundary
CI Viscosity

Subscripts:

-
e
P-am Quantity
Quantity at Mge of B o u n w Layex
W Quantity at wall
Separation Reattachment
INTRODUCTION GFZCOSI8-l-V-1P*
Efficient high-lift systems arc critical in achieving
optimum aerodynamic performance of future generation
military and commclrial aircraft. Well designed high-lift
Figure 1. Typical High-Lifl Configuration
 ~

8-2
0
One of the difficulties encountered in computational
analysis of high-lift systems using the Navier-Stokes
equations is the generation of suitable high quality grids
h t each element to capture all relevant flow phenomena.
Tbrsc basic appmaches to generating grids and solutions
about complex configurations have k n widely reported in
the literature: zonal patched (or blocked) grids,
M S ~ grids,
N and
~ overlapped (chimera) grids (Figure
2). The w r d approach (e.g. Refcrence 1) bas becn widely
demonstrated on a number of geometries. and bas shown
excellent accuracy in the calculation of many complex
viscous flows. Such an appmacb utiIizes structured grids
generated in separate zones or blocks which transfer
information across common boundaries. Its major
drawback in application to analysis of high-lift systems is
the necessity of matching b o u n W s at zonal interfaces.
POI typical high-lift geometries the boundary matching
places a large demand on grid generation capabilities and
can lead to degradation in grid quality (and consequently
solotion accuracy) at boundaries in close pmximity to the
high-lift system components. The unstructured grid
appmach (e.g., Reference 2) readily generates grids about
high-lift configurations, but can suffer in accuracy from
highly stretched triangular cells generated in the boundary
layer region. Unstlvcrurui grid solvers BIC the most recent
approach to the analysis of complex geomchiea and show
p a t pmnise. but they have not yet demonstrated the same
wide applicability to complex viscous flows as smctund
flow solvers. The chimera grid scheme (e.g., Reference 3)
utilizes overlapping. WucNnd grids which arc not required
tomatchathoundan'es. Structured gids h t cach element
can be generated independently with no compmmise in grid
quality. and the expriencc gained in the application of
Muchued grids to canplex viscous Bows can be utilized

The intent of the present study is to develop and evaluan a

technique for the rapid and efficient analysis of multi-
element high-lift systems. The chimera approach was
chosen because it allows for s i m p f i d generation of high
quality grids and the use of pmven shuehucd grid solvers.
Using the chimera approach, regions where more grid
resolution is required may be assigned additioual Overlawed Grid
overlapping grids without modification of existing grids.
changes in high-lift configuration can be rapidly analyzed
since independently gridded elements can be added or
subtracted at will, and movable surfaces can be easily
modclled by translation or rotation of existing grids (e.g.
Reference 4). The CPL3D EulerNavicr-Stokes solver
(Reference 5 ) modified to treat overlapping grids was
selected due to its wide application to complex viscous
flows and the generally excelIent results obtained with its
W. Initially a Baldwin-Lomax algebraic turbulence model
was used for its simplicity and easc of application.
However a Baldwin-Barth onecquatiou model (Refewnce 6)
was incorporated to address deficiencies noted with the
simpler Baldwin-Lomax model. Transition was not treated
(i.e. flows were considered fully turbulent at all points).
The chosen approach is capable of treating three-
dimmional geometries without modification.

METEOD flgure 2. Unstructured, Blocked, and

The chimera scheme is a domain decomposition scheme Overlapped Gilds About Two-Element Alrloll
&st proposed by Ben& and Steger (Reference 3). Its
application can be illustrated by considering a general
 m -
airfornflap cohfigurat&n as depicted in Figure 3.
Independent grids an generated about the airfoiland flap
elements. The flap grid is embedded within the airfoil grid
and its outer boundary is updated by interpolation from the
airfoil grid. To keep the airfoil grid from calculating a
solution inside the solid surface of the flap, airfoil grid
points which fall within the flap surface an designated hole
points and removed from the solution. An artificial
boundary is established in the airfoil grid surrounding the
hole points and is updated by interpolation from the
overlapped region of the tlap grid. In this way a two-way
communication is established between the airfoil and flap
grids, and the complete flowfield excluding regions within
solid surfaces can he calculated. The interpolation process
is outlined in Figure 4 which details a portion of the
overlap region between the airfoil and flap grids. At each
I --
Airfoil Grid
Hole Cut in
Airfoil Grid

Flap Outer Boundary

Interpolated From
Airfoil Grid

Flap Grid

Airfoil Hole Boundary

-

-
8-3

boundary two points are interpolated. Since the flow Interpolated From
solver difference stencil at any point uses information from Flap Grid
at most two neighboring points in any direction, the
solution is completely isolated from points within the Figure 3. Deflnltlonof Chlmera
hole. lntelpolatlonBoundades

Ihc chimera mlution takes part in two steps. In the F i t

step a pre-processor code is executed which detennioes the
location of hole points. determines outer boundary and Airfoil Grid
hole boundary points which must he interpolated, and
calculates the interpolation factors needed to update
boundary points. The pre-processor. code for CFL3D is
called MAGGIE (Reference 7). Bounday points an updated
using a bilinear interpolation scheme which utilizcs the
four sunounding points of Ihc interpolating grid. It should
be noted that this interpolation scheme is not
consenrative. and that interpolation errors an proportional
to the ratio of the a n a of the cell k i n g interpolated to the
fxca of the cell providing the interpolation data (Reference
8).

The second step in the chimera solution is the flow solver,

CFL3D. In this study CFL3D is used to solve the thin layer
nppmximation of the three-dimensional. time dependent, 0 Flap outer boundary pin1
conservation law form of the Reynolds averaged, updated fromairfoil grid
compressible Navier-Stokes equations. Ihe code solves Airfoil hole boundary @I
the discretized flow quations implicitly using an upwind- updated hom Rap grid
biased spatial differencing scheme with either flux
difference splitting or flux vector splitting for the Figure 4. Delall of Alrfolllflap Gild Overlap
convective and pressure terms. and cenupl differencing for Reglon lllustratlngTwo-Polnt Overlap
the shear stress and heat transfer terms. 'Ihc Roe-averaged lnterpolatlon
flux difference splitting scheme was employed in this
study. 'Ihc Baldwin-Lamax algebraic model or the Baldwin-
Barth m e quation model was used to calculate turbulent selection of the points used for interpolation. New
viscosity. In application to multi-element flows, the algorithms were implemented for determining which grids
Baldwin-Lomax model was applied independently on each would be searched for interpolation factors in complex
element grid with no attempt to match turbulent viscosity overlapping grid regions. A new algorithm for
muss chimera boundaries. It was found neca%saryto limit determining hole points was incorporated. X i s new hole
the search for the location of maximum vorticity which s e m h algorithm g m t l y improves the hole determination
d e f i e s the length scale of the Baldwin-Lamax turbulence procedure (especially in three dimcosions) and will be
model to the first peak in vorticity and not the absolute described in detail.
maximum of voniciry. 'hbulent viscosity forthe Baldwin-
Bslth model was interpolated at chimera boundaries in the The original bole s e m b algorithm implemented in
m e manner as the flowfield variables. MAGGIE is based on d e f h g boundaries which create a
surface enclosing the hole points. The outward normal
Many improvements to the MAGGlE grid pre-processor direction for each boundary is prescribed to define an
code were required to increase the speed of generating 'outside' of the boundary. Initially all candidate points an
suitable chimera boondades about complex configurations. sssumcd to he hole points. and each point is tested against
Biedmn (Reference 9) improved the search routine for the boundary to &ermine if it lies on the 'outside' of the
8-4
e
boundary and is therefore outside of the hole creation RESULTS
d a c e . A point is determined to be outside the boundary if The CP23D chimera code was applied to three multi-
the dot product of the vector from the closcst point on the clement airfoils for which test data is available. Each test
boundary to the candid& point with the outward d m was conducted to investigate physical phenomena
nonnal vector at the hole boundary point is positive associated with multi-elemsot airfoils or the detailed
(Pigure 5). 7 h i s algorithm requires some user experience in performance of a specific multi-element airfoil
defming hole creation surfaces and the division of the configuration. Due to the nature of such testing. special
surfaces into boundaries (e.g. Reference 10). The new attention to test technique and long estaMishcd guidelines
algorithm also requires the definition of a hole creation for two-dimensional airfoil testing were obsctved Since
surface enclosing the hole points, however there is no ne& the testing is aimed at acquidng aidoil data, there is no
to specify outward normals or to divide the surface into error associated with Reynolds number scaling or
boundaries. To determine if a point is within the hole geometric fidelity which would be present in scale model

-
creation surface. the number of times a ray fmm the point testing of P development tlight vehicle.
to infinty intersects the hole creation surface is determined
(Figure 6). An odd number of intersectioos indicates the Two-Element Airfoil:
point is within the hole creation surface. and an even The first test case analyzed was a modified NLR 7301
numbor of intersections indicates the point is outside of the supercritical aidoil with a slotted trailing edge flap. The
hole. In practice the new hole point algorithm has p v e n airfoil was modified such that no flow separation was
to bc much morc robust than the origind. e dlow speed (Reference 12). Tcst data is available
o b s ~ ~ at
et a hhch number of 0.185 and Reynolds number of 2.51
million based on airfoil chord. No boundary layer
transition device was used. Dsta were collected at scvcral
angles of attack for two flap gap distances (1.3% and 2.6%
ofairfoilchord) with a 20' flap deflection. A small laminar
separation bubble near the leading edge of the airfoil WUI
observed experimentally. Transition from l a m h u to
turbulent boundary layer was generally seen to OCCUI just aft
X Candidate hole point of the separation bubble on the airfoil upper surface at
abom 2.5% airfoilchord, at about 709b chord on the airfoil
O b l e creahbn boundary point

-<> 0 point outride boundary (nomhole point)

r n c 0 point inside boundnfy (hob point)
GFZM578&VW
lower d m . and 50% flap chord m the flap upper d a c e .
Ihe flap lower surfse exhibited laminar flow st all times.
The main element and tlap bcmdary layers were found to be
c o d u e n t for the 1.3% flap gap. The experiment also
Flgure 5. Orlglnai Hole Determlnatlon Algorlthm showed the flow in the flap cove region of the airfoil lower
surface to be near esparatiw at the enuance and to be
experiencing re-laminarkation of the boundary layer o w

r Ray From
Cadidate point r Hole Creation
Sudace
the flap leading edge.

Representative computatiooal grids used for this case me

shown in Figure 7. Main clement and tlap grids are of C
topology, with the main clement gtid extending to the
X candidats hole point farfield. Two grid dcnsitics were evaluated. The c ~ ~ grid
n e
had dimensions of 205x45 00 the airfoil and 179x33 on
Odd Number of Intersectionsof Hole Creation Surface the flap, while the corresponding fine grid dimensions were
and Ray From Candidate Point Indicates Point Is 424x91 and 199x41. Initial point spacing off the solid
Inside Boundary GP2cO518bV-iT4
surfaces was adjusted to yield a y+ of approximately one for
both the fme and coarse gdds. Come grid solutions .uw
Figure 6. New Hole Detennlnatlon Algorlthm made using only the Baldwin-Lcmax turbulence model.
Convergeoce of the solution showed a large dependence on
In order to increase computatioual efficiency, improved the hlmulcncc model employed ( P i p 8). The Baldwin-
farfield b o u n d q conditions based 00 the method of Barth model required appmximately 2 m iterations to
asymptotic expansions (Reference 11) have been achieve force convergence. while the Baldwin-Lomax
incorporated into the flow solver. These boundary model required in excess of 8000 iterations. It required
conditions zue derived from the @der equations about a approximately 40 minutes of Cray I "CPU time to rn
constant pressure rectilinear flow. They have been ZOO0 iterations on the h e grid using the Baldwin-Barth
demonstrated to produce more accurate results when nubulace model. wMe the Baldwin-Lamax model requid
compared to those obtained using standard farfield approximately 20% less CPU time per itemtion.
boundary conditions on a given grid Conversely. similar
ascuracy can be obtained on a smaller grid, reducing the Experimental pressure distributions ate available at 6.0'.
number of points in the f d e l d grid and thereby reducing 10.1'. and 13.1' angles of attack, and comparison of the
computation time. To apply the improved boundary predicted and experimeotal pressure distributions zue in
wndition, the farfeld boundary must be defined by analytic excellent agrsement for both the coarse and fme grids at
curves, and U currently limikd to parabolic curves for both flap gap distances ( P i p s 9 and lo). The fine grid
inflow boundaries and linear ~ulycsfor outflow boundaries. solutions did show a slight impmvement in the prediction
 8-5

To investigate the effect of turbulence model on the

Airfoil and Flap Grids solution, find grid c a l c W o n s were made for the 2.6% flap
gap case utilizing the Baldwin-Barth hubulence model.
Pnssure diatributions (not shown) are virtually identical
with those obtained using the Baldwin-bmnx model.
Improved behavior near stall is obtained using the Baldwin-
Barth model, with the maximum lift occurring at 14.1'
angle of attack, but wirh a m m gradual onset of stall than
indicated in the experiment (Figure 13). No oscillatory
behavior was noted in convergence using the Baldwin-
Bmth model. The Baldwin-Barth model also shows a
significant improvement in prediction of turbulent
boundary layer velocity profdes on the airfoil and flap
upper surfaces (Figure 14). with a much improved
prediction of boundary layer thickness.
Flgure 7. Computetlonal Gllds for NLR 7301
Allloll With W Flap Comparison of predicted and experimental skin friction
distributions show reasonable agreement for both the
B a l d w i n - h a x and Baldwin-Bath models on the airfoil
upper surface ( P i 15). Expcdmental skin friction
cafticients are available only for turbulent flow regions.
and therefore are not available for the major podon of the
lower d a c e . Comppdsnn of experimental and predicted
drag coefficients, however. rue not in good a p m e n t for

10-9
10-10 -3
0 2 4 6 8 1 0 1 2 1 4
Iterations x 1,000 o-5(Hv~

Flgure 8. Convergence History for NLR 7301

Alrfoll, 2.6% Gap
IM, 0.185, do, Se = 2.51 x 106, Flne Grid t
i I I I 1 I I I I
-1 6
of the suction peak pressure for the 1.3% flap gap 0"' -12 a- 10.V
.pameby. Comparimo of experimental and calculated lift
coefficients also show good a m e n t (Figure 11). Coarse
grid results tend to overpredict the lift for both flap gap
distances, while the fine grid greatly improves the
agmment for the 2.6% flap gap and gives a lomewhat
smaller improvement for the 1.3% gap. The 1.3% flap gap
exhibited confluent boundary layers on the flap upper
surface as shown by the flap boundary layer not recovering
to the potential velocity before merging with the airfoil
wake (Figure 12). "he relatively worn agreement of
pndicted and experimental lift for the 1.3% gap may be
attributed to a greater sensitivity of this case to the
boundary layer development, which was not modelled
correctly duc to the lamioar flow on the flap upper mnfnce
and n-lamiwizatiOn of the boundary layer in the tllp cove
region. Coarse grid rolutinns would not converge at angles
of attack bcyond 13.1'. Fie grid r e d t s a r r e obtained at
15.1' angle of attack with the 2.6% gap case exhibiting an
oscillatory convergence. At 17.1' angle of attack the 1.3% 4
gap case a l a 0 exhibited oscillatory convergence, -0.2 0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
Computations f d to ptedict the abrupt stall o b m e d Relative Chordwise Location. WC
Gp2CoIl&a.Vha
experimentally at 15.1' angle of attack. The oscillatory
convergence and fdum to pmdict the abrupt atall is Flgure 9. Computed end Expermental Pressure
attributed to the inadequacy of the Baldwin-Lomax Dlrtrlbullonsfor the NLR 7901 Airfoll
~RWCIICC modo1 in calculating the onsct of separation and
2.6% Gap. M, = 0.15, Re = 0.62 x106
the effects of massive sepadon.
8-6

-8 3.5
0 CFWDX.Gap= 1.3%
U" -6

4
a=6.V .
c
m
.-
-2

0 Coarbe Grid

I I I I I I I

-1 6 1.5
8, -12 3.5 I
a = 10.1'
c
C
.-m
$ 4

s-4
?
$ 0
a I I I I I I I
4
-16 r

2 a = 13.1' 3
Angle-of-Attack- deg
GFZ40518~1l-V-~

Figure 11. Lift Curve for NLR 7301 Airfoil

With 20" Slotted Flap
M, = 0.185. Re = 2.51 x 106
Y
I I I
4' I I I I Three-Element Alrfoll:
The second test case was a GAW(1) airfoil with a slotted
Relative Chordwise Location - WC flap and add-on leading edge slat (Reference 13). The
G4,-o5,8-io-v,t&
airfoil was tested with and without the slat present for a
Figure 10. Computed and Expermental Pressure variety of flap and slat deflections. The test was conducted
Distributionsfor the NLR 7301 Alrfoii at low speed for a Reynolds number of 0.62 million.
1.3% Gap,M, = 0.185, Re = 2.51 x106 Transition was obsemed to occur at 5% airfoil chord on the
main element upper surface, and between 50% and 65%
either the Baldwin-Lomax or Baldwin-BarIh models, with chord on the lower surface. Transition was fixed at 10%
the computational nsults two to three times higher than slat chord for runs in which the slat was employed.
the experimental values. The higher predicted drag may bd
due in part to the extensive laminar flow which exists in Computations were ma& with the flap deflected 40' with
the experiment but not modelled in the computations. and without the slat. When employed, the slat was
deflected 42'. Grids were again of C topology about each
The improved analytic boundary conditions were also element with the main element outer boundary extending to
applied IO the NLR airfoil. A grid with a parabolic outer the fdield and the slat and flap grids embedded (Figure 17).
boundary which extended only one chord length upstream Addition of the slat to the basic airfoil and flap
of the airfoil leading edge and 1.7 chord lengths configuration required no modification or additional grid
dumstream of the airfoil trailing edge was generated about generation effort to the main element or flap grids. The
the main element. A larger reference grid was also grid sizes were 303x63 on the main element. 217x29 on
generated which incorporated the smaller grid as a subset. the flap, and 193x29 on the slat. Additional e-s with a
but extended the outer boundary to 7 chord lengths refined slat grid with dimensions of 193x57 were also
upstream of the leading edge and 10 chord lengths performed. Results were generated using the Baldwin-
domstrcm of the trailing edge. Pressure distributions and Lamax turbulence model as the Baldwin-Banh m&I failed
force coefficients generated on the smaller grid using the to converge for the slatted cases,
improved boundary conditions are in good agreement with
those generated on the larger grid using the standard Without the slat computed and experimental pressures are
boundary condition. whde the standard boundary condition in good agreement (Figure 18). although the main element
used on the smaller grid shows considerable degradation in pressure peak is generally overpredicted. Some
the solution (Figure 16). The smaller grid reduced the discrepancies also occur at the minimum pressure location
number of points i n the main element grid by of the trailing edge flap. In addition, a spike in pressure is
approximately 20%. with a corresponding reduction in generated in the computations by the sharp corners
CPU time. introduced by the way in which the blunt trailing edge of
 8-7

3.5 r I

1.5
0 2 4 6 8 10 12 14 16 18
Angle-of-Aitack deg
~

G?24-3510-1+V-bcb

flgure 13. NLR 7301 Alrfoll2.6% flap

48 Gap M, 0.185, Flne Gild
-WUe[CFL3DX)
0 UN.(Exprimeot)
E 40 Lift Coefficient vs Angle-of-Attack
E I
20
16
Airfoil E 12
E
A 8
4
0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3
Flap Boundsly 0
u/u, Layer
GP2U)SlbiP.Vlp

flgure 12. Upper Surface Boundary Layer

Veloclty Ploflles at 69?4 flap Chord
NLR 7301 Airfoil, M,= 0.185,
Alpha = 6". Re = 2.51 x lo6,Fine Grid
the true flap geometry was modelled (Pigure 19).
Computations for the caws in which the leading edge slat
was present are not in as good agreement (Pigm 20). In
general the computations tend to overpredict the pressure
4
0
U-.
48
peak and overall loading of the slat and underpredict the
loading of the main aidoil and flap. This behavior is 40
clearly evident fmm the pmsurc distributions at 5 3 and 32
9.S angle of attack, while at 14.07' the computational
result appears to be more highly loaded on the main 24
element than the experiment. ' I l i a can be attributed tn a N
16
separation on the main element which occurred in the
experiment but not in the computations. 8
0
Several attempts were made to improve the calculated 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2
results for the slatted cases. The grid about the slat was
retined and different hole cutthg schemes were attempted to
ensure that the separation which occurs nuder the slat
d a c e was contained entinly within the slat grid and not
Figure 14. computed ~ w n d a r y Vebcity mer
Profilea Uslng Baldwln-Bmh and Baldwln-bmax
near the hole boundary of the main element grid. In Turbulence Models on NLR 7301 Alrfoll,
addition. the slat geometry was modified such that the
separation underneath the slat was completely eliminated. 26%Gap, M,= 0.165, a = So, Re 2.51xl@
Neither of these had any signZicant effect on the computed hole strategy considerably improved the agreement
results. between the experimental and computed pressure
distributions (Figure 23). The reason for the difference
Four-Element Airloll: appears to be the different levels of turbulent viscosity
To further validate the solver. the fourclement airfoil of which the flap grid perceives in the twn different hole
Reference 14 was analyzed. This airfoil consists of a schemes. This is confirmed by the fact that the Baldwio-
leading edge slat. a main nixfoil element, a tlap vane, and a Banh turbulence model, which interpolates turbulent
trailing edge flap. The aidofi was tested at a Mach number viscosity at chimera bounduies. did not show the same
8-8

1.1-1, level of differences. Due tu its better agreement with

i 0.032 h81 expedmcnlal data. the second hole stmtegy WM usd for nll
suhaequent c a l d t i ~ ~ .

Compnrisons of experimental and computed pressum

distributions arc generally good (Figure 24). At 8.13’
angle of attack there is a large discrepancy between
expedmental and p d i c t e d pressursa on the VMC element,
apparently due to a scparation on the vane upper surface
which was not predicted computationally. At 0.0’ ~ g l of c
attack this separation has apparently not yet developed.
and at 14.25’ angle of attack the separation region hra
Chordwise Locstion WC - GpI+Os~I)-,5V~
been reduced significantly and the computations axe again
in good agncment with experiment. Ihc likely culprit for
missing the separation on the vane ia the Nbulence model.
Figure 15. Upper Surf8ce Skin Frktbn
Results at 8.13’ angle of attack using the Baldwin-Bmh
Coefficient Distribution turbulence model show yirmally no difference with tho=
NLR 7301 Airfoil, 2.6% Flap Gap, presented for the Baldwin-Lomax model. The pressure
M, 5 0.185, Re = 2 . 5 1 ~ 1 0 ~ results presented here (using the B a l d w i n - h a x turbulence
of 0.2 and Reynolds number of 2.85 million. model) arc c ~ n s i ~ t ewith
n t those of Rogers, et al (Reference
Computational results for this airfoil have also been 15). who employed a Baldwin-Bath model.
reported by Rogers et d (Reference 15) using a chimera
approach with an incompressible Navier-Stokes solver CONCLUSIONS
employing a Baldwin-Bath turbulence model. A tcchniquc for the rapid analysis of multi-element high-
lift systems utilizing the chimera approach has been
For the present analysis C type grids were generated about evpLoatcd Navier-Stokes solutions have bccn obtained for
each element with the main airfoil grid extending to the two-, three-,and four-elemcnt airfoils using the technique.
fadield (Figure 21). lo the vane/flap region, the grids of and varying agreement with expedmentd data has been
the main airfoil. vane. and flap all overlapped. The grid demonstrated as a function of airfoil geometry. Accurscy of
interactions in this region an complex, and the solution nulti-elcment airfoil analyses depends to a large extent on
exhibid a dependency on the way the holes were d e 5 e d the complexity of the flow to be modelled For the two-
about the various elements when using the Baldwin-Lomax element case without confluent boundary layer flow.
hubuleuce model. Two different holes were cut in the vane excellent prediction of lift is obtained up to stall, with
and flap grids near the hailing edge of the main airfoil maximum lift obsewed. at the same angle of attack as the
(Figure 22) and solutions were obtained at 8.13’ angle of experiment and the predicted value of maximum lift within
attack. The fumt hole was cut around the airfoil trailing 2% of the expcrimenral value. Gwd prediction of turbulent
edge such that the vane and flap grids extended above the boundary layer velocity profiles and skin friction is also
main airfoil wake, while the second hole removed all VMC obtained. although drag is not well predicted Less accurate

I
and flap points above the airfoil wake line. The second results are obtained for the case exhibiting confluent

-
--- Law Grid. Standard BC
smell orid,Improved BC
CI
2.377
2.379
cd
,0472
,0425
’*---* SmallGrid,StandardBC 2.320 ,1362
0 Experiment 2.366 .ozs

Pressore
Coefficient

CP -2

2 I I I I I I
0.0 0.2 0.4 0.6 0.8 1.o 1.2 1.4
Relative Chordwise Location - WC
OPi)MSlblkV/,p#

Figure 16. Application ol improved Boundary Conditions

NLR 7301 Airfoil, 1.3% Flap Gap, M, = 0.185, Alpha= 6.0
 Figure 17. Glids About Three-Element AlrfOll

-CFWDX
0ExmsrimPnt

I
,
I I
2‘ I I I I I
-10 I

Figure 1% Tralllng Edge Closure of GAW

Alrfoll Flap

c
Bald--Barth model belter predicts the detailed boundary
-0.2 0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 layer panmeters and gives a more positive indication of
Relative Chordwise Location - WC stall, with none of the o s c i l l a t o ~convergence
~ exhibited
GPN051S-18-VW by the B a l d w i n - h a x modeL I b e Baldwin-Bmth model
bas also demonstrated a positive effect on convergence rate
Flgure 18. Computed and Experimental which can be signiiicant for certain cases.
Pressure Dlstrlbutlons for the
GAW Alrfoll Wlth Slotted Flap The chimera approach has shown itself to be a versatile
b= 0.15, Re = 0.62 x 106 tool which can greatly simplify the grid generation effort
boundary layer flow, and consequently modelling of flap and improve the grid quality for the Navier-Stokcs analysis
gap effect is not accurate. Calculation of gap effects may of multizlement airfoils. Accurate and efficient solutions
require modelling of transitional and re-laminanzing can be obtained about complex configurations. h i t c d by
boundary layers. Pressure comparisons for the thrw- the applicability of the flow models employed.
element airfoil are not good. probably because of the
separation which occurs on the undersurface of the leading REFERENCES
edge slat. Experimental and predicted pressurc 1. Raj, P., Olling. C.R.. Sikora. J.S.. K e n . J.M..
distributioos are in better agreement for the four-element Singer, S.W.. and Brennen. J.E.. W Dimensional
airfoil which exhibits no separation on the slat. Better Euler/Navier-Stokes Aerodynamic Method (TEAM).”
prediction of separation, stall. and post-stall results Volume I: Computational Method and Verification,
depends on better modelling of turbnleoce. WRDC-TR-87.3074,June 1989.

l%eBald--Baah tumulence model has been shown to be 2. Mavriplis, D., “Turbulent Flow Calculations Using
superior to the Baldwin-hmax turbulence model. The UoWuchurd and Adaptive Meshes.” ICASE Report
Baldwin-Bmth model allows turbulent viscosity to be No. 90-61. September 1990.
matched at chimera boundaries, which reduces the
dependence of the solution on the way hole boundaries an
deked about the various elements. In addition. the
8-10

3. Benek, J.A.. Steger. J.L.. and Doughexty. F.C.. "A

Fl&ble Grid Embedding Technique with Applicatious
to the Eder Qmtions:' AIAA-83-1944, J U 1983.
~ ~ 0 EXPERIMENT

4. Duugheriy. F.C.. and KUM. J.H.. "TmsoNc Store

Separutiou Using a Three-Dimensional Chimera Grid
Scheme:' AIAA-89-0637, January 1989.

5. Thomas, J.L..Taylor. S.L.. and Anderson, W.K..

"Nuvier-Stokes Computations of Vortical Flows h r
Low Aspect Ratio Wings." AIAA-87-0207, January
1987. - 6 1

6. Baldwin. B.S.. Barth. T.J.. "A One-Quation

Turbulence Transport Model for High Reynolds
Number Wall-Bounded Flows." NASA TM 102847.
August 1990.

7. Baysal. 0.. Ponludi. IC,and Lessard, V.R., "Multigrid

and Upwind Viscous Flow Solver on Three-
Dimensional O v e r l q p d and &bedded Grids." &
Journll,Vol. 29. No. 6, 1991.
8. Tukeshi. S.. and Jamesou, A.. "Multi-Body Flow &Id
Calculations with Overlapping-Mesh Method." AIAA-
4 L, - I

89-2179, 1989. -6

-4
9. Biedron. B.. A. S & M Inc.. Private C-unicatious.
-2
10. Suhs. N.E., and Tramel. R.W., "Pegsus 4.0 User's
Manual," ADC-TR-91-8. November 1991. 0

2
11. Verhoff, A., Stookesbeny. D.C., and Agrawal, S . , -0.442 0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6
"Far Field Computational Boundary Conditions for Relative Chordwise Location. WC
Transonic Flows.'' Fourih International Symposium QPzcMte-Z&v&b
on Computational Fluid Dynamics. September 1991. Rgure 20. Computed and Experimental Pressure
12. Van Den B q , B.. "Boundary Layer Measurements on
DlStrlbuUons for me GAW
Alrfoll Wlth Slotted Flap and Slat
U Two-Dimensional Wing with Flap." NLR-TR-
79009-U. January 1979. M,=0.15, Re=0.62~106

Airfoil
Trailing Edge

op2*ablbll-v-*

Figure 21. Four Element Alrfoll Odds Figure 22. lllustratlon of Two Hole Cutting
Technlquea for Four-Element AMOII
 8-11

13. Braden. J.A., Whipkey. R.R.. Jones, G

and Lilley,
D.E., ”Experimental Study of the Separating
Confluent Boundary Layer.” NASA CR 3655. June
1983.

14. b a r , E.. et al, ‘Two-Dimensional Wind Tunnel Tests

Airfoil with Various High Lift
of a NASA Supe~~riticsl
Systems - Volume Il Test Data.” NASA CR 2215.
April 1977.

15. Rogers, S.E.. Wiltbcrger. N.L.. Kwak. D., “Eftkient

Simulation of Incompressible Viscous Plow Over
4 Multi-Element Airfoils.” Pifth Symposium on
0 0.4 0.8 1.2 1.6 2.0 Numerical and Physical Aspects of Aerodynamic

(p21Q51epby9c
-
Relative Chordwise L d o n X/C Flows, J M U 1992.
~

ACKNOWLEDGEMENTS
Flgure 23. Effect ot Dlffared Hole Boundsrles On

M, - -
Four-Element Alrtoll Pressure DLItdbutlOn
0.2, Alpha 8.13‘
This work was suppoaui under funding of the MCDOMCU
Aircraft Independent Rcscarch and Development pmgram.
Computer resources were provided by the Numerical
Aemdynamic Simulation Progrsm at NASAAmsr Research
Center and by CRAY Research. Inc.

I I I I I
-1 2
0 -CFUDX
0 a Experimnl

B a18.W

Relative Chordwise Looation - WC

QPzco618-2cvM

flgure 24. Measured and Predicted Pressures

for Four Element Alrfdl
M, = 0.2,Re = 2.85 X I O6
 9-1

Numerical Solution of the Navier-Stokes Equations for

High-Lift Configurations on Structured Composite Grids
T. E. Nelson D. W.Z i n g G. W. Johnston
University of Toronto Institute for Aerospace Studies
4925 Dufferin Street, Downsview, Ontario
Canada, M3H 5T6

Summary Prandtl number

pressure
A numerical method is presented for the s o solution vector
lution of the compressible Reynolds-averaged, Reynolds number
thin-layer Navier-Stokes equations on structured eigenvector matrix
composite grids as applied to high-lift configu- Cartesian velocities
rations. The method is an adaptation of an contravariant velocities
implicit approximate factorization algorithm for Cartesian coordinates
block structured composite grids. Interfaces be-
tween blocks are treated hy overlapping the grids (I angle of attack
and takiig one layer of points from neighboring P relaxation constant
blocks. Turbulence is treated using the Baldwin- 7 ratio of specific heats
Barth one-equation turbulence model. 6 deflection angle
High-Eft applications presented for comparison 4,JfI spatial central difference operators
with wind tunnel data include: a NACA 4412 air- A eigenvalue matrix
foil with NACA 4415 flap, a GA(W)-1 airfoil with €9 v curvilinear coordinates
a 29% chord flap at 30 degree flap angle and two P density
gap settings, and a GA(W)-1 airfoil with 15% T nondimensional time
chord slat and 29% chord flap. Good agreement
with experimental data is obtained for cases with Subscripts
fuUy attached flow or small regions of separated f flap
flow. For cases with extensive regions of flow sep- m main airfoil
aration, the thickness and extent of the separated S slat
regions are underpredicted.
Superscript
n time step

Nomenclature
speed of sound 1 Introduction
flux Jacobians
chord length The flow around twc-dimensional high-lift con-
drag coefficient figurations is complex, often containing multi-
lift coefficient ple separated regions, confluent boundary layers
pitching moment coefficient and compressibilityeffects, even at low onset flow
pressure coefficient speeds. The performance of high-lift devices, ea-
flux vectors pecially close to stall, can he difficult to predict
PP and requires the solution of at least the Reynolds
time step averaged Navier-Stokes equations with a turbu-
Jacobian lence model. This is an expensive computational
Mach number task which is made even more time consuming by
overlap the necessity of generating a field grid.
9-2

Considerable progress has been made in solv- ditions required by the grid generator and solver
ing the Navier-Stokes equations for high-lift con- are all determined automatically. The result is
figurations. Recent papers on the topic include an E-type topology about the configuration. An
methods for the incompressible Navier-Stokes example of the block decomposition and the con-
equations [l] and for the compressible Navier- nections that result m shown in Figure 1for the
Stokes equations [Z],[3],[4],[5].Investigators have case of an airfoil and flap. Fewer blocks could be
employed structured, patched grids [l], [2],[3], specified by allowingmultiple connections or mul-
overlapping Chimera grids [l], [5] and unstruc- tiple boundary conditions on block sides. Bow-
tured grids [4]. Turbulence models used include ever, to maintain simplicity for the flow solver,
the Baldwin-Lomax model [l], [Z], [3],[4],[5]and blocks with only one connection or boundary con-
the Baldwin-Barth model [l],151. dition on each side have been permitted.
However, many dficulties still remain. Turbu- Grids of types other than H may be generated
lence models which are simple enough for prac- by specifying the block boundaries manually. For
tical computations can be inaccurate for com- example, a stacked C-type grid could be gener-
plex flows. Unsteady flow may occur on high-lift ated by interactively defining the block bound-
configurations especially near stall or post-stall. aries with a CAD program. The advantage of
Such cases require time accurate computations. the automated approach is the ease with which
This paper applies a well-established numeri- new geometries may be gridded. For instance, a
cal method for the Navier-Stokes equations to matrix of gaps and overlaps for a flap optimiiza-
several high-lift configurations. Extensive com- tion study may be quickly gridded with approxi-
parisons with experimental data are made in- mately the same number of points and resolution.
eluding turbulence data, thereby contributing to In contrast, specifying the block boundaries in-
the evaluation of the Baldwin-Earth turbulence teractively k a lengthy process for each geometry
model [SI. and would be onerous for optimization studies.
The disadvantaee " of the automated Drocedure is
that the A-type topology requires more points to
2 Grid Generation resolve the flow than C or 0 type grids. This
increases memory requirements and computing
A computational grid is needed in the flow field time for the solver.
surrounding the high-lift system of airfoils in or- The second step in the grid generation proce-
der to solve the Navier-Stokes equations. Gen- dure is to distribute points on block sides and
erating an appropriate grid around a practi- produce an interior grid. Points may be dis-
cal multi-element geometry can be a difficult tributed by a variety of methods, but usually
task. Problem encountered include blunt trail- hyperbolic tangent distributions are used so the
ing edges, various gaps and overlaps between el-
spacing at each end of a block side may be spec-
ements, detachable slats, and cove regions with
ified. Also, if desired, points may be clustered
sharp corners. part way along a block side. Within each block,
A single structured body-conforming grid can- a starting grid is generated using a simple alge-
not usually be generated around practical con- braic technique, transfinite interpolation.
figurations so either unstructured grids or struc- Finally, the composite grid is smoothed using
tured composite grids are required. By first di-
an elliptic grid generation method [9]. Block
viding the domain into topologically rectangular boundaries are usually held fixed, with fixed
blocks, astructured grid may be generated within points on each side. However, they may also be
each block. This approach is taken here and is allowed to move freely aa part of the solution of
known as multi-block. It has been used by a num- the elliptic system. Or, alternatively, the bound-
ber of researchers for practical problems in two ary points may be allowed to move constrained
and three dimensions [7]. to the original boundary curves.
An automated procedure has been used to di- The final grid is smooth, with point and slope
vide the domain into topologically rectangular continuity of grid lines at block boundaries. Also,
blocks and generate grids [E]. The grid gener- grid lines are usually orthogonal at block edges.
ation is accomplished in three steps. First the The exception is at special points, where slope
domain is divided into blocks by tracing stream- continuity cannot be maintained. An example of
lines and equipotential l i e s determined from the a complete grid for the GA(W)-1 airfoil and flap
solution of the potential flow problem, computed in a wind tunnel is shown in Figure 2.
using- a panel
. method. Curves specifying
. . -block
boundaries, block connections and boundary con-
 9-3

3 Numerical Technique for The matrices 2,E and &?are the flux Jacobians,
defined by
Solvine the Navier-Stokes
2
.

Equations
3.1 Solution Method
The solution method used here for the thin- The diagonil form the
- bf is
layer Navier-Stokes equations is an adaptation
of the diagonal form of the Beam and Warming +
Tf [ I h6cA~] [I h b A,,] ,'A$ R + =2
implicit approximate factorization algorithm in (10)
generalized coordinates, as used in the program where
ARC2D (10). The method is summarized here to =T ; ' ~ T ~ (11)
enable discussion of interface conditions.
A,, = T; ET,, (12)
The thin-laver Navier-Stokes eauations in Ken-
~~
-
eralized curvilinear coordinates are and
?f = T'T,, (13)
8,g + a(@+ a,,P = %-la,$ (l) The matrices Ar and An are diagonal matrices
where whose elements.&e the eigenvalues oEhe flux Ja-
cobians. The viscous flux Jacobian M-cannot be
diagonalized with the flux Jacobian B so it has
4 =J- (2) been dropped from the left hand side of equa-
tion (10). However, a term approximating the
viscous eigenvalues is added to the diagonal, as
A

[ PUU
PU
+
&P
pvu+(yp
+
] described by Pulliam [lo]. With combined sec-
(3) ond and fourth order implicit nonlinear dissipa-
tion, the diagonal algorithm requires the solu-

1
P) - C p

I
tion of scalar pentadiagonal systems of equations
r nV 1 rather than block pentadiagonal systems, which
r.
are required for equation (7).
P= J-1 Puv+vzp
(4)
p v v + vYP
- +
v ( e P) - vtp 3.2 Application of Boundary Con-

[ ]
~

0
ditions
v n l + vYma For implicit centered difference codes with ex-
3 = J-' %ma vym3 + ( 5 ) plicit boundary conditions all variables must be
v=(uml+ vma m4)+ specified at each boundary. As a result, the num-
+ + +
vy(uma vm3 m d ber of numerical boundary conditions required
and may exceed the number of physical boundary
conditions. In a multi-block scheme, if each block
ml = -
~ ( 4 7 1 ~ 7 4 2vyv,,)/3
~
is to be solved independently then additional con-
ditions must also be specified at grid interfaces.
ma = p(vyu,,+v=vq) The boundary conditions used here closely fol-
m3 +
= ~ ( - 2 v . u ~ 4vyvq)/3 (6) low the work of Pulliam [lo]. The exception
m4 = pPr-l(-y - l)-'q&,,(aa) is a viscous outflow boundary for internal flows,
ms = p P r - ' ( y - 1)-'qYt$(aa) specifically the case of an airfoil in a wind tunnel;
then the static pressure is specified. If both the
With implicit Euler time marching, approximate upstream and downstream pressures are known
factorization and centered spatial differences, the from the experiment then they may be speci-
delta form of the algorithm is fied directly. Otherwise the downstream pres-
sure is adjusted such that the inflow pressure is
[I+h6,ji"l[I+ha,,~'-h%-'~,, J - ' ~ " ] A O "= f?" pm. Thus for internal flow at a subsonic viscous
(7) outflow boundary p, pu, and pu are extrapolated
where the right-hand side is from the interior and

R" = -h[6cEn + 6,P" - R-16,,3n] (8)

n
Poutflow
"-1
= Poutnow +PAP"-' (14)
9-4

where @ is a relaxation constant and are obtained for the block boundaries. An inter-
mediate update is then applied so that
AP = ~m - Pinflow (15) - 1
and pinem is defined to be the average of the
AQ?, = 5 (A&
A@z) + (18)
upper and lower Wall Pressures just downstream the new values are then l d e d into both
of the inlet. adjacent blocks for use in the next sweep. Af-
ter the JI sweep is completed, the values on the
3.3 Interface Conditions at Block boundaries are averaned again.
Boundaries
In a structured data base. the solution is stored
1
A@ = (A@ + ~a)
in a series of n x m x 4 arrays with the bound- and this average value is used to update the mlu-
ary between neighboring blocks stored in both tion on both blocks. Although this simple inter-
blocks. The solution technique, however, re- face condition is not nonreflecting, the resulting
quires additional information from the neighbor- multi-block procedure is as robust as the origi-
ing blocks if a solution is to be obtained for the nal algorithm and, most importantly, converges
interface points. This introduces the concept of to exactly the same solutiou obtained on a sin-
halo points. If a block is surrounded by an ex- gle block grid. At the interface, disturbances are
tra layer of points, then prior to the next solution partially reflected but these are most noticeable
step, the halo points may be filled with data from at start-up when the boundary conditions are
the adjacent block. These datamay then be used ramped in. Over several hundred iterations they
as required. appear to be no worse than the reflections due to
Halo points can be used in several ways. The the explicit boundary conditions and the rate of
usual approach is to determine how far away the convergence is not adversely effected.
algorithm needs data and then to copy the a p It is very important that the metria and d k
propriate number of rows or columns of points. sipation be treated in the same manner a c m a
Normally, artificial dissipation is the constrain- block interface. Otherwise, the program may fail
ing quantity and for the form of dissipation used to converge completely or may diverge. Also, if
here, three extra rows would be required. Alter- the intermediate update is not applied then the
natively, just one row of halo points can be used. stability of the scheme is severely degraded.
Rather than computing the dissipation in one
step, intermediate quantities can be computed 3.4 Turbulence Model
and stored in workspace arrays and data passed
to halo points of the workspace arrays. The ad- The choice of turbulence model can have a sig-
vantage of the first approach is that only one copy nificant impact on computing h- and acm-
between blocks is required at the beginning of racy. usually, simple algebraic models such as
each step. In the second approach, which is used the Baldwin-Lomax model have been employed
here, less memory is needed for halo point data [Ill. These models have been used successfully
but additional copies are required. Care must for single element airfoil computations includ-
be taken when copying data between blocks be- ing shock/boundary-layer interactions, small s e p
cause the storage location and sign of the data =ations [I21 and computations near maximum
may change depending on the orientation of the lift 1131.
blocks and the data type. For high-lift computations with multiple ele-
Once the dissipation terms and the right hand ments, simple algebraic eddy viscosity models
side have been determined for each block, an im- are inadequate. Confluent boundary layers, large
plicit time step may be taken and equation (10) separated regions and separated wakes cannot
%ay be solved in two steps. If the product matrix be treated properly. One would expect that
N is split into its facton Ti'T,, then equation improved results could be obtained with one-
(10) may be rewritten as shown equation or two-equation eddy viscosity models.
There are indications, however, that even these
[I + h6<At] T;'A@ =2 (16) models may be unable to predict the interaction
of a wake and a boundary layer [14] or large sep
Tq[ I + h6,,A,] T;'A$ = A@ (17) arations (151 and that algebraic Reynolds stress
the strong interaction of the near wake of a slat unnel wall effects, so the wind tunnel was in-
with a main airfoil boundary layer may result in cluded in the computations. Inviscid boundary
'negative eddy viscosity' [14]. This phenomenon conditions were applied at the wind tunnel walls
cannot be modelled by an eddy viscosity model. to avoid computing the wind tunnel wall hound-
Such effects are quite local, and should not dom- ary layer. The wind tunnel test section extended
inate the flow. As a result, one-equation or tw- two chords upstream of the leading edge and
equation models may be adequate in many situ- three chords downstream, so the computational
ations. boundaries were placed at the same locations.
The turbulence model used in this paper is Approximately 70,000 points were used for the
that of Baldwin and Barth [SI. It is a one- grid and the off-wall spacing was set to 2 x 10-5c
equation model that is fairly easy to implement to ensure that y+ < 1 for the first point off the
and has recently been applied to high-lift prob- surface.
lems by Rogers et al. [l]for the incompressible In initial computations, transition was fixed at
Navier-Stokes equations and by Rentze et al. [5] the locations specified in the experiment. A lam-
for the compressible Navier-Stokes equations on inar separation bubble was predicted on the main
Chimera overset grids. airfoil. Unfortunately, the bubble w a not stable
Prior to implementing the model in'the multi- and was occasionally shed downstream, resulting
block code, it was tested for a single element air- in an unsteady solution. In order to obtain steady
foil at high lift and compared to the Baldwin- results, the transition point was moved forward
Lomax turbulence model. The test case se- to the leading edge.
lected, which was previously examined by An- The computation was performed for the NACA
derson et al. using the Baldwin-Lomax tur- 4412 and NACA 4415 airfoils with both sharp
bulence model [13], is a NACA 0012 airfoil at and blunt trailing edges. For sharp trailing edge
IR = 3.0 x lo6 and M = 0.3. The computed and airfoils, the computation converged very well.
experimental lit coefficients are shown in Fig- With blunt trailing edges, some unsteadiness oc-
ure 3. As may be seen from the plot, the pre- curred in the wake region which prevented com-
diction of CL,,, is lower for the Baldwin-Barth plete convergence. A comparison of the conver-
model than for the Baldwin-Lomax model, which gence history for the two cases is shown in Fig-
is very close to the experimental results. At an- ure 4. Differences between the two solutions were
gles of attack higher than 16 degrees the solution small and the results presented helow are for the
became unsteady for both turbulence models and blunt base airfoils.
steady post-stall solutions could not be obtained. A comparison of the pressure distribution with
experiment is shown in Figure 5. The pressure
distribution on the flap agrees quite well with
4 Results experiment and the flow separates at 87% flap
chord, close to the experimentally observed value
4.1 NACA 4412 Airfoil with of 93%. The computed suction peak on the main
NACA 4415 Flap airfoil is hieher than in the exoeriment and re-
sults in additional lift. Wind tunnel wall pree
The first multi-element case considered is the
sura are shown in the same figure and the agree-
airfoil and flap experiment of Adair and Horne
ment is fairly good except near the upstream and
[16]. The experiment consisted of surface pres-
the downstream boundaries. The discrepancy
sure measurements and flow field measurements
near the boundaries indicates that the location
of velocity and turbulence quantities using both
of the boundaries may have affected the compu-
hot-wire and laser anemometry. The airfoil sec-
tational results.
tion was a NACA 4412 with a NACA 4415 flap.
Comparisons of velocity profiles at five stations
The model was positioned at (I = 8.2', 6, =
are shown in Figure 6. As may be seen in the fig-
21.8', G', = 3.5%c, 01 = 2 . 8 % ~and tested at
ure, the boundary layer is predicted quite well
IR = 1.8 x lo6, M = 0.09. Transition was fixed on the main airfoil at stations 1 and 2. The
at z/c = 0.025 and z/c = 0.010 downstream
flow velocity through the slot, station 4, is un-
from the pressure minimum on the main airfoil
derpredicted but the wake of the main airfoil is
upper and lower surfaces respectively. On the
predicted fairly well. At the flap trailing edge,
upper surface of the flap, transition was fixed at
station 12, the computed flow has a separated
z/c = 0.008downstream of the minimum pres-
boundary layer profile but the region of reversed
sure; on the lower surface transition was free.
flow is tbinner than seen in the experiment. Sim-
The data presented were not corrected for wind
ilar results for profiles and pressure distribution
were obtained by Fbgers et al. [l]wing an in- Table 1: Summary of GA(W)-1 airfoit and flap
wnfigurations.
comoressible Navier-Stokes solution metbod with
the Baldwin-Barth turbulence model.

4.2 GA(W)-1 Airfoil with 28%

chord Fowler Flap
 0
9-1

fairly insensitive to cove shape [18]. case resulted in 31 blocks. A total of 90,000
For the wide gap cases, there is a large sepa- points were used for the grid and the off-wall
ration on the flap. For case B1, o = 4 O , the flapspacing grids was set to 2 x 10-'c.
separation is underpredicted, as shown in Fig- A comparison of the computed pressure distri-
ure 11. Because the flap separation is underpre- bution and experiment is shown in Figure 15. As
dieted, the flap carries more load than it should may be seen in the fipure, the pressure distribu-
which in turn increases the loadiig of the main tion agrees very well for the main airfoil and the
airfoil. Profiles of the velocity in the confluent flap. For the slat, however, the suction on the
boundary layer over the flap are shown in Fig- upper surface is overpredicted. This result is ob-
ure 12. The main airfoil wake is initially c a p tained because the computed separation on the
tured well but becomes progressively worse. The lower surface of the slat is too small. As may
flap separation may be seen clearly at station be seen from the experimental pressure distribu-
(+/c)t = 0.27. Although the computed profile tion, stagnation pressure was recorded at 13.5%~.
is separated, the extent of reversed flow is much In the computations, however, the reattachment
too small. This result is similar to that seen for point is further forward, at 1 0 % ~Therefore,
. ef-
the cove separation on the main airfoil. However, fectively, the slat has more camber which results
high-lit performance is sensitive to flap separa- in higher loading than seen in the experiment.
tion whereas the cove separation is not as signif- The details of the computed flow may be Been in
icant. Figure 16 which clearly shows the large separa-
A comparison of the lift coefficients for the tion on the lower surface of the slat and the point
computed cases with the experimental data from of reattachment. The flow is attached at the u p
Braden et al. is shown in Figure 13. The effect per surface trailing edge of both the main airfoil
of gap is underpredicted because the flap separa- and flap, so the resulting pressure distributions
tion is underpredicted. Also included in the plot for these elements agree fairly well with experi-
are corrected experimental lift coefficients from ment. The separation in the cove region of the
Wents and Seetharam, which were obtained at main airfoil is underpredicted as before.
W = 2.2 x lo6 [19]. The computed l i t coeffi- Comparisons of velocity and Reynolds stress
cients are close to the Lift curve from Wenta and profiles for the upper surface of the main air-
Seetharam because in the higher Reynolds num- foil are shown in F i y e 17. The effect of the
ber experiment there is less extensive separated merging slat wake on the boundary layer pr-
flow. file is clearly e e n in the computations but is not
well resolved in the experiment. However, it a p
4.3 GA(W)-1 Airfoil with Slat and pears that the wake develops and merges more
quickly in the computations than is seen in the
Flap experiment. This is consistent with the results of
The three-element configuration considered here Rentre et al. [5] for a threeelement configuration
is a GA(W)-1 main airfoil with a 1 5 % ~slat at with the same turbulence model.
6, = 42', G, = 1.5%c, 0,= 2.5%c, and a 29%
Fowler flap at 6 j = 40°, Gj = 1.5%c, 01 =
1.5%. The entire configuration is at (I = 5.3' 5 Conclusions
and corresponds to case F-1 of Braden et al. [17].
A method for solving the compressible thin-layer
The flow field about a three element mnfig-
Navier-Stokes equations around high-lift config-
uration is considerably more complicated than
urations on structured composite grids has heeu
for a flap alone. The leading edge slat depresses
presented. The method is based on the approxi-
the suction peak on the main airfoil which tends
mate factorization technique used in the program
to delay stall. However, the slat wake interacts
ARC2D. Interfaces between blocks are treated
with the boundary layer on the upper surface of
using a simple overlapping procedure with an in-
the main airfoil which promotes early boundary
termediate update. Turbulence is modelled ns-
layer separation. The merging wakes then inter-
ing the Baldwin-Barth oneequation turbulence
act with the slot flow over the flap. The flow is
model. Grids were generated with the aid of an
further complicated by the separation in the cove
automated procedure which divides the domain
region of the main airfoil and on the underside of
into blocks.
the slat. A diagram of the three element case is
Good agreement with experiment is obtained
shown in Figure 14.
when the flow about the configuration is attached
The block decompoeition for the three element
or regions of separated flow are small. Compu-
9-8

tational difficulties encountered include laminar 191 Thompson, J. F., Warsi, Z. U. A. and
separation bubbles, multiple regions of separated Mastin, C. W., Numericol Grid Gencro-
flow and slightly unsteady flow in wake regions. Lion, Foundations ond Applicotions. North-
Results tend to he poor for cwes with large re- Holland, 1985.
gions of separated flow in critical areas such as
the flap upper surface or on the slat. Generally, [lo] Pulliiam, T. H., “Efficient Bolution meth-
the extent and thickness of rev,,& flow regions ods for the Navier-Stokes equations,” Lee-
are underpredicted. The Baldwin-Barth turbu- ture Notes For The Von Karman Insti-
lence model works well in most instances but tute For Institute for Fluid Dynamics Lec-
tends to underpredict eddy vis,.osity in w&e re- ture Series: Numerical Techniques For Vis-
pions and overpredict eddy vismity in the outer cous In Turhomaehinery
portion of boundary layers. Bladiings, January 1986.
[ll] Baldwin, B. and Lomax, E.,Thin-layer sp
References proximation and algebraic model for sepa-
rated turbulent flows,”AIAA Paper 76-257,
[l]Rogers, S. E., Wiltberger, N. L. and Kwak, January 1978.
D., “Efficient simulation of incompressible [12] Maksymiuk, C. M., and Pulliam, T. H.,
viscous flow over single and multi-element “Vimus trahonic airfoil workshop results
airfoils,” AIAA Paper 92-0405, 1992. using ARCZD,” A I M Paper 87-0415, Jan-
uary 1987.
.121_Chow, R. and Chu. K., “Navier-Stokes solu-
tion for high-lift multielement airfoil system 113) Anderson, W.K., Thomas, J . L., and Rum-
6 t h flap =paratio%” AIAA Paper 91-1623- sey, C. L,,~ ~ ofthin-layer
~ ~ ~l i~ ~~ ~i
CP, June 1991. Stokes eauations near maximum lift.” AIAA
[3] Shima, E., “Numerical analysis of multiple Paper 8 b o 4 9 1January 1984.
element high l i t devices by Navier-Stokes [14] Squire, L. C., “Interactions between wakes
equation wing implicit TVD finite volume and boundary-layers;” pmg. A~~~~~~~ sei.,
method,” AIAA Paper 8&2574CP, 1988. vol. 26, pp. 261-288,1989.
[4] Mavriplii, D. J., “Euler and Navier-Stokes [15] Davidson, L. and Rizsi, A., “Navier-Stokes
computations for airfoil geometries using un- mmputation of airfoil in stall using alg+
structured grids,” CASI Journal, vol. 36, braic Reynolds-stress model,” AIAA Paper
pp. 62-71, June 1990. 92-0195, 1992.

[6] Baldwin, B. S.,and Barth, T. J., “A one- [17] Braden, J. A., Whipkey R. R., Jones,G. S.,
equation turbulence transport model for and D, E., “Experimental study ofthe
high Reynolds number wall-bounded flows,” separating confluent boundary-layer,” Vol-
NASA TM 102847, August 1990. nme I - NASA CR 3655. Volume 11- NASA
[q Weatherill, N. P., “Mesh generation in CR 166018, 1983.
CFD,” Computational Fluid Dynamics, von [l8] Wentz, Jr., W. H. and Ostowari, C., “Ad-
Karman Institute for Fluid Dynamics Lee- ditional flow field studies of the GA(W)-l
ture Series, 198404. airfoil with 30-percent chord Fowler flap in-
[SI Nelson, T. E., Zingg, D. W., Johnston, G. eluding slotgap variations and mve shape
w.,‘‘Automated @id generationfor high- modifications,” NASA CR-3687, 1983.
lift configurations,” Canadian Aeronautice 1191 Wents, Jr., W. and Seethararn, H.C., “DE
and Space Institute: Proceedings of the velopmnt of a ~~~l~~flap systemfor a high
Third Canadian Symposi= on Aerodynam- performanee general aviation airfoil,” NASA
ies, November 1991. CR2443, December 1974.
 a 9-9

TOP WIND TUNNEL WALL

2
> d
g
U

3
3
iE
2
3 X

BOTTOM WIND TUNNEL W A U

Figure 1: Automated block decomposition and the resulting block connections for the GA(W)-1
airfoil and flap in a wind tunnel.

Figure 2: Near field grid for the GA(W)-1 airfoil and 29% chord Fowler flap in a wind tunnel,
(plotted with every second grid line in each direction).

la

1.4

12 .
1.0 .
c, 0 8 ;.

OD . ’ ’
6 10 16
Anple d Atla&

Figure 3: Computation of CL,,, for NACA 0012 airfoil using the Baldwin-Barth and Baldwin-Lomax
turbulence models.
9-10

1O"O -

Iteration

Figure 4: Convergence history for Adair and Horne test case of NACA 4412 airfoil with NACA 4415
flap. Results with sharp and blunt trailing edges.

: :Ki
1.00 1.10 1.20 1.30 20 -1.0 0.0 1.0 2.0 3.0
wc YJC wc

Figure 5: Surface pressures and wind tunnel wall pressures for Adair and Horne airfoil and flap test
case.
-

o"i
Siation 1, XK: = 0.4% S l P h 2. XIC 0.756 Station 4, WC I 1.WO Slalion12.WC=1.308 a a h 16.xIc=1.364
0.07 0.12 0.17 0.w

Ydc
0.07
0.1
0.07 0

Ydc Ydc
0.0 0.30
0.02 0.02
. . 0.050
0.020.0 0.5
0.000 1.0 1.5 2.0 0.40 r:'

0.
0.02

0.000
..-
0.60
0 . 5 0.0 0.5 1.0 1.5 0.0020.40.80.81.012
UN, UN. w. UN,

Figure 6: Comparison of velocity profiles for Adair and Horne airfoil and flap test case.
 9-11

Figure 7: Comparison of experimental and computed pressure distributions for GA(W)-1 airfoil 29%
chord Fowler flap case B4.
Stallon 62. WC = 0.60 Slation 64, WC I0.80 station 62. xx:= 0.60 SlaliGn &(.wc 1 0.80
0. 0. 0.

1-
0.

ydc
0.03 0.02
~~ 0.0

0.02 0.0

0.01 0.01 0.01

~ : ~
:
0. 0.0
~~ 0.6UN,1.0 1.5 0.
0 *:0.6 0.0
O. -1.0 1WWVN',) 0.5
' ~
0.0 0.6 1.0 1.5 -1.0 0.6 0.0 0.5
UN, lOO'(WN',)

:1!
Figure 8: Boundary layer and Reynolds stress profiles on the upper surface of the main airfoil for
the GA(W)-1 airfoil and flap case B4.
slmionm.(xx:),=o.o3 staiion7~~
(m),=0.27 stallon 68, (WC), = 0.03 Slalim 70. (XI=,
0.27
0.1 50 O.1M 0.1

0.125

YJC
0.07

o m 0.075

ydc
0.060
0.050 0.050

0.025 0.026 0.02

oaw 0.
0.0 0.6 1.0 1.5 -1.0 4.5 0.0 0.6 1.0
UN, UN, lOV(UVN',)

Figure 9: Boundary layer and Reynolds stress profiles on the upper surface of the flap for the
GA(W)-1 airfoil and flap case B4.
9-12

4.b_;D4.0O2n
Station278. WC = 0.66
0.00

YJC
4.

4.0

4.04

4.10
0.00

4.04

4.

4.10
Station 279. wc = 0.71
0.w

4.02

4.04

Ydc

4.04.5
4.1
0.08
Station 282. WC = 0.76
.
\

0.0 0.6 1.0

4
0.w

.04
YdC

4.m

4.08

4.10
Station 284. XX: = 0.88 Station 285, wc = 0 . a
0.00

4.0.

YdC
O.W

4.W
4.10
4 .6 0.0 0.6

4.6 0.0 0.5 1.0 4.6 0.0 0.6 1.0 4.6 0.0 0.5 1.0
'"U '"n WUn UN, '"n
Figure 10: Boundary layer profiles in the cove region of the main airfoil for GA(W)-1 airfoil and
flap case B2.

6.0

6.0

4.0

3.0
cp
2.0

1.o
MAIN AIRFOIL
00

-10
0.00 0.26 0.50 0.76 1.w 0.00 0.10 020 0.30

0.071
(wc). (WCI,

Figure 11: Comparison of experimental and computed pressure distribut,ions for GA(W)-I airfoil
29% chord Fowler flap case B1.
statim55. (WC),=o.as StalionS6. (WC),=O.O3 Slaticn57.(xIc),=O.16 Statlon58. (XC)F-02?
0.100 0.1 0.100 0250

0.076

0.050
0.02
YJC 0.050 0.050
YJC 0.1w

0.02 0.025
0.0

0.000 0.000 0.wo 0.000

0.0 0.6 1.0 1.6 0.0 0.6 1.0 1.6 0.0 0.5 1.0 1.6 4.6 0.0 0.5 1.0 1.6
UN, UN, "'Jn "4
Figure 12: Boundary layer profiles on the upper surfaee of the flap GA(W)-1 airfoil and flap case
B1.
 9-13
1

1.o
0 2 4 6 8 10 12 14 16
a
Figure 13: Lift curves for GA(W)-1 airfoil aiid flap teat cases.

,Confluenlbundary Layers ,

Main Airfoil
SeparpliMl

Figure 14: Flow field interactions around an airfoil with slat and flap.

MAIN AIRFOIL

0.00 0.05 0.10 0.15 0.03 0.25 0.50 0.75 1.00 0.00 0.10 030 0.M
W)* (W), (W).

Figure 15: Comparison of experimental and computed surface pressures for GA(W)-1 airfoil with
slat and flap.
 Figure 16: Streamlines on the lower surface of the slat for the GA(W)-1 airfoil with slat and flap.

0.0

0.0

0.02

0.01

"."8.0 0.5 1.0 1.6 2.0 2.5 0.0 0.5 1.0 1.5 2.0 2.p
0.

0.

0.01

0.00
-1.0 4.5 0.0 0.5
i(i
OS%

0.02
0.03

0.01

0.W
-1.0 4.5 0.0 0.5
UN, UN, lOO'(@GN'.) ~WJ*(IXN'~)

Figure 17: Velocity and Reynolds.stress profiles on the main airfoil Cor GA(W)-1 airfoil with slat
and flap.
 High Reynolds Nninber Configuration Development of a 1ligh-Lift Airfoil

Walter 0. Valniezo and Chet .I.Dnminik

Acrndynamics Research and Technology
Douglas Aircraft Company
38.55 Lakewood Blvd.. Long Beach, CA 90846, IISA

and
Robert J. McGhee and Wesley L. Goodman
Fluid Mechanics Division
NASA Langley Research Center
Hampton, VA 23665.5225, USA

SLJMMARY 1.0 INTRODUCTION

A n experimental program has been conducted to asscss Increased performance requirements for high-lift systems
(!le performance of a transport multielement airfoil a t in the commercial transport sector have caused renewed
llight Reynolds numbers. The studies were performed a t emphasis in multielement airfoil aerodynamics. The
chord Reynolds numbers as high as 16 million in the focus of the more recent research has been on under-
NASA Langley Low Turbulence Pressure Tunnel. standing Reynolds and Mach number effects on high-lift
Sidewall boundary-layer control to enforce flow two- optimizationl-5 with the ultimate aim of improving max-
dimensionality was provided via an endplate suction imum lift capability as well as lift-to-drag ratios.
system. The basic airfoil was an 11.55% thick supercrit-
ical airfoil representative of the stall critical station of a
new-generation transport aircraft wing. The multiele- The majority of the reported investigations have dis-
cussed performance effects due to Reynolds and Mach
ment airfoil was configured as a three-element airfoil with number variations on generic high-lift configurations.
slat and flap chord ratios of 14.48% and 30% respec-
tively. Testing focused on the development of landing Additionally, these studies have concentrated on
configurations with high maximum lift capability and the reporting the results of rigging optimizations
assessment of Reynolds and Mach number effects. Also (gap/overhang) and their effect on airfoil performance.
assessed were high-lift performance effects due to devices These investigations have greatly added to the existing
such as drooped spoilers and trailing-edge wedges. The high-lift database with relevant results at flight and near
present experimental studies revealed significant effects flight conditions.
on high-lift airfoil performance due l o Reynolds and
Mach number variations and favorable lift increments at This document reports on experiments conducted at high
approach angles of attack due to the use of drooped Reynolds numbers to develop performance concepts for
spoilers or trailing-edge wedges. However, no substantial high-lift transport aircraft applications. The concepts
improvements in maximum lift capability were identified. trentcd in this investigation are modifications to the
A recently developed high performance single-segment trailing edge of the multielement airfoil and include
flap was also tested and results indicated considerable
segment flap design.
-~
deflected smilers. trililina edae wedges. and a new Finde-
improvements in lift and drag performance over existing
airfoils. Additionally, it was found that this new flap
shape a t its optimum rigging was less sensitive to The work documented in this paper is the result of n
Reynolds number variations than previous designs. cooperative experimental program conducted hy the
Douglas Aircraft Company and the NASA Langley
NOMENCLATURE Research Center. The experiments were conducted in the
NASA Langley Low Turbulence Pressure Tunnel (LTPT)
C airfoil chord (slat and flap stowed) in Hampton, Virginia. The aim is to augment thc existing
database of Reynolds and Mach number effects on high-
Ce lift coefficient lift performance and new concepts.

Q",, maximum lift coefficient 2.0 MODEL GEOMETRIES

CP pressure coefficient The model tested utilized an I I.SS"/o thick supcrcritical

airfoil representative of the estimated stall-critical
M Mach number location on an advanced design subsonic transport. The
model spanned the width of the test section and had a
OH overhang clean (stowed) airfoil chord of 22 inches. The single-
element and the multielement airfoils tested arc shown in
RN Reynolds number Figure I . The slat chord ratio was 14.48% and the
single-segment flap chord ratio was 30% of the stowed
U angle of attack (degrees) airfoil chord. Pressure orifices were located along {.he
centerline of the model (146 taps for the three-element
USTALL stall angle (degrees) configuration) and along or near the trailing edge of each
component. Surface pressures were measured via nine
6, flap deflection (degrees) Electronic Sensing Pressure (ESP) modules calibrated to
IS, 30, 45, and 100 psig as required by model aero-
6s slat deflection (degrees) dynamic loading. Accuracy for the ESP's was 0.1% *
of full scale value. Integration of pressure measurements
6SP spoiler deflection (degrees) yielded the forces presented in this puhiication. Four
10-2

rows of streamlined support brackets for the high-lift element and angles of IS", 30", 45", and 60" were tested
devices (Figure 2) were required due to the very high on the three-element configuration. The wedges were
loads (approximately 15,000 pounds) developed at some glued to the airfoil lower surface at the trailing edge. All
of the maximum lift conditions. The nomenclature wedges were sealed with Room Temperature Vulcanizer
defining the key geometric parameters of high-lift systems (RTV) to prevent flow through the wedge/airfoil juncture.
is shown in Figure 3. A sketch of a 30" trailing edge wedge applied to the
single- and three-element airfoils is shown in Figure 4.

Figure 1. Airfoil Gcometries Tested in the LTPT

3.0 LTPT TEST FACILITY

The Langley Low Turbulence Pressure Tunnel is a single
return, closed-throat wind tunnel (Figure 5 ) that can be
operated up to IO atmospheres thus allowing very high
Reynolds number capability. The test section is 3 feet
wide by 7.5 feet high by 7.5 feet long.

Figure 2. View of High-Lift Model Support Brackrts

(-) Shown Eunm

Figure 5 . Schematic of Low Turbulence Pressure Tunnel

To promote two-dimensional flow over the model in view

of its low aspect ratio and strong wall interference, a new
side-wall boundary-layer control (BLC) system was
installed in the LTPT'. The BLC system employed the
differential pressure between the test section and the
atmosphere to provide suction of the sidewall boundary-
Line layer through porous endplates (Figures 6 and 7). The
system yielded good quality two-dimensional flow over
the model for the Reynolds numbers tested as indicated
by the flap trailing edge spanwise pressures shown in
Figure 8. The criterion used to define "two.
dimensionality" was the observation (at run time) of suf-
ficiently flat trailing edge pressures in the spanwise
direction (*0.5% of leading edge peak). Examination of
Figure 8 reveals that there are two peaks in the spanwise
pressures at 15% and 85% span for the case with
Figure 3. Nomenclature for Multielements sidewall suction. These spikes were attributed to the
presence of the outboard brackets (10.7% and 89.3%
The model could also be configured with positively span) and were found to be relatively insensitive to
deflected spoiler surfaces with angles of 5" and IO" as sidewall suction rate. Hence, these peaks Were not con-
shown in Figure 1. The spoiler geometry was removable sidered when assessing the two-dimensionality of the
and each spoiler deflection consisted of a steel model part flow. Representative trailing edge pressures with these
at the appropriate deflection. two spanwise stations removed are shown in Figure 8.
While this procedure may appear somewhat arbitrary, it
In addition to the above-mentioned configurations, was found that removing the two pressure spikes caused
trailing edge wedges which were 3% of chord in length by the brackefdid improve spanwise pressure dis-
were also tested on both the single- and three-elcment tributions that were otherwise not flat (see also Figure 8.).
airfoils. A wedge angle of 30" was tested on the single-
 10-3

program was to measure the performance effects due to

deflected spoiler (5". IO') applications on a three-element
landing configuration. The flap gap and overhang were
held constant to isolate the effects due to spoiler
deflection.
It was found that for this airfoil application, drooping the
spoiler (5' or IO') actually decreased the maximum lift
capability and stall angle (Figures 9 and IO). However,
over most of the lift curve, the lift coeflicient at a given
angle-of-attack was considerably increased by the
drooped spoiler (Figure I I). The extra lift was generated
largely by the main element, as shown in Figure I I , and
is a consequence of its increased aft camber. Besides

-
generating more lift, drooping the spoiler changed the
character of the stall. leadinn to a sharper stall.

Figure 6. Boundary Layer Venting System

Figure 9. Effect of Spoiler Deflection on Maximum Lift

of a Three-Element Airfoil,
6s = 30" & = 30"

Figure 7. Porous Endplates of Boundary-Layer Control

System
46
Q1
.o.a
Q.2
4 1
'' M
a1
U
01
Figure IO. Effect of Spoiler Deflection on Stall Angle
01
of a Three-Element Airfoil,
OS
o 1 o a o a o a m m m m m i m as = 30" SF = 30"
Nondlmenslonal Model Span (X)
Figure 8. Comparison of Trailing Edge Spanwise
Pressures With and Without Sidewall Suction, 4.6
a = 16" 4.0
3.6
4.0 RESULTS AND DISCUSSION 90

The following discussion reviews results obtained for

' 26
drooped spoiler effects, trailing edge wedge applications, 20
and the performance of a new generation single-segment 16 F
flap design. The test results were obtained transition free
a t 0.20 Mach number and a chord Reynolds number of 10
9 x 106, unless otherwise specified. This condition is r e p 0.6
resentative of full-scale at the estimated critical wing sta-
tion for low-speed stall on an advanced medium range "." -
nn
transport aircraft. o z 4 e ~ m t z n m m m ~ ~
a (DEGREES)
4.1 DROOPED SPOILER EFFECTS

Positively deflected spoiler surfaces (trailing edge down) Figure 11. Effect of Spoiler Deflection on Lift of a
have been used successfully on military aircraft to Three-Element Airfoil, 6s = 30' 6 p = 30'
enhance lift generation. An objective of the present
Two key observations can be made from these results.
First, the increased loading on the main element leads to
reduced loading on the flap (downwach) and increased
loading on the slat (upwash). Second, increased loading -e-= h r = 5-
on the slat leads to the slat stalling won after the main
element stalls. This effect is not present for the unde-
flected spoiler case, where the slat does not stall at all.
To further illustrate these observations surface pressures
at 8" angle of attack (approach) are shown in Figures 12, vur
13. and 14. Figure 13 clearly indicates increased circu- -2
lation on the slat with increasing spoiler deflection. The
authors attribute upwash from the main element as the -1
cause for this effect. Conversely, the flap surface pres- 0
sures of Figure 14 indicate reduced flap circulation with
spoiler deflection. Again, the mechanism for this efTecl 1
is believed to be downwash from the main element. In
general, application of drooped spoilers to the present msawmmnoll6!20rz6lm~uo
transport airfoil indicated maximum lift decrements h u t wc (%I
substantial increases in lift at a given angle-of-attack.
Figure 14. Effect of Spoiler Deflection on Trailing-Edge
Surface Pressure Distributions of a Three-
Element Airfoil, & = 30" & = 30"

Summary results of Reynolds number effects a t max-

j!
-6
imum lift for the three conf8urations with spoiler
deflections of O", Y, and IO" are shown in Figure IS. The
overall trend with increasing Reynolds number is similar
(increased maximum lift) to that obtained with the
MAIN E L t W m N,
-3 baseline configuration.
-2
4d
-1
M=O.ZO
0
1

Figure 12. Effect of Spoiler Deflection on Surface

Pressure Distributions of a Three-Element
6r = 30' RN (x 10 *)
Airfoil, = 30'

Figure 15. Effect of Reynolds Number on Maximum

Lift, as = 30' dF = 30'

--
4.2 TRAILING EDGE WEDGE EFFECTS

To simulate the efTect9 of a small (3% chord) split flap

on low speed performance, wedges were attached to both
the single- and three-element airfoil. A wedge with an
angle of 30" was tested on the single-element airfoil and
IS", 30",45" and 60" wedges were applied to the flap of
the three-element airfoil as described in Section 2.
The intent of this study was to identify the fraction of the
maximum lift increment realized on the single-element
airfoil (Figure 16) that is retained when the high-lift
components are deployed. It can be seen from Figure 17
that the wedge provides a lift increment throughout the
0 6 l O S ~ a S M 3 6 4 0 .
lift curve that diminishes with angle of attack. The max-
m u m lift coefficient for this three-element landing con-
we (%I figuration is only marginally increased (+0.02) by the use
of the wedge. It is worth noting that most of the increase
Figure 13. Effect of spoiler Deflection on Leading-Edge in lift 1s generated by the main element a d not the flap
Surfacepressure Distributions of a Three- as might be expected. The mechanism for the increase
Element Airfoil, as = 30" bp = 30" in lift is the augmented loading on the much larger main
element caused by upwash from the flap.
 10-5

A lift curve comparison of the baseline and the 45" wedge

results is shown in Figure 19 for a Reynolds number of 9
x 106. As observed with the 30" wedge application, the
majority of the lift increment is due to the main element
of the airfoil, with an increment in maximum lift of
f0.07.
6.0
TOTAL

"Y
ao
02

o z 4 e 8 x )
U (DEGREES)
n w w
P
4.0
36
3.0
26
20

Figure 16. Effect of a 30' Wedge on the Lift of a Single- 10

Element Airfoil
0.6
6.0 0.0
4.6
0 2 4 t 1 8 w n u m ~ m n 1
U (DEGREES)
4.0
3.6 Figure 19. Effect of a 45" Wedge on the Lift of a Three-
3.0 Element Airfoil, SI = 30" 6, = 30"
c,26
20 4.3 ADVANCED SINGLE-SEGMENT FLAP

An advanced flap concept developed a t Douglas Aircraft

to was also tested and is shown in Figure 20. The new
shape is significantly dirfcrent than the baseline flap. The
0.6
advanced flap was designed to reduce flow separation (on
0.0 the flap surface) for landing flap deflections. A compar-
0 2 4 e 8 m u Y 1 6 1 8 2 0 2 2 2 4 ison of flap surface pressures measured a t an approach
U (DEGREES) condition (8, = 30",a = 8") is shown in Figure 21 and
demonstrates the improvement over the baseline flap. It
Figure 17. Effect of a 30" Wedge on the Lift of a Three- can also be seen that the advanced flap was able to SUS-
Element Airfoil, BS 30" 6, = 30'
~

tain an increased suction peak while maintaining

attached flow. This increased suction p a k was present
Results ohtained with 15", 30",45". and 60' wedges on up to and including the maximum lift condition (Figure
the three-element airfoil for a range of Reynolds numbers 22), where both flaps display attached flow. This
are shown in Figure 18. Increases in maximum lift increased suction peak increases the loading on the flap,
capability (from the baseline) are more apparent for the especially a t the lower angles of attack. The performance
45" wedge application. Additionally, the maximum lift improvement due to the advanced flap is shown in Figure
levels out a t wedge angles of 45' and greater (for a 23. The total lift of the three-element airfoil increases
Reynolds number of 9 x 10s). The results obtained a t 5 with the advanced flap. This includes a maximum lift
x IOLindicated a significant increase in maximum lift increase of +0.07. Notably, drag levels are reduced,
between the baseline (no wedge) and the 60" wedge especially around the approach condition (Cc = 3.1).
application. However, at the higher Reynolds number (9 This decrease in drag a t approach is shown in Figure 24
x 10s) this increment all but disappears. Thus, testing at to be 29% over the baseline flap.
a Reynolds number of 5 million produces trends which
do not occur n t the higher Reynolds numbers.

4.1
42 1
Figure 20. Advanced Flap Shape
Figure IR. Effect Reynolds Numbcr on Maximum Lift,
6s = 30" 6,; = 30'
10-6

4.0 r 01 = 8" 0.05

6, = 30'
-3.6
-3.0
1- @ '2VS,,

-25 0.04

-20
cp -1.5
0.03
-1.0
-0.5 Cd
(TOTAL)
0.0 0.02
0.5
in .
m m BO 8 5 8 0 ss XK)
0.01
wc (%)
Figure 21. Flap S u r f x e Pressures a t an Approach
Cnndition, = 30" d r = 30" 0.00
BASELINE ADVANCD
FLAP FLAP
4.0
a = 21" Figure 24. Total Airfoil Drag Reduction Due to
-3.5 Advanced Flap Shape, = 300
-3.0
-25 It is interesting to note the Reynolds numher trends of
-20 Figure 25 since they indicate decreased sensitivity for the
advanced flap. It is expected that these results are a
CP -16 consequence of the absence of flow separation on the
-1.0 advanccd flap surface.
-0.5
M = 020
I
0.0 4.5 I
0.5
1.0
13

1.2
1 / \ 1
Fieure 22. Flap Surface Pressures at Maximum Lift, 1.1
Ks = 30" 6 , = 30"
4.0
0 1 10 16 ?o
RN (x lo1)

1.0 Figure 25. Effect of Reynolds Number on Maximum

Lift, 6, = 30" dr = 30"

3.0
The new flap shape exhibit5 different optimization trends
c,2.1 than that of the baseline flap'. Because of large-scale
10 separation on the baseline flap, the best locations for
1, maximum lift occurred with positive overhangs (overlap-
LO ping) and large gaps (1.5 to 2.0%). The optimization for
0.3
the advanced flap are shown in Figures 26 through 29.
0.0
Unlike the haseline flap, the advanced flap optimizes a t
o,m o.m ~.~~ ~ . o.09
~ =o . ~.(oI
~ negative (or zero) overhangs with gaps between I and
C. 1.5"/0. The dashed lines in Figures 26 through 29 are
estimated lines of constant Ce.,,. For the four
deflections tested, the maximum lift decreases as the
flap/spoiler overlap increases (more positive overhang).
Figure 23. Comparison of the Lift and Drag for the The flow about the advanced flap is better behavcd than
Bascline and Advanced Flaps, that for the baseline flap, thus allowing the flap to take
6, = 30- ar = 30" advantage of additional Fowler motion.
 10-7

5.0 CONCLUSIONS

-
An experimental program to assess the performance of a
Od I transport multielement airfoil at flight Reynolds numbers
has been conducted by Douglas Aircraft Company and
the NASA Langley Research Center. The goals of this
effort were to establish new high-lift technologies as well
as to augment the high-Reynolds number database on
multielemenl airfoil configurations. Several conclusions
are drawn from this study:

ld 1 ia i 0s 0 4.5 .l
'
I
.1.5
1. The drooped spoiler (5" and IO') decreased the max-
Rap Overhang (X C) lmum lift capability of the three-element airfoil
landing confipuration. Additionallv the droooed
~~~~~ ~ ~~~~~~~~~~ ~~~~~ ~ ~~~~~~~ ~~

Figure 26. Maximum Lift Coefficient Contours for an spoiler decreased the stall angle from that of (he
Advanced Flap Shape, dp = 30" baselioe (undeflected spoiler),
2. Drooping the spoiler did increase the lift of the airfoil
RN = e II l o ' 8. = 30 dag 8. = 35 dag at pre-stall angles of attack. This additional lift was
U = 03 generated by the main element and is a consequence
0.0
I of its increased aft camber.

3. The Reynolds number effect on the drooped spoiler

/ / I configuration were similar to those obtained with the
undeflected spoiler. Maximum lift increased with
increasing Reynolds number.

.-
9.0
II
ld 1 id j Od
' '
0 Od
a .l .IS
I
4. Trailing edge wedge applications provided a signif-
icant increase in Ce and Ce,, on the single-element
airfoil.
Rap Omhang (X C)
5. For the 30" wedge application, the maximum lift
increment measured for the single-element airfoil
Figure 27. Maximum Lift Coefficient Contours for an (0.17) was not retained for the three-element airfoil
Advanced Flap Shape, ar = 35" configuration (0.02). The maximum lift increment for
the 45" wedge case was +0.07. Most of the lift
RN=exloa Y=OZ S.=30dag Sr=37.5dag increment was due to the main element. and the
flap.

6. Reynolds number effects on the three-element trailing

edge wedge application were significant. The lower
( 5 x lob) Reynolds number data indicated increasing
maximum lift with increasing wedge angle (up to
60"). However, the 9 x IO6 Reynolds number data
exhibited a much smaller increase than the 5 x I O 6
data. Additionally, the trends with wcdge angle were
not reproduced betwccn the two Reynolds numbers.
W
W 2 1.S I Od 0 Od .) -Id
Rap O m l u n g (X C) 7. A new advanced single-segment flap shape was
tested. The new shape led to significantly reduced
flow separation for the single-segment landing flap
case. Total airfoil profile drag at approach was
Figure 28. Maximum Lift Coefflcient Contours for an reduced by 29% over the baseline design.
Advanced Flap Shape, = 37.5"
8. The advanced flap was not as Reynolds number sen-
sitive as the baseline flap at the higher Reynolds
numbers tested.
9. The optimum rigging of the advanced flap shape was
substantially different than that of the baseline
design. The advanced flap optimized at smaller gaps
and more negative overhangs for increased Fowler
motion.

6.0 REFERENCES

LO I. Morgan, Harry L. Jr., Ferris. James C., and McGbee,

9.6 1 1.8 1 Od 0 0.6 .l .1.6 Robert J., "A Study of High-Lift Airfoils a t High
Flap Overhang (X C) Reynolds Numbers in the Langley Low-Turbulcnce-
Pressure-Tunnel," NASA TM-89125, July 1987.

Figure 29. Maximum Lift Coefficient Contours for an 2. Wedderspoon, J.R., "The High-Lift Development of
Advanced Flap Shape, d r = 40" the A320 Aircralt," ICAS Proceedings 1986.

3. Fidders, S.P., Kirby D.A., Woodward D.S., and

Peckham, D.H.,"Investigations into the Effects of
 Scale and Compressibility on Lift and Drag in the
RAE 5m Pressurized Low-Speed Wind Tunnel,"
Aeronautical Journal, Paper No. 1302, March 1985.
Valarezo, W.O., Dominik, C.J., McGhee, R.J.,
Goodman, W.L., and Paschal, K.B., "Multielement
Airfoil Optimization For Maximum Lift at High
Reynolds Numbers," AIAA Paper 91-3332.
Valarem, W.O., Dominik C.J., and McGhee, R.J.,
"Reynolds and Mach Number Effects on Multiele-
men1 Airfoils," Fifth Symposium on Numerical and
Physical Aspects of Aerodynamic Flows," California
State University, Long Beach, January 1992.
6. McGhee, R.J., Beasley, W.D., and Foster. J.M.,
"Recent Modifications and Calibration of the Langley
Low-Turbulence-Pressure-Tunnel," NASA TP-232R,
1984.

I. Paschal, K.,Goodman, W., McGhee, R., Walker, B.,

and Wilcox, P., "Evaluation of Tunnel Sidewall
Bovndary-Layer-Control Systems for High-Lift
Airfoil Testing," AIAA Paper 91-3243.
 11-1

UNE METEODE D'INTERACTION VISQUEUX NON-VISQUEUX

POUR ECOULEMENTS INCOMF'RESSIBLES HYPERSUSTENTES
SUR PROFILS MULTI-CORPS
ENREGIMEDEDECOLLEMENTPROFOND
-----
( A VISCOUS-INVISCIDSOLVER
FOR HIGH-LIFT INCOMPRESSIBLE FLOWS
OVER MULTI-ELEMENTAIRFOILS
AT DEEP SEPARATION CONDITIONS )
_____
J.C.Le Balleur et M. Neron
ONERA. BP 72.92322 wtinm ccdw ( ~ ~ ~ 1

Abstrnct. La dthode est v d s e par ropporr b une erperimentorion

ONERA, sur I'hypersustenroteur lriple-cops RA16SCI. mec lo
A viscous - inviscid interoction nwnericd method for in- gkontem'e de bec shhre AMD. Une non-unicitk rkoliste des
compressiblepOws over multi-elements oicfoils, which is on ex- solutions d'kcoulements dkcollts a ktk mise en lm'hre par lo
tension of the nwnerical method previously suggested for dthode de cdcul. Un accord cdcul-expkrience sarisfoisont est
compressible otrached or seporared or even sidled flows over obtenu, mec le mo&le de turbulence b 2 kquations proposk.
oicfoils, is presented. The robust algorithms of the nterhod ore
c@le now ro converge os ne11 for anached flows or mas-
sively seporotedpOws. such as induced by slope discontinuities
of oicfoils or slots. or such as induced by stall. The viscous -
inviscid opproach introduces o self-odopkztive viscous grid in 1. INTRODUCTION.
both n o m 1 and streomwise directions along the displacement
swfaces, with everywhere o streomwise grid-resolm'on of the la mtthodc numMque d'interaction visqueux-non
some order os the seporoting boundnry [oyer thichnesses, even visqueux pour les tcoulements inwmpressihles d'hyper-
at slat opexes, which is believed to eliminate ony aleatory sustentateurs prcSenth id, dont I'aigine m o n t e ? lairef. 121
efect of numerical viscosity. des m b e s auteurs, a &e depuis lors deVelopptC en pmlonge-
The method is vdidnred with respect IO ONERA experiments, m n t direct de la dthcde numMque prop& par ailleurs
on the Ihree-element RAI6SCI high-lift device, with the severe pour Ics baulements compressibles attaches ou decOll6s de
AMD slat-geometry. A reolistic non-uniqueness of the seporot- pmfls d'ailes, Le Balleur [31[11P][9], dont les pro&s h n t s
ed flow solutions hac been exhibited by the colculation method. [3][1][9] ont permis de donna accb BU calcul des
A satiflactory agreement between theory and experiment is ob- dtcollements massifs et du d b h a g e pmfond.
mined, with the suggested 2 equation turbulence model. Dans le cas de I'hypersustenntion. M plus du p g d s
n h s s a i r e vets le calcul des ecOulemcnts massivement
d&oIl&, qui 6tait d&j&rencontr6 sur les pmfils d'ailes au
d b h a g e , le calcul doit W e r la plus grande wmplexite
Resume. des ewulements visqueux, ainsi que la plus grande complexite
- des geodtries. qui inwqment souvent des singularit&s.par
Une dthode n d r i q u e d'interaction visqueux - non exemple SLITles k s .
visqueux pour les kcoulements incompressibles sur Ies projls La pdsente d t h c d e n d r i q u e montre que la simulation
multi-corps, qui est un prolongentent de la &tho& n d r i q u e n d r i q u e par interaction visqueux - non visqueux p u t Cue
suggkrke prkckdemment pour les kcoulements contpressibles rendue o p h t i m e l l e , p o r n que les progn%srtcemment ob
anaches ou &coUks voire dlcrochcs sur les projls ddles, est tenus [3][1], & la fois sur les algorithms de wuplage cf SLIT
plpsenike. Les o l g o r i t h s robustes de lo ntethode sant m ' n - I'extension des modkles de wuche-mince. soient introduits.
f e ~ n tcopobles de convergence font pow les kcoulemena La simulation par interaction visqueux - non visqueux met
onoches que msivement dkcollks, tek que ceca induits por de plus I'accent SLIT les wnwintes de furesse de didtisation
les discontinui?ks de p e a des p r o m ou des becs, ou que ceux ntcessahzs, qui exigent que les pas d'espaoc en directMn de
induifs por le dkcroclurge. L'opproche dinteraction visqueux - l'&oulement dans le voisinage des decollcments soient
non visqueux introduit un m i l l ~ visqueux
g~ auro-adaptatif b lo toujours maintenus du meme ordrc que I'tpaisseur de la
fois en direction n o m l e , et dans lo direm'on de I'kcoulement wuche incidente qui va d&oller, du moins si un d c u l f i i s t e
le long des swfaccs de dlplmement, mec portout une de ces processus de dtcollement cst souhait&. G t t e wnoainte
rksoluion de mcrilloge dons lo direction de I'kcoulement du d'6chelle de didtisation. d'aigine physique, introduit des
& ordre que l'kpoissew des couches limites w le point de tchelles exugtnemnt petites sur les becs anguleux. Elk est
dlcoller. mhnc U a p a des becs onguleux, ce qw' (limine evidemment toujours &sente dans Ics calculs dnns Ies
I O U ~.$et d k m i r e de viscositk n d r i q u e . dthodes d'intuaction visqueux - non visqueux, w m e dans
11-2

les methodes de dsolution "directe" d'Equations de Navier-

Stokes. Le defaut d'une telle rCsolution di&tis& A khelle
h e reviendmt h faire reposer le processus de dkollement cal-
cule sur un effet ali?atoin de viscositC num&ique, B moins
d'inuoduire une mod6htion locale du type *' volume de wn-
uole".
S'il ne peut Ztre fiable de fah repser le calcul des zones
voisines du dbllement sur la dissipation num&ique,
I'introduction 6venNelle d'une modelisation additionnelle dans
les khelles fines ne doit pas Sue rejet&. Cependant, la
dthodologie d'interaction viqueux - non visqueux semble
maintenant offrir la pssibilit6 d'introduin A la fois l ' d s a
I'khelle de dsolution fine wmcte au dkollement, et I'acds
au calcul du dhllement massif, tout en wnsewant ces avan-
tages habituds de &s faible diffusion num6rique. Dans cette
optique, les raffinemnts locaux necessahs du maillage sont
d ' w abord plus ais6 si I'on maintient un calcnl non-visqueux
quasi-incompressible. d i d t i & par methode de singularit&.

= ps2+ w + pulw. - pulw. (3)

2. EQUATIONS ET MODELES.
PqEcos w + PW. ,*t.z.)

La pdsente d t h o d e de calcul, A l'exception du calcul

non-visqueux incompressible, est identique A la mtthode
n d r i q u e props& pour les hulements wmpressibles
attaches ou deCoU6s de profils d'ailes, incluant IC cdcul du
decrochage profond, Le Balleur [31[71[11. On pourra se repn-
-a=
as
COS y. -.axa 1
sin w = -
w.
4
er h [I] pour une pr6sentation d6taill6e.

2.1. Theorie de "Formulntion-De&itnire".

Les Equations de Navier-Stokes sont dissaci6es en 2

systkmes coupEs Equivalents en utilisant la thcorie de
"Formuladon-Deficitain" de Le Balleur [41171[91[31[11. La
W r i e est utili& ici en inucduisant une m c a m de
wuche-mince dans les Equations. Cette mncahue n'inucduit
pas de restriction majaue une fois &rite en "R€fbCntiel de En plus des Equations de masse et de mouvemenf le
Deplacement", voir [3] ou [l]. systtme (3), choisi ici de rang-3, inclut I'Equation
d'enuainement, qui est w e collocation le long de la surface
xz = ab,, avec I'Equation locale de muvcment du systkme (1)
23. Equations locales en "coucho-minee". selon x i . Conaairement A de nombrcuses affirmations, les
equations int6graIes sont m i n s rcsrricrives que celles de
Rappelons que la th&rie de "Formulation-Deficitre", F'randtl, parce que p,p . p. U J sont ici supposts 6ue Cvolutifs
props& vers 1980 et compli?t& par I'inucduction du selon xz au sein de la couche visqueuse, Le Ballcur [41[811121.
"Wf6rentiel de Deplacement" 131, permet de gh5aher cette extension ne dsultant que du choix optimal de definition
l'emploi d'Equations de wuche-mince au decollement massif. des 6paisseurs indgrales ("Formulation-Deficitaid' par rappon
Dans ce df&rentiel,la direction x z devient sensiblement nor- aux flux).
male A la surface de deplplacement (iconnue avant calcul).
x z = 0 est alms la surface de &placement et non-plus A la
2.4. Mod& de profils de vitesse turbulents pram?lriques.
pami, qui devient x2=,?,&I). Les Equations visqueuses, en
theorie de "Formulatto ' n-Deficitaire" [3][1], lorsqw Yon choisit
de les tronquer par des approximations de wuche-mince, On utilise la description analytique originale des profils de
s'ecrivent, si r l z dksigne la conwinte de cisaillement et si 3 , vitesse moyenne turbulents suggkee pnkddemment, Le Balleur
r;, designent respectivement le jawbien et les symboles de [71[91[81[31, moddisation wnstruite tant pour Ies hulements
Christoffel des cwrdonn6es cunilignes : attaches que mssivement dCwU&, voir [9]. Elle est ici
appliquk sous sa forme la plus compl&te[31. avec sous-couchc
laminaire et terme D d'aumrtissement :

-[ 1 - W ] . D + D 9 Ra [<] loi knninaire

de p ~ ~ i
 0
I- 11-3

avcc : F ( z ) = F * [ +I] -, Z F * ( z ) = [ ~ - z ' . ~ ] ~k =, O A l -r ./a

Cme d l i s a t i o n des profils de vitesse foumit aussi une

mod6lisation ghCrale de I'enuainemnt turbulent d'&uilibre
[3][71[11, compte tenn de la definition de E dans (3) :
0 1 0 1

3. METHODES NUMERIOUES.

Les nm?thodes nu&ques utilisCes pour la wuche

visqueuse et pour I'algorithme de couplage soot identiques h
cclles d6veloppees en fluidc compressible pour les profils
d'ailes. Lc Balleur [31[11[7l.
25. MoMe de lurbulence k-tau.
3.1. M6thode n d i q u e visqueuae.
Lcs effets de turbulence hm+uilibn wnt calcd6s eo
ajoumt le md-ile & 2 &uations de wosport de Le Balleur
[7][31. qui p u t Cfn utili& aussi bien dans Ies rdthodcs-locales Rappelons que la dthode n d q u e de couchc visqueusc
(calcul de 1'1). et dans les rdthodes-int6grdes (enwinement et est de nature hybride entre m&h&s-integrales et rdthodcs-
dissipation). Ce modae sugg&6 en 1981 a pour originalitc. locales.
d'une pan de "foxer" le d e de Launder-Rodi par la A chaque station ( X I ) , les profils de vitesse paradhiiqucs
d l i s a t i o n des profils de vitesse. ce qui le rend mieux adaptc turbulents des relations (4) wnt didtisCs selon x 2 entre
aux ecoulements dtcoll6s, et d'autre pan d'effectuer un calcul x2=zwet x 2 = 8 , avec un maillage en x2/S qui est autc-
en mmes dUcan aux valeurs dorm& par le mod&le adaptatif & 8 et h la forme dn pm6l de vitesse. La didtisation
d'&nilibre (conrmiptes r&z et enuainement E ~ ) L'&art . auto-adaptative utilise ici 37 nocuds optimiseS, d6pendants de
~&uilibreest suppost atre invariant en dinction 2,et S a cd- a!, avec des sch6mas implicites de ua&zes pour lier les gran-
d 6 selon X I par 2 &uations (int6grales) de transport. Cdes-ci deurs locales et integrales.
calcdent des effets d'histoire pour l'hergie dnetique Nr-
bulente moyenn6e en x 2 notee E(xl) , et pour la conuainte de De station en station ( X I ) , le s y s t h e d'euations integks
CisaiUement de Reynolds moyenn6e en x2 not& 'Or') : en x 2 est di&tiSC sur IC maillage curviligne Or1) de surface de
deplacwnent, projection sur cette surface du maillage c d g n e
de paroi (s). Notons que I'intCgration en x 2 (normale B la
couche visqueuse) dest plus ici normale h la p m i . Lc Balleur
[31[91 :

i = 1. 5, j = 1, 5, (50)

4 ?fcosw

3k
8
.I , '
b i = E* -ws
k 'r
t
b'

bS
siny=W./q
11-4

La prtsente mtthcde reprend la discrttisation robuste en (s , x') Le Balleur [3][1][101, en variables complexes :
dtvelopp4e depuis de nombreuses anntes, Le Balleur = o,f+la,,,..op= 1
[71[81[31[11, smarching et non-lintairement implicite en s, x'. 1 - (a,,,.,)
Elle commute difftrents schtmas numtriques en direction s, de
second-ordre, ou de premier ordre si le maillage devient gros-
sier, voir [I].
Les inconnues "non-visqueuses" du probleme visqueux
(5a), not& avec un signe chapeau, (9 , Gt,/q ), doivent etre
coupltes aux quantites correspondantes du probkme non-
visqueux. qui sont nottes ( 9 . wJq). Pour tviter les problhes
de singularit6 de dtcollement, le probPme visqueux est risolu.
avant couplage. soit en mode direct, wit en mode inverse. A
chaque noeud (i. I ) et iteration de couplage N , le systkme
d i s d t i d (5a) est fermt par la relation (Sb). elk-meme
discrCtisCe par des schtmas cen& pour (J$/Js) et Dans ces expressions, le terme p correspond i I'amplification
tventuellement d&en&s-aval pour (aq /as ) : complexe de I'algorithme de point fixe "direct" (p8 Ctant de ce
E = O :Direct fait I'amplification complexe de I'algorithme de point fixe "in-
E = I : Inverse verse"). On notera que la relaxation U', 0' ci-dessus
correspond i la forme originelk de I'algorithme "Semi-
,N inverse", Le Balleur [6], c'est B due B une relaxation optimale
od = I+-', mais que I'tcriture ici en variable complexe de o
inclut tgalement I'analyse de Wigton. Holt [ I l l , voir [12]. On
La commutation d'un mode direct (E = 0) vers un mode remarquera enfin que l'algorithme "Semi-inverse Massive-
inverse (E = I ) est asservie au p a r a d t n de forme du profil de separation" conduit, meme en subsonique, B ce que les 2
vitesse (a' > 0.28 par exemple). Le Balleur [71[31. termes o ' et o3soient actifs et non-nuls, Le Ballcur [31[10].

33. Couplage : Algorithme "Semi-inverse Massive-separation". 33. Calcul des sillnges.

A chaque iteration de couplage. la vitesse normale h la Le calcul visqueux direcdinverse et le couplage en effet
paroi non-visqueuse w: (ou bien la discontinuitt de vitesse de dtplacement wnt effectues sur les sillages de la meme fa-
normale non-visqueuse SUI les sillages <w:>), qui uaduit I'effet con que sur les couches limites, et les dtcollements tvennrels
de dtplacement gtoCralid, est mcdifite en chaque noeud de induits par les gradients de vitesse peuvent y Sue calcults. Ce
couplage ( i . 1) de m i b e explicite p i c e B I'algorithme calcul des sillages est conduit ici dans I'hypothkse du uaite-
"Semi-inverse" de Le Balleur [61[31[101[11 : ment simplifit issu de la seule partie symttrique des Quations,
voir Le Balleur [7]. L'effet de courbure est pris en compte
dans la mtthode, comme dans [71, mais tous les calculs ont ici
616 conduits sans cet effet.
En plus de I'algorithme de couplage, on utilise pour
chaque sillage I'algorithme de mise en Quilibre propost pour
les profils mono-corps, Le Balleur [7]. repris en dttail dans
[11[31.

o<o<2,
La prtsente mtthode utilise l'extension rbcente "Semi- 3.4. Methode nudrique non-visqueuse.
inverse Massive-separation" de I'algorithme. Ce nouvel algor-
i t h m gtntralise I'algorithme originel dans le cas oh wn/q Le calcul incompressible non-visqueux est effectut par la
cesse d'&mpetit, mais peut devenir de I'ordre de I'unitt, Le d t h o d e de singularitts proposte par le second auteur, dtjh
Balleur [31[11. pdsentte en (21.
Rappelons de plus que, dans la thtorie numtrique du Dans cette mtthode, le conditionnement numtrique est
couplage associte B l'algorithme "Semi-inverse", voir [I], la amtliort en distribuant sur les parois des profils. dans les mnes
technique numtrique s-marching du probkme visqueux permet de faible tpaisseur prks des bods de fuite, une double
B chaque station ( i . 1) une tlimination des variables internes dpartition de singularitts sources et tourbillons.
"purement visqueuses", Le Balleur [51[61[71[31. ce qui donne L'augmentation correspondante du rang du systErne est utiliste
p' en chaque noeud : pour imposer localement des conditions de dgularitt aux
&oulements fictifs que la distribution de singularitCs induit B
I'intdrieur des profils. L'emploi d'une telle mtthcde de
singularitts mixte dans ces zones de faible tpaisseur. associte
Les coefficients w:. of. o: ci-dessus sont, dans B des conuaintes suppltmentaires appropriies pcrmet en effet
I'algorithme "Semi-inverse" de Le Balleur, calcults en chaque d'utiliser les singularitts de f q o n optimale, en limitant l'erreur
noeud ( i ) de couplage. Ce calcul est tel que w:'l corresponde de discrttisation.
B la solution (lintaride) de couplage exact sur le nombre Un traitement numtrique satisfaisant dans les zones de
d'onde maximal [61[71[3] a,,. du maillage. Une surrelaxation faible Cpaisseur est tout particulihnent nkessaire ici dans le
de 2 est alors la limite de stabilitt [61[7]. Pour I'algorithme cas de profils multi-corps, dont les corps principaux et les becs
"Semi-inverse Massive-sepmtion". les coefficients s'tcrivenh prisentent p r h de leurs bods de fuite des tpaisseun
 11.5

g6orhkment faibles. 43. Conditions des mmparaiaons falculsxpglence.

NumCriquemnt, pour les ithtions de couplage. In
matria inverse du systbme non-visqueux d i d t i & (sillages Les calculs sont c o m p d s nux rcSultats d'essais I131
exclus) est invnriante, et stock&, aussi longtemps que le mail- obtenus B I'ONERA dans les souffleries SI Ca en 1980 et
hge (adaptatif) n'est pas dactualid. F1 en 1984. Ces essais ont et6 effectots entre panncaux (b =
1,Ol m). sur une maquette de 0.5 m de corde de d f h n c e , B
une vitesse de 40 4 s . et B des nombres de Reynolds de 1 2
I@ et l,9 I@. ~ e sessais ont et6 effectu& en transition
naturelle, les effortsttant obtenus par intdgration des pressions
4. RESULTATS. sur les pmfils.
La comparaison aux rbdtats experimentaux &it 2-
Les dsultats de calcul cidessous ont et6 obtenus sur un effectuhe en tenant compte des difiicultes exphimentales. En
pmfil d'hyQersustentateur consmit h panir du profil supercri- effet. sur une telle configuration multi-corps. pour pdscrver B
tique RAl6SCI de I'AhspatiaIe, expckimente avec un volet et I'6coulement un caractkre bidimensionnel. il est nkcessaire de
diff-ts types de becs [13]. Dans le but d'6valuer les perfor- disposer des caissons d'aspiration sur les parois latkrales. Les
mances et les pmgrks r&ents du code de calcul (code VISlBb), essais ont monM que dans cenaines configurations. par aspira-
la configuration de bec la plus Ah,dotee du bec A.M.D. ,et tion I a t W e B I'exaados du volet, il etait possible de passer
so w de dhllements pmfonds. a et6 Selectionnk comme cas d'une solution d h l l 6 e sur le volet h une solution non-decoll&
test. Les configurations calculees correspondent aux valeurs de par simple variation de la pression d'aspiration.
-20" et -40° de I'angle de braquage du volet he,.
Par ailleurs, lors de certains essais, des dHormations de
gbm&ie ont kt6 constatees, influant notamment sur la largeur
41. Description de I'hypersustentateur tripleeorps de fente corps-volet, (cas L , - Z O " ) , ou sur I'angle de bw-
quage du bec, dont la valeur mesur6e ap&s essais ttait de 32'
Conhaimnent B la configuration double-corps, trait& B l'emphture au lieu de 3(P, la valeur en murs d'essai h mi-
antkiemment [2] avec une version beaucoup moins avancbe envergun. vraisemblablement plus &v6e nUtnnt pas connue
de la methode de calcul, et d'une complexit6 ahdynamique avec certitude. Enfin, les effom globaux obtenus par
intCgration des pressions induhnt une incertitude nu niveau du
bicn mindre, la configuration mplecorps (fig.1) inucduit m e
difficult6 majeure par la g6om&ie du bec. Celui-ci prbsente bec.
en effet h I'inaados une singularit6, B I'aval de laquelle se situe
une cavitk. La difficult6 de calcul dside mins dans la singu-
Isrit6, et dans le d6collement qu'elle dblenche dans la cavit6,
que dans sa geOmtuie particulihn, Ies parois nmont et aval y 4d. Courbe de portance. Volet braqu6 P 2V.
prcSentant en effet un angle aigu.
Dans cette premihe configuration, les angles de braquage
du bec et du volet ont t'espectivement pour valem 30' et -20".
42. Conditions de enlcul.
Sur la courbe de portance, les dsultats de cald. f i g d s par
des mix. ont tt6 obtenus jusqu'B I'incidence de 20.Zo, fig. 5.
La m6thode de calcul des couches visqueuses doit traiter B Les dsultats exphimentaux (F1 1984) sont reprcSent6s par des
panir de cette singularit6 un dhllement profond de cavit6. cercles. A 16'. les deux B U ~ M cmix reprCsentent les r&ultats
me couche de d h g e apparaissant B In singularit6. avant de obtenus avec des valeurs de I'angle de braqunge du bec 8, de
ncoller vers I'aval. Dans un tel cas, une technique de calcul 32' et de 33'. L'ttoile nprescnte le rCsultat obtenu dans la
simplificnuice souvent adopt& consiste par exemple B configuration de df6rwce avec le mod& de turbulence B 1
modifier artificiellement la gh&& de la cavit6. de fwB equation de uanspar.
adoucir le proassus de decOUune.nt.
Notons qu'au voisinage du dtcrochnge, dans cette
Le recours B ce typ de cnlcul simplilicatevr n'est ici pus configuration corm~edans les suivantes, le calcul est effectue
par continuit& en initiant les calculs ? i de dsultats o b
panir
nhssaire. La demarche retenue, permise par la dtbcde. @Ice
h sa th6nie de "Formulation Dcficifaire", voir 111, et nu tenus B une incidence voisine.
"Rcfcrentiel tangent B la sutfnce de deplafement". a consist6 i
catserver la forme cxacte de la cavitk, et B ne modifier la 45. Incidence 16O. Volet braqu6 P 20".
geomCtrie qu'en 6moussant l'apex de la singularit.5.
L'6moussement est ici obtenu en inucduisant un an: de cercle
de trts faible rayon (1,36 IO-'). mail16 ici par 60 nauds sur Sur la Fig. 6, les mef6cients de pression Cp(x/c) calcults
environ 150° (fig. 32.4). Sur cet 6moussemnt, le processus sont cornpar& B I'expCrience. Lcs positions des points de
physique de decollement (compression) p u t dm effectivement dhllement et de recoueslent duns les couches visqueuses ont
disdtist. Cette didtisation h k h d e h e , dictec par &k i n d i q u h sur la reprGmtati~du profil. ainsi que sur les
I'khelle de la couche limite juste en amont, assun que le pro- sillages f i g d s dans leur position d'esuilibre h convergence.
cessus de decollement et de naissance d'une coucbe de L'acwni est excellent sur IC volet et la quasi-totalit6 du
melange soit ici effectivement calcule et ne soit plus d6tamink corps principal. Sur le bac, ainsi que sur le corps principal B
par Ies seuls effets de viscosit6 num6rique. l'endroit de la pointe de Cp. la cornparaison est moins borne.
Le8 calculs ont kt6 obtenus en laissant la nansition se
Toutefois, confirmant les mnsrques des exfirimentateurs. les
dtveloppa de fapon naturelle, et sans mala les ~paissemdes calculs ont monM I'influence du p a r d t r e de l'angle de b-
couches limites. Us sont systhatiquemnt effectuks, sauf indi- quage du bec, les dsultats ttant trts diff&nts lorsque cet an-
cation contraire, avec le mcd.?.le de turbulence h deux equations gle prend, nu lieu de la valeur nominale de 30°. la valeur de
de transport. 32'. Fig.8, ou de 33O, Fig. 7, compatible avec Ies d 6 f m t i o n s
de maqunte.
114

Pour ces deux valeurs conigCcs du braquage de bec, tant de mhercher une modificatiw dcs umditions d'initialisation
le niveau de Cp dans les pointes de survitesse, sur le bec et le du calcul. en agissant sur Ies conditions initiales ou sur le pm-
volet. que le Nveau de pression dans la zone de dhllement B cessus de convergence, mais m s modification aucune de
l'inuados du bec, sont beaucoup plus proches de I'expkience. l'ensemble du traitetnent numerique, afin d'explonr I'existence
la milleure comparaison semblant etre obtenue pour la valeur d'une eventuelle autre solution. Cette seconde solutiw. par-
33", Fig.7. Remarquons qdouue I'orientation du bec, des faitement convergk, a &? uouvk. effectivement, par simple
parimbues tels que les largeurs des fentes, non consideds ici, diminution de la frcpuence B laquelle la g6omCtrie des gillages
peuvent avoir une influence imprtmte. En ce qui conceme la est remise en cause en vue de leur 6quilibrage (reanUalisation
&&ode de calcul, l'absence aotuelle de prise en compte des toutes les 300 i t h i o n s , au lieu de 20). Cette modification a
couches limites confluentes peut Cue egalement responsable pour effet de figer la geOmetrie des sillages pendant Ics 300
d'une panie de I'Ccan observt. p r e m i k s ithations, pendant lesquelles s'effcctue la plus
grande partie de la convergence. La seconde solution A obw
Sur les figures suivantes, la distribution de pression nue ainsi pdsente alas un dkollunent pCnMis€ A I'extrados
Cp(x/c) sur le bec a et6 representee A diffhntes Cchelles, tout du volet Fig.21. 23, mais ms aucune trace du decollmmt
d'abord dans la mCme orientation que sur I'hypersustentateur. pl.ec6dmuaent observC dans le sillage du c a p s principal (Solu-
Fig. 9, puis a6n de mieux visualiser la solution calculk au tion B : Fig.20, 22).
voisinage de la singularit6 (6m0ussk)~ ap&s une rotation de
U a ainsi 6t6 mis en tvidmce I'existence de 2 solutions
l'image placant verticalement la bissectrice de I'angle f m 6
diferentes dans une m2mc configuration. stables toutes deux
par les dew parois, de p m et $autre de cette singularit&,Fig. par ramort B l'algorithme de calcul. I'obtention de l'unc ou de
IO, 11, et 12. La position des points de decollement et de l'autre ne dependant que de I'initialisation.
recollement a kt6 indiquke. Un second dhllement, &s lim-
it& swvient dam le sillage, B panir du bord de fuite, Fig.9 et
IO. La Fig. 12 montre le detail de I'eCoulement autour de la 4.7. Courbe de portanre P 2 brandtea Volet P 400.
singularid 6mousSee. Le dkollemnt s w i e n t sur I'arrondi.
a@s le pic de surviresse, au maximum relatif de Cp. On ob-
Cette m€me obserfanon d'un dedoublunent de solution a
serve ensuite un petit plateau de pression, toujours sur
et6 effectuk. A d'autres incidences. puis en conduisant les cal-
I'arrondi, qui semble se raworder seulement plus en aval nu culs par condnuid au sein de chacune des deux f d w de
plateau de pression associe B la cavit.5. Fig.12 et 11.
solutions, chaque calcul Ctant initialise A panir d'un calcul A
La figure 13 permet enfin de companr les r6sultats de cal- une incidence voisine.
cul obtenus sur k m2me cas de dfbrence avec le modUe de Db lors que la lnfurcation vets I'un ou I'autre type de
turbulence A 1 Quation de transport. Sur ce cas. les dsultats
sont exui?mment p b e s . Cette conclusion ne saurait cepen-
solution s'est muv& effCCN&. une modification 6vcnhtelle de
la fdquence de r&ctualisation des sillages n'a plus aucune in-
dant 2tre gentrale. Avec le modWe B 1 huation, le niveau de
cidence SUI le type de solution, chaque calcul resfant du memC
Cp est trzs ICg8rement inf&ieur sur la partie ani& de
type que celui ayant mi B I'initialiser. Lcs deux branches. A
I'exuados des uois corps. en pardculier sur le bec.
et E, sont totalement disjoints sur la large plage d'augle
Sur les figures suivantes sont pdsentees des vues de d'incidence deCrite ici. Fig.24.
I'eCoulement autour du bec. Les profils de vitesse. Fig.14. font L'interpd~on physique du dedoublement de solution
apparaitre la difficult6 du calcul, inaccessible aux metbodes de
calculc parait simple. Le braquage Clev6 du volet Sde=-4o0
touche llmite classique, Yotientation du vecteur vitesse par
impose B la partie de I'bulement comprise e n m le bec et le
rapport B la pami, dans la couche de melange @s de la singu-
volet une forte dcviation, inaccessible en dui& visqueux sans
Mt6, etant voisine de 90'. Les q u a figures suivantes
pdsentent B convergence les lignes de courant, respectivement decollement de I'une ou l'autre couche visqueuse qui le borde.
non-visqueuses et visqueuses, Fig.15 et 16, ainsi que les isD Le dtcollement se pmduit donc, soit dans le sillage du corps
mach Fig.17 et 18. Elles font appmitre en pardculier la mne principal, soit sur la couche limite extrados du volet,
de recirculation 6paim et la position du point de recollement I'eCoulement ayant la double possibilite de sui= soit la dircc-
L'examen de I'koulemnt a u t m de la singularit6 &mouss& tion du sillage, soit ceUe de l'exuados du volet. Le coaa8lc de
m e t de visualiser la petitess des khelles physiques de la bifmaaon entre ces deux familes de solutions doit pouvoir
decollement, qu'il est nhssaire de discrttisa, sous peine de s'effecmer par modification difftnntielle de la pression nu sein
remplacer le processus physique par un processus de dissipa- des deux couches visqueuses. C'est-Adire par aspiration eVou
tion num&ique, Fig. 19. soufflage. ceci dans le calcul comme dans I'expCrience.
L'examen des doctunenu d'essais, qui mentionnent la pos-
sibilite observte d'obtenir sur I'extrados du volet une solution
4.6. Non-unicitd des solutions. Volet braque P 400. d h l l 6 e ou non par simple modikation de la pnssion
d'aspiration sur les parois latMes, A l'extrados du volet.sem-
Dans la seconde configuration 6tudi& de ble confirmer cette hypothbse. La d6termination du domaine
I'hypersustentateur. I'angle de bmquage du volet est de d'incidence et d'angle de braquage du volet dans lequcl cette
=ao, pour un m€me angle de braquage du bec S k d O o . Dans double solution survient reste ici empirique.
ce cas B fort deCollement, I'analyse par le calcul a mis en
Sur l'une des branches, comespondant au cas du
bidence et interp&d un phenombne de non-unicid de solu- d6collement dans le sillage du c a p s principal, 11 a et6 possible
tion. de poursuivre les calculs au delS. de I'incidence de denochage
Le premier cas de calcul aboni6, effeaU6 ti une incidence du profil hiple-corps. Fig. 24. Sui I'autre branchc. et pour
a=16', dans les m2mes conditions que pdcedetnment avait en Le,=-20'. cela n'a pas encore et6 possible. plusieurs
effum i s en evidence une solution comportant, B l'muados du difficulth se conjuguant lorsque l'on se r a p p h e de
volet, un d6collement ds limit6 p&s du bord de fuite Fig.20, I'incidence de decroChage, celm-ci semblant s'effecmer de fa-
22, conmcirement B la solution attendue, et comportant par con- con plus brutale que sur la branche prtcedente.
tre un d6collement intense, B facteur de forme maximal
H40.8,dans le sillage du corps principal. Il a donc kt6 tent6
 11-7

4.8. Incidence 16'. Volet brnquC h 40". gumme d'incidence, correspond h unc bifurcation entre. 2 t y p s
de d6wllemnts massifs, siN& m t 8 t dans le sillage du corps
En dectionnmt le cas de dfCrence exp&imental, c'est-a- principal, tant6t sur le volet.
dire la solution prtsentant un dbllement ghcralisC sur le
volet, la comparaison avec les rcSultats expCrimentaux est ex-
cellente. en pnrticulier en ce qui conccmc la zone dkollCe & Brmerekments. Lw OUICIU~adressenf ici leurs remrciemew h
l'exuados du volet (fig.23). Comtne dans le cns B S-,=-20°, la D. Blaise, P. Girodrow--Lavigne. H. Gassot, pour leur concours
cornparaison est m i n s borne sur le bec et sur le corps e c i - pr6cieua dnns Ies walremnu g r & q u e s . a i d qu'h S. Henry. qui a
pal, l'endroit de la pointe de Cp. Le nivcau de Cp & wavaillL sur ce code lors de son passage 6 I'ONERA.
I'extrados du volei, en mnont du decOllement, est egalement
plus faible que dans l'expkience.
Toutefois, et plus e n m que dans le cas prMdent, REFERENCES.
l'accord calcul-ex&ience s'amCliorc lorsque l'mgle de bra-
quage du bec pend. au lieu de la valeur nominale. la valeur
&u=320, Fig.26. La dpnrtition de Cp(dc) sur le bec, Fig. 25 [I] LB BAUSUI( 1.c - Viscwphviseid calcvlafion of high-lift sepmtcd
mmp'esdblc flows ova airfoils and wings. FWcedmgs AGARD Symp.
et 21, pdsente le &me componement que dans le cas m High-lift dcviccs scmdynamics. Banff, Csnsds. 5-8 Oclobo 1992.
pr&dent (Sd,=-2OD : Fig 10 et 12). La cornparaison des AGARBQ415. Panu 26.10 soaar.
lignes de courant et des isc-mach pour chacune des deux solu-
tions obtenues B I'incidence a=16O, pour Sdd=-400, Fig. 28,
30 et 29, 31, (solutions dkjB illusw6es par les figures 20.22, et
21, 23), montre les deux types d'hulemmt possibles, avec un
dkcollement soit dans le sillage du corps principal, soit &
l'exuados du volet

49. A& au calcul du dkrochaee.

La d t h c d e a montd sa capncite unitu des cas de

dBcollemnts gCn&lids sur une telle configuration, dp-5 la 161 LE BALLEki J.C - Viscaus-inviscid flow mtchin!Z : Numerical muhod
;dd appl~%& I0 Nmdim~nOnslwRl(ruc and SE c flows. . La
d i f f b c e considCrable d'6chelle entre les epaisseurs ReJrrck Auospaurlc 1978.2, p. 67-76 (March 197d$%h. OT English
visqucuses sur la singularit6 bnouss4e U dans le d6wUement WnSL €SA-TT-496.
principal. Fig. 32, 33. et 34d. 34g. Ces figures pdsmtent
mpectivemcnt le maillage de calcul du h i d e visqueux, le
champ de vecteurs et les iso-mach, & l'incidence de 19".
La cornparaison des isomach B divesses incidencw, au
cours du deauchage, fait a p p m k e deux Ctapes dans la pene
de portnnce, dtj&visibles sur la Fig. 24, la p r e m i h cornspan-
191 U WUR J C - Vl0cw.Inwcid mtuacum solvw and wmpulauon
dmt B la remontk vers I ' e x d o s du pmfl du decollement si- of b hly SC led flows .' S t d m of V M e X Lhmwwcd Flows'. cha 3.
tu6 dans le sillage du corps principal, In seconde comspondant p 1st 192. !
& d ICASE symp NASA Langley F d d . USA. Ouly 810.
& sa g6nCralisation & tout l'exuados, Fig. 34.

5. CONCLUSION.

La d t h c d e n u d q u e d'interacticm visqueux-non
visqueux pdsentk ici (code VIS18). inep Ies pm&s
dalis6s sur les profils monocorps dans le cas de dkollements
massifs.
Elle a pu €@emise en oeuwe avec s u d s sur un hyper-
sustentateur triple-corps, dans le cas d'un bec pdsentant une
singuluit6 fonement prononck, l'intrados du bec formant une
cavitC. Les calculs ont pu etre conduits sur la geomctrie rW1e.
simplemnt b n o u s s k nu niveau du point singulier par un ar-
mndi de t& faible rayon. Mal& la difficult6 de discr6tisation
& &s petite khelle impode par cette SinguluitC, la d t h o d e a
p m n i s le calcul de deux configurations multi-corps dans une
large gamme &incidences.
Elle s'est av& capable de traiter des cas d'kcoulements
p m f o n d b t dkcollts autour d'un pm6l multi-corps, nu-del8
de l'incidence de &mhage.
Elle a permis de meme en kvidence, sur le cas du pmIl
RA16SC1 avec brquage du volet a -40'. une non-unicite
d i s t e de la solution, les 2 solutions obtenues Ctant parfaite-
ment convergCes. Cette double solution,persistante sur toute la
11-8

\
Fig. I . GCodirie. (RA16SCI. Sk=3O0 I Sda=-200).

Fig. 2. DCrail de la poinre du bec (singularit&houss&).

Fig. 3. Gkodrrie du bec AMD. Profrl RAl6SCI.

Fig. 4. Derail de I'arrondi 6 la poinre du bec.

3.0

25

2.0

I.J

1.0

0.5

0.
LPH
0. 4. 8. 11. 16. x . z,,

Fig. 5 . Courbe de porrancc

(RA16SCI. M=0.123, Re=I,8 lo6, Sydet=-2O0). Fig. 6. Dismburion de pression
(M=0.123 Re=1,8 IO6. ~ = 1 6Sk=3Oo,
~ ~. Sde,=-20-).

10
FCP

Fig. 7.Distribution de pression h la poroi Fig. 8. Disfriburion de pression 6 la pami

(M=O,123, Re=1,8 IO6. a=164 S k , = U o , S,,=-20°) (M=O,I23 , Re=I,8 IO6, a=16'. SkC=32', SV,,=-2O0).
 8.

JCP

.I
S.
I.

3.-

1 -

I. -
0. -
kr.'
........ ;.,
.....
..:e
..:....;..
.....
.i
TCP
7.

.,I[..........
11-9

-8.I

I/
-.-.-........
-..__.
/---

0.S-
(1.

-
-0.5

- L
Fig. 12. Distribution de pression sur la singularirk imoussie
Bec (M=0,123 , Re=1,8 IO', a=16'. &,=3

Fig. 13. Disrriburion de pression

(M=0,123, Re=1.8 IO6, a=16', 6kc=30", 6,,=-20"
Profi1.s de viresse

Fig. I S . Dkcollement mossifb I'inrrados du bec

Lignes de couranr non virqueuses Lignes de couranr
(M=0,123, R e = I 8 106, a,-16", 6hc=30a I 6"&,=-20"). (M=0,123, Re=1,8 IO', a=16'. 6kc=300,6v&=-200J.
11-10

Fig 17. DCcollemenr mmsifh l'inrrdos du bee

kgnes iso-mach non visqueuses Lignes iso-mach
(M=0,123 , Re=1,8 IO6,a=169 6,=30°, 6de,=-200). (M=0.123 Re=I.8 lo', u=16O. S,,=3Oa,
I 6,re,=-200).

Fig. 19. Derail du dtcoilemrnr sur la .\ingularlri 6moussEe

Lignes iso-mach
(M=0,123 , Re=l,X NI6, a=16O, 6k=30" , =-20").

10.-
I..-

--CP --CP :~ ( l r i l 2 I O ' J

- - i
- -
~

-
~

-
. .
- -
- -
- -
- -
-0,- -1. -
 11-11

'1
05

00
,
6
,
I
,
12

Fig. 24 Courbe de porronce.

.
,
16

(RA16SC1,M=0.123 Re=1,8 106, S d a = 4 O o ) .

A,LPHi
20 24

Fig. 25. Bec. Distribution de pression sur la singdarilt

tmousste. Solution A (orienrarion aprks rorarion. del&
(M=0.123 , R e d . 8 IO6, ~ ( = 1 6Sk=300,
~. SV,,=-4O0).

0. P., 0.. *.I 0.4 ... 0.8 0..

0.. I., I.. IS 1.4 I.'
(.I#.I I.. I..

Fig. 26. Distribution de pression :Solurion A Fig. 27. Bec. Distriburion de pression sur la singdarilt
(M=0,123 , R ~ l . 8lo6, a=16', Sk=32' Sh,=40"). imoussie. Solurion A (orientalion aprks rotalion. derail)

Fig. 28. Non-unicirk. Lignes de couranl :Solution B Fig. 29. Non-uniciti. Lignes de courant :Solution A
(M=0,123 , Re=1,8 IO6, a=16', SkC=3O0, 6,de,=-4O0). (M=0,123 Re=1,8 lo6, a=16'. S,=3Oo,
I &&,=40°).

Fig. 30. Non-uniciri. tignes iso-mch :Solution B Fig. 31. Non-unicirt. tignes iso-mch :Solution A
(M=0,123 , R e d . 8 lo6,a=16'. ,,S =300 , =40"). (M=0,123, Re=1,8 lo6, a=16",Sk=300, &,=4Oo).
11-12

.....-
a : a=17,5"

Fig. 32. Maillage vis uellxauro-Mqrarif :Solution B

(M=0.123, Re=1,8 ,'OI a=19', Sk=3Oo, 6vda=-400)
,_..

.. , . . . . . .. . . b : a= 17.P

Fig. 33. Profils de viresse :Solurion B

(M=0,123 , Re=I.8 10'. u=19', Sk=300, , ,a =40°).

c : a= 1 8 7

f: a= 1 8 7

d : a= 19'

g : a= 19'

h : a= 19.1" e . a= 19.1'
Fig. 34. DCcrochage. Li nes is0 m c h :Solurion B
! --
(M=0,123, Re=1,8 10 , 8,,-30Q, SYd,=-4O0)
 12-1

HIGH-LIFT SYSTEM ANALYSIS METHOD

USING UNSTRUCTURED MESHES

K. de Cock
National Aerospace Laboratory NLR
P.O. Box 90502
1059 CM Amsterdam, The Netherlands

SUMMARY calendar time needed for grid

generation. One valid approach of
A 2D High-Lift configuration generating grids is the structured
analysis method is described. The approach, as long as it is
flow model used is based on the sufficiently automated. Examples of
Euler equations, discretized on High-Lift results obtained with
unstructured meshes. The generation block structured meshes can be found
of the unstructured meshes is based in Refs. [l] and [2]. Another
on the principle of successive grid approach consists of the use of
adaptation with respect to the unstructured meshes (see for
geometry. This approach makes later instance Ref. ( 3 1 ) . It is clear
extension towards fully integrated that CFD methods on unstructured
grid adaptation with respect to the meshes use more computer resources
solution straightforward. The main than their structured counterparts
characteristics of the Euler solver (interconnections between the grid
are upwind flux-difference splitting nodes need to be stored for
of the convective part of the Euler instance). The continuously
equations (second-order accurate increasing computer power is one of
discretisation in space) and four the reasons why the use of
stage Runge Kutta local time unstructured meshes in CFD is
stepping. Results obtained with this booming now. It is also clear that
analysis method are shown for the full automation is more readily
NACA0012 airfoil and three element achieved with unstructured meshes
airfoils. Conclusions are drawn. (Ref. [4]) since the grid generation
is based on autonomous algorithms
and not on partial differential
1. INTRODUCTION equations that include interactively
tuned source terms.
The complexity of High-Lift system The current paper shows flow
flow phenomena requires appropriate solutions based on the Euler
analysis methods, able to simulate a equations and obtained on
number of relevant flow mechanisms. automatically generated unstructured
The current High-Lift flow meshes. It is shown that the grid
simulation models vary from models generation is not dependent on the
based on panel methods to models geometrical complexity of the High-
based on the Reynolds averaged Lift system.
Navier Stokes equations. A Following cases are treated:
consequence of the complexity of the
geometries under interest is a case1
strong pressure on the turn-around- NACA0012 single airfoil, Mach
time of the High-Lift system number at infinity equal to
analysis part in High-Lift system 0.85 and angle of attack equal
design methods. This turn-around- to 1 degree.
time requirement partly explains the
success of panel methods in CFD. For case2
field methods, a computational grid NLR422 three element airfoil,
covering a 2D or 3D domain is always Mach number at infinity equal
required. In order to attain to 0.20 and angle of attack
sufficient accuracy while limiting equal to 10 degrees.
the number of grid points, the grid
should contain most of the grid case3
points near to the boundaries of M 1 three element airfoil,
interest. Given the geometrical similar to case2 but
complexity of High-Lift devices, extensively measured in the
grid generation is often a hard NLR HST tunnel. Mach number at
task, resulting in a large amount of infi'nity equal to 0.22 and
12-2

angle of attack ranging from - A third algorithm, hereafter called

2.5 to 25 degrees. successive grid refinement, does not
have the previously cited drawbacks.
It will be shown that grid
2 . GRID GENERATION generation is considered as
adaptation of an unstructured grid
2.1 Context with respect to measurable
properties of the geometries under
Various algorithms for unstructured interest. Hence adaption with
grid generation exist. The two most respect to measurable properties of
frequently used algorithms are the solution is straightforward.
Further the successive refinement
algorithm automatically produces
a) the Delaunay triangulation coarsening of the grid towards the
algorithm: basically this algorithm outer boundaries of the grid and a
tells how to connect a given set of surface grid suitable for
points into .a valid triangular grid. aerodynamic purposes.
2.2 Grid generation by
b) the advancing front algorithm: successive refinement.
given the initial front (the surface
grid on the geometry under interest The following assumptions are made:
€or instance) the algorithm
prescribes a method to advance the A1 : Let M be a given geometry
initial front towards the outer around which the flow is to be
boundaries such that an unstructured calculated.
grid results.
A2 : Let the geometry M consist of
Both algorithms have been K,,..., ,...,
Ki 5 , a set of n
successfully used in the past and non-intersecting contours.
are the subject of constant further
development and improvement. There A3 : Let each contour Ki consist of
are however some drawbacks of both
algorithms.
...
mi segments S,, I Sj, ...
,S,.
(This allows to treat C, type
contours, see Fig. 1).
Both algorithms a) and b) do not
form a natural basis for grid A4 : Let Saj be the analytical
adaptation with respect to the flow representation of a segment Sj.
solution while a fully integrated The analytical representation
approach of grid generation and grid Saj of the segment Sj can be
adaptation is believed to be used to generate pj points
advantageous. P,I... ...
,Pk, ,Ppj on segment Sj.
Algorithm a) needs an interactively
made so called back ground grid to A5 : Let Ssj be the spline
ensure certain properties of the representation of a segment Sj
based on the points
grid like coarsening of the grid
towards outer boundaries. This -
PI,. . ,Pk, ,Ppj-
introduces user interaction,
diminishing the automation level of A6 : Let Spj either be the polygonal
grid generation based on the representation of a segment Sj
Delaunay triangulation algorithm. based on the points
.
PI,...,PI,,..,Ppjor the convex
Algorithm b) needs a suitable hull of the spline
initial front. In case the surface representation S B j of a segment
grid is chosen as initial front, the sj .
construction of the surface grid
becomes critical for the resulting A7 : Let all polygonal
grid quality in the neighbourhood of representations Spj further be
the surface. A mere triangulation of chosen such that the contours
the CAD/CAM surface description is K,,...,K,,...,K,, are closed.
often not suitable enough for (Remark that the task of
aerodynamic purposes. So the closing all contours is not
generation of the initial front can trivial. All current flow
limit the automation level of grid simulation systems based on
generation based on the advancing field methods, structured or
front algorithm.
 12-3

unstructured, need contours convergence rate of the Euler

which are closed.) solver.
A8 : Let G be an unstructured grid The requirements for the polygonal
covering the geometry M. Two representation Sgj of the segments Sj
equivalent representations of based on the points of grid G can
grid G are used: also be identified a priori.
Important examples are:
G,: node table
node connection table Rsl: Each Sgj should not be empty.
per node
Rs2:. Each Sgj should at least
Gb: node table consist of three points.
node connection table
per triangle Rs3: The grid spacing along the
geometry M should be
A9 : The grid G is considered as sufficiently small in order to
given. It can either be a have enough resolution on the
default grid consisting of 7 geometry M. If A is the total
nodes (start from scratch) or arc length of the geometry M
a previously generated grid dnd AA is the arc length
(restart). between two points of any Sgj
then :
A10 : Let Sgj be the momentary
polygonal representation of a
segment Sj based on the points
of the grid G that would lie
on the polygonal
representation Spj of segment
Sj, in case the grid generation should hold.
would be stopped at that
moment. Rs4: The grid curvature along the
geometry M should be
All: Let F be a given flow solution sufficiently small in order to
on grid G. In the case at hand accurately represent gradients
F is the solution of the Euler of the flow in the solver. If
equations. a, b and c are three
subsequent points of any Sgj
In order to ensure with a given flow and alpha is the angle between
solver a successful and reliable the vectors ab and bc then
simulation of the flow around the
given geometry M I some requirements
of the grid can be identified a
priori:
Rgl : The far field boundary of the
grid G should lie sufficiently
far from the geometry M.
(usually 30 - 100 chords away
from the geometry M). This
distance is an input of the should be valid.
grid generator.
Rs5: All the segments Sgj together
Rg2 : The grid G should have most should form closed and non-
resolution in the intersecting contours with the
neighbourhood of the geometry same topology as the segments
M. Coarsening of the grid G SPj
towards the outer boundary is
still inevitable, even with The basics of the current grid
nowadays powerful computers. generator can now be summarized:
Rg3 : Each node of the grid G should - So called "grid generation" is
lie in the centre of the adaptation of the given grid G
finite volume surrounding it. with respect to the properties of
This grid smoothness the polygonal representations Sgj
requirement, if met, will of the segments Sj of geometry M.
favour the accuracy and
12-4

The polygonal represen- of finding the intersecting pairs

tions Sgj of the segments Sj of of N line segments does not
geometry M are used to require O(N*N) but O(N log(N))
determine the refinement operations Ref. ( 5 1 . In the
flags of the grid nodes. particular case at hand, all the
intersecting pairs consisting of a
- So called "grid adaptation" connection of representation G, of
is adaptation of the given the grid G and a line piece of the
grid G with respect to the polygonal representation Spj of
properties of the given flow any segment Sj of the geometry M,
solution F. should be found. If the grid G
The flow solution F is used consists of K connections between
to determine the refinement nodes and the polygonal
flags of the grid nodes. representations Spj of the
segments Sj of the geometry M
Hence grid generation is the result consist together of L line pieces
of the recursion of following then N = K + L. As a consequence
algorithms: the construction of Sgj can be
implemented in a efficient way.
algol: assignment of an "in-contour"
flag (Ref. [6]). algo3: control of the grid generator
based on Sgj.
Given a node i of the grid G,
detect whether node i lies inside Test if all Sgj meet requirement
or outside the closed contours
...
K,, .
,Kit.. ,%. This algorithm is
known as the Shimrat's Algorithm
Rsl.
If the answer is negative, then
all nodes of the grid are
Ref. [ 7 ) . It requires O(K*Q) flagged.
operations, K being the total If the answer is positive, then
number of line segments of the
polygonal representations Spj
...
forming together the contours Test if all Sgj meet requirement
... ..
K,, , X i , . ,K,, and Q being the
number of nodes for which the "in-
Rs2.
If the answer is negative for a
contour" test is used. It can be particular Sgj, then all nodes
shown that the required order of having intersecting
operations O(K*L) can be reduced interconnections with the
to O((K+L) log(K+L)) (see algo2). corresponding Spj are flagged.
If the answer is positive, then
algo2: construction of Sgj. ...
Assume that a connection of Test if all Sgj meet requirement
representation G, of the grid G Rs3.
intersects the polygonal If the answer is negative for a
representation Spj of a segment Sj particular Sgj, then flag all
of one of the contours Ki of nodes that have intersecting
geometry M. interconnections with the
If the grid generation would be corresponding Spj causing the
stopped at this point, the
intersection point would be one
problem.
If the answer is positive, then
data point of Sgj, the polygonal
representation of the segment Sj
...
based on the grid points. So by Test if all Sgj meet requirement
collecting in an array all Rs4.
intersection points, Sgj can be If the answer is negative for a
constructed (only grid points that particular Sgj, then flag all
have a negative "in-contour" flag nodes that have intersecting
can become member of Sgj). interconnections with the
In essence algorithms algol and corresponding Spj causing the
algo2 are equivalent to the so problem.
called geometric intersection If the answer is positive, then
problem which has extensively been
studied for linear programming,
...
hidden line elimination Test if all Sgj together meet
algorithms, wire layout etc. It requirement Rs5.
has been proven that the problem If the answer is negative, then
 12-5

flag the nodes that cause the "in-contour" indicator algol. This
problem. algorithm only has local
If the answer is positive, then influence. It requires O(R)
... operations, R being the number of
flagged nodes.
Stop the refinement of flagged
nodes and jump to algorithm algo6. It is obvious that the grid
generation as described is not
sensitive to the complexity of the
This algorithm consists of a geometry M to be treated. Generation
logical if-then structure and of a grid around a complex High-Lift
requires only testing of device is as easy as generating a
properties of Sgj. The number grid around a single airfoil.
operations required is negligible.
For what concerns the extension
algo4: a refinement algorithm for towards 3D applications, two
flagged nodes of grid G. algorithms need special attention
namely the "in-surface'' indicator,
If i and j are two flagged nodes algol, and the construction of Sgj,
of grid G a new node is inserted algo2.
between i and j. Further the For the "in-surface" indicator,
interconnection table of algol, the intersection of K
representation Ga of grid G is elementary triangular planes of the
updated. This algorithm has only 3D poly surface representation Spj
local influence and requires O(M) with L connections of representation
operations, if M is the number of G, of the grid G should be found.
flagged nodes. This can be done within the same
order of operations as for the 2D
algo5: a smoother. analogy of algol.
For the construction of Sgj, algo2,
Move all internal nodes i towards the same holds true.
the centre of gravity formed by
node i and its neighbours. 2.3 Grid generator DZGRID.
Algorithm algo5 helps to meet
requirement Rg3. Characteristic memory requirements
of the current two dimensional grid
Recursion of algorithms algol, generator DZGRID are:
algo2, algo3, algo4 and algo5 until
a jump is made to algo 6 results in reals per node: 5
a grid which is automatically integers per node: 18+2*KB
refined towards the contours,
without any user defined tuning. KB being the maximum number of
Hence requirement Rg2 finally will neighbours of a node i occurring in
also be met. The final grid is grid G. Typically KB equals 10.
obtained by applying algorithms
algo6 and algo7 the grid: Some run time characteristics of the
current two dimensional grid
algo6: a boundary reconstruction generator D2GRID are:
algorithm.
- The lowest grid level of the multi
Move grid nodes linked to the grid V cycle is generated in one
polygonal representation Sgj on to grid generator run.
the spline representation Ssj of
the segment Sj and give them the - Uniform refinement is used to
corresponding solver boundary obtain the subsequent finer grids
code. So the user has to determine of the multi grid V cycle. As a
in the geometry description file consequence the subsequent grids
for each segment Sj whether it of the multi grid V cycle are
concerns a part of a solid wall, a nested, such that no additional
symmetry wall, etc. This algorithm interpolation data is needed.
algo6 requires O ( S ) operations, S
being the number of nodes on the - For the NLR422 grid of Fig. 5 the
boundaries. grid generation took about 2000
CPU seconds on the NLR Cyber 9 6 2 .
algo7: a node killing algorithm.
Remove all nodes flagged by the
12-6

Conclusions:
un, + vn, = 0 (4)
- The memory r e q u i r e m e n t s d o n o t
form t h e b o t t l e neck f o r t h e
current g r i d generator. and t h e f a r f i e l d boundary c o n d i t i o n
- For t h r e e dimensional applications
a n e f f i c i e n t i m p l e m e n t a t i o n of t h e
g e o m e t r i c i n t e r s e c t i o n problem i s
important.
I n e q u a t i o n ( 4 ) nx and ny are t h e
- T h e v e c t o r i s a t i o n l e v e l of t h e C a r t e s i a n components o f t h e u n i t
algorithm i s l o w due t o t h e v e r y normal t o t h e s o l i d w a l l boundary.
d e e p IF-THEN s t r u c t u r e s . E q u a t i o n s ( 3 ) , ( 4 ) and ( 5 ) s h o u l d be
s o l v e d f o r t h e v e c t o r of p r i m i t i v e
variables <. In a shorter notation
3 . EULER SOLVER e q u a t i o n ( 3 ) reads as:
3.1 Solver algorithm.

The c u r r e n t f l o w s i m u l a t i o n s y s t e m
i s b a s e d on t h e u n s t e a d y E u l e r
e q u a t i o n s ( 3 ):
I n t h e grid generation phase t h e
f l o w domain i s d i v i d e d i n t o f i n i t e
volumes, on a n a v e r a g e h a v i n g a n
h e x a g o n a l s h a p e . An example i s g i v e n
i n Fig. 2 . I n each grid p o i n t i t h e
<
v e c t o r of p r i m i t i v e v a r i a b l e s is
unknown. I n F i g . 3 t h e g r i d a r o u n d
t h e s l a t o f F i g . 1 i s shown. T h e
g r i d p o i n t s are ( a p p r o x i m a t e l y )
lying i n t h e centres of t h e i r
1 s u r r o u n d i n g c o n t r o l volumes, h e n c e
t h e p r e s e n t method c a n be
c h a r a c t e r i s e d as a v e r t e x c e n t r e d
f i n i t e volume method. E x p r e s s i n g
t h a t equation ( 6 ) should hold within
e a c h f i n i t e volume n of t h e g r i d
yields:

(7)
(3) To o b t a i n t h e l e f t hand side o f
I n equation ( 3 ) p is t h e d e n s i t y , p e q u a t i o n ( 7 ) two a s s u m p t i o n s are
made :
i s t h e s t a t i c p r e s s u r e , U and v are
t h e C a r t e s i a n x and y components o f - t h e shape of t h e f i n i t e volumes
t h e v e l o c i t y v e c t o r i n t h e two d o e s n o t change i n t i m e .
d i m e n s i o n a l space and 7 i s t h e
specific heats ratio. Density,
- t h e unknowns are c o n s t a n t w i t h i n
t h e f i n i t e volume.
C a r t e s i a n v e l o c i t y v e c t o r components
and s t a t i c pressure form t h e v e c t o r The l a t t e r a s s u m p t i o n i s d u e t o o u r
of p r i m i t i v e v a r i a b l e s e, a vector i n t e r e s t i n t h e steady solution.
i n t h e four dimensional s o l u t i o n G e n e r a l l y , a s e c o n d order a c c u r a t e
s p a c e . E q u a t i o n ( 3 ) i s t o be d i s c r e t i s a t i o n o f t h e r i g h t hand
supplemented w i t h t h e perfect s l i p s i d e o f e q u a t i o n ( 7 ) i s o b t a i n e d by
boundary c o n d i t i o n on s o l i d w a l l s assuming a l i n e a r s p a t i a l v a r i a t i o n
of t h e primitive v a r i a b l e s vector <
w i t h i n t h e f i n i t e volume. T h i s i s
 12-7

p r a c t i c a l l y a c h i e v e d by r e p r e s e n t i n g The u n d e r l y i n g idea of t h i s a p p r o a c h
each component tk o f t h e v e c t o r of i s t h a t d i s c o n t i n u i t i e s ( s u c h as
p r i m i t i v e v a r i a b l e s by a p l a n e i n s h o c k s f o r i n s t a n c e ) a r e located on
t h e x,ylckspace. This plane is t h e i n t e r f a c e s between t h e f i n i t e
c o m p l e t e l y d e t e r m i n e d by t h e volumes of t h e g r i d , and a r e n o t
component ck of t h e v e c t o r of a l l o w e d t o cross t h e f i n i t e volumes
( t h e Gauss theorem i s n o t a p p l i c a b l e
primitive variables < i n t h e centre i n case d i s c o n t i n u i t i e s cross t h e
o f t h e f i n i t e volume and t h e k t h f i n i t e volume).
component o f t h e u n i q u e s p a t i a l Applying t h e Gauss theorem t o t h e
g r a d i e n t v e c t o r of t h e p r i m i t i v e r i g h t hand s i d e o f e q u a t i o n ( 7 ) it
variables vector < in the finite transforms t o
volume. So t h e p r i m i t i v e v a r i a b l e s
vector is defined within t h e f i n i t e
volume by e q u a t i o n ( 8 ) :
s l i n i= -
dt
1
Si
(Fn, + Gn,) dS

T h i s e q u a t i o n is shortened as

T h e s p a t i a l g r a d i e n t v e c t o r of t h e
primitive variables vector is < d!: = -
n

J N ~ S
unique s i n c e t h e l i m i t i n g i n t h e dt
p r e s e n t scheme i s a p p l i e d d u r i n g t h e Si
approximation of t h e s p a t i a l
g r a d i e n t vector of t h e p r i m i t i v e w i t h N t h e normal f l u x a t t h e
variables vector <.
An example f o r i n t e r f a c e S of t h e f i n i t e volume.
o n e component of t h e s p a t i a l E q u a t i o n (12) c a n b e i n t e g r a t e d i n
gradient vector is given i n equation t i m e by a s t a n d a r d f o u r s t a g e Runge
(9): K u t t a l o c a l t i m e s t e p p i n g method
(see f o r i n s t a n c e Ref. [ 9 ] ) .
Convergence a c c e l e r a t i o n c a n be done
by a F u l l M u l t i G r i d a c c e l e r a t i o n
t e c h n i q u e based on d e f e c t
correction. W e w i l l further
c o n c e n t r a t e on t h e a p p r o x i m a t i o n o f
E q u a t i o n ( 9 ) i s t h e r e s u l t of t h e t h e i n t e g r a l i n t h e r i g h t hand s i d e
Gauss theorem a p p l i e d t o t h e f i n i t e of e q u a t i o n ( 1 2 ) . For each p a r t of
volume a r o u n d node i. Si i s t h e t h e c e l l s u r f a c e S t h e normal f l u x N
s h o u l d be c a l c u l a t e d . C o n s i d e r t h e
s u r f a c e of t h e volume ni, nx and ny c e l l i n t e r f a c e Sij between nodes i
a r e t h e C a r t e s i a n components of t h e and j. L e t N i be t h e normal f l u x
u n i t y outward normal on t h e s u r f a c e b a s e d on t h e p r i m i t i v e v a r i a b l e s
S and <Lim,u,r. i s t h e l i m i t e d vector t h e primitive variables
e x t r a p o l a t i o n of t h e p r i m i t i v e vector e x t r a p o l a t e d from i towards
variables vector < towards t h e
t h e c e l l i n t e r f a c e Sij w i t h e q u a t i o n
s u r f a c e Si. I f t h e n e i g h b o u r s o f node
(8):
i are c a l l e d j, e q u a t i o n ( 9 ) c a n b e
made d i s c r e t e :

A s i m i l a r d e f i n i t i o n h o l d s f o r Nj.
W i t h i n a n upwind f l u x d i f f e r e n c e
scheme t h e normal f l u x Nij a t t h e
c e l l i n t e r f a c e Sij i s d e f i n e d as:
1 ( xj-xi
c
j (xi + 7
(10)
LIM i n e q u a t i o n ( 1 0 ) s t a n d s f o r a
l i m i t e r , i n t h e p r e s e n t work t h e Van
Albada l i m i t e r w a s u s e d ( R e f . [ 8 ] ) .
12-8

Euler solver are

- More than 4 orders of
magnitude reduction in
maximum residual after 1000
multi grid V cycles for a
three level 45512 nodes grid
around a three element
In equation (14) 1A1 is the airfoil.
eigenvalue matrix of the discrete
Jacobian of the normal flux with - 74 p-sec per multi grid V
respect to the primitive variables cycle and per node of the
vector <.
R and L are the right and top level grid of the multi
left eigenvector matrices of the grid cycle.
same Jacobian. The discrete Jacobian
of the current solver is based on - Vector operation ratio on
the polynomial character of the the NLR NEC SX3-12 99.5%
components of the normal flux with
respect to the primitive variables - Vector length 127.6
Ref. [lo].
Equation (13) clearly shows that the - 573 MFLOPS ( FLoat ing point
present scheme goes back to Van Leer Operations per Second)
‘s MUSCL scheme [ l l ] such that this.
scheme can be expected as being Conclusions can be drawn from these
monotone and second order accurate f igures:
in space.
The use of 2D unstructured
3.2 Solver D2EUL grids requires 47 percent
more storage of reals and
Some memory requirements of the 1000 percent more storage of
current two dimensional multi grid integers. The use of 3D
Euler solver are unstructured grids requires
143 percent more storage of
reals per node: 47 + 4*KB reals and 3333 percent more
integers per node: 3 + 3*KB storage of integers.

KB being the maximum number of KB should be as low as

neighbours of a node i occurring in possible. The grid generator
grid G. Typically KB is equal to 10. can be extended with a post
For the case of a structured grid 2D processing algorithm that
Euler solver with the same reduces KB towards the
characteristics (same scheme, same natural average of 6 in two
vector computer, same compromises dimensions and 12 in three
between speed and memory): dimensions. The percentages
would respectively be 20,
reals per node: 47 + 3*KB 600, 54 and 1600 percent.
integers per node: 3 The lowest possible KB
occurs in case the control
In this case KB is equal to 4. volumes are chosen
For a 3D multi grid Euler solver the triangular.
following would be found:
The largest increase of
reals per node: 64 + 6*KB memory use is due to the
integers per node: 3 + 4*KB storage of interconnections
since the interconnections
Typically KB is now equal to 25. For are no longer known a
the case of a structured 3D Euler priori. This increase should
solver with the same be payed off by the
. characteristics: advantages of using
unstructured grids, mainly
reals per node: 64 + 4*KB by the automation of the
integers per node: 3 grid generation.

In the latter case KB is equal to 6.

4. APPLICATIONS
Some run time characteristics of the
current two dimensional multi grid 4.1 Casel.
 12-9

The computational grid has 10433 potential method, in which the slat
nodes. The Mach number is 0.85 and and wing cove shape are changed (by
the angle of attack is 1 degree. The an iterative method) such that the
Mach number distribution on the pressure coefficient at the bubble
surface is shown in Fig. 4. It can edges is constant Ref. (12). It is
be seen that the scheme is monotone observed that the solution on the
in the neighbourhood of the shocks fine grid matches better the
and that it captures the shocks with potential solution than the solution
one interior node. The present on the coarse mesh. The streamline
solution is compared with Ref. [13]. following the bubble edge of the
The Mach number distribution is very slat cove in the potential
sensitive to the total pressure calculation is plotted in Fig. 12,
losses due to the dissipative together with the is0 total pressure
character of the scheme. Once loss lines, found on the coarse
generated (at the nose of the grid. Fig. 13 shows the wing cove.
profile) the total pressure losses It can be seen that for the slat
are convected downstream, causing a cove the bubble edge found with the
uniform difference between the two potential method does not coincide
Mach number distributions. with the edge of the bubble found
with the Euler method. For the wing
4.2 Case2. cove the agreement is much better.
Fig. 14 and 15 are the corresponding
Two grids around this airfoil are figures in case of the fine grid. It
used. A coarse grid with 11673 nodes can be observed that the level of
and one global refinement of this the total pressure losses diminishes
coarse grid with 45512 nodes. The on the finer grid, except in the
Mach number at infinity is 0.20 and coves where the peak value increases
the angle of attack is 10 degrees. A and the shape of the isolines
global view of the fine grid is changes.
shown in Fig. 5. A detail of the
coarse grid around the slat has been 4.3 Case3.
shown in Fig. 3. A summary of the
convergence history is given in The grid used is very similar to the
Figs. 6 and 7. Figure 6 shows the coarse grid around NLR422 and
maximum residual of the mass contains 15351 nodes. The Mach
equation occurring in the whole flow number at infinity is 0.22 and the
field. It can be seen that the slope angle of attack ranges from -2.5 to
of the curve deteriorates going from 25 degrees. Fig. 16 shows the cl -
the coarse to the fine grid, for alpha curve for M = 0.22, Fig. 17
both the single grid and the multi the cd -alpha curve and Fig. 18 the
grid method. Far more iterations are cm - alpha curve. Due to the effect
needed to converge in the single of the dissipative character of the
grid method case. Figure 7 shows the scheme the calculated curves fit
lift coefficient history. The lift relatively well to the measured
coefficient is slightly higher for curves. Figs. 19 and 20 give the
the fine grid, indicating that in numerical and experimental values of
principle another refinement of the the pressure defect at the slat and
grid is needed. For the fine grid flap cove rakes for 9 degrees
the Mach number distribution around incidence. Due to the perfect slip
the whole configuration is shown in boundary condition of the Euler
Fig. 8. Several dips in the is0 Mach equations the pressure defect
number lines near to the surface can decreases toward the solid wall. The
be seen. These are due to the total peak level of the pressure defect
pressure losses to which the Mach however corresponds reasonable with
number is sensitive. The total the experiments.
pressure losses are caused by the
dissipative character of the scheme
used (mainly the limiting of the CONCLUSIONS
scheme, reducing the scheme locally
to a first order accurate scheme). It is shown that the described flow
The pressure coefficient simulation system is capable to
distributions around the slat, wing successfully handle complex 2D
and flap are shown in Figs. 9, 10 geometries such as encountered in
and 11. In these figures a High-Lift systems. The main
comparison is made between the characteristic of the unstructured
current solutions on the coarse and grid flow simulation system is the
fine grid and the solution of a automation of the grid generation
12-10

procedure. This paper illustrates 'An automatic euler solver

the application of unstructured grid using the unstructured
generation to solve the Euler upwind method.', Computers
equations for complex geometries in and Fluids, Vol. 19, nr.
two dimensions. For a typical three 34, pp. 273-286, 1991.
element test case with 16000 nodes
the total grid generation plus 5. Shamos, M.I. and Hoey, D.,
solver turn-around-time of the 'Geometric intersection
current flow simulation system is problems', in 17th Annual
approximately 2 hours. The Symposium on Foundations
computational results are compared of Computer Science, IEEE,
with experimental results and show 1976.
qualitative agreement, due to the
dissipative behaviour of the 6. Milgram, M.S., 'Does a
solution of the discrete Euler point lie inside a
equations. A future critical issue polygon', J. Compt. Phys.
is to account for the dissipative 84, 1989.
effects in a proper way. It requires
an extension of the mathematical 7. Shimrat, M., Commun. ACM,
model to Reynolds averaged Navier pp.434, 606 ,1962
Stokes equations.
8. Van Albada, G.D., Van
Leer, B. , Roberts, W.W. ,
ACKNOWLEDGEMENTS 'A comparative Study of
Computational Methods in
The author wishes to thank Dr. B. Cosmic Gas dynamics',
Oskam for the fruitful discussions Astron. Astrophys. 108,
and the ideas about the applications 76-84 , 1982.
of the analysis method, Ir. P.
Termes for the comparison between 9. Batina, J.T., 'Accuracy of
calculations and GARTEUR AD (AG08) an Unstructured-Grid
results and the members of the Upwind-Euler Algorithm for
GARTEUR AD (AG08) for their the ONERA M6 Wing', J.
permission to publish results Aircraft, Vol. 28, NO. 6 ,
obtained within the action groups. 1991.
10. Dick, E., 'Multigrid
REFERENCES formulation of
polynomial flux-
1. Haase, W. , Brandsma, F. , difference splitting
Helsholz, E., Leschziner, for steady Euler
M., Schwamborn (Eds.), equations', J. Comput.
'EUROVAL - A European Phys. I 1, 161-174, 1991.
Initiative on Validation
of CFD codes', notes on 11. Van Leer, B., 'Towards the
numerical fluid dynamics ultimate conservation
volume XX, to be published difference scheme VI a
by Vieweg. second order sequel to
Godunov ' 8 method', J.
2. De Cock, K., 'Effect of Comput. Phys., 32, 101-
wing sweep on the solution 136, 1979.
of the continuous and
discretized euler 12. Oskam, B., Laan, D.J.,
equations for a High-Lift Volkers D.F., 'Recent
configuration',NLR CR advances in computational
91176, 1991. methods to solve the high-
lift multi-component
3. Mavriplis, D.J., airfoil problem', AGARD
Martinelli, L., 'Multi Conference Proceedings,
grid solution of No. 365 , 1984.
compressible turbulent
flow on unstructured 13. 'Test Cases for Inviscid
meshes using a two- Flow Field Methods', AGARD
equation model', ICASE Advisory Report, No.211,
Report No. 91-11, 1991. 1985.
4. Nakahashi, K., Egami, K.,
 12-11

Fig. I Three element airfoil NLR422 (case2)

Fig. 2 Finite wlumes around the slat of the NLR422

three element airfoil (case 2)
12-12

e
.........
...........
. .<a.?
'i:
......-
.
-
........
.:........
.<.e?;..%
..........
SF:;:;;:.
U&$:::.
vr.,-s>.
b
&
.>
:..
_.,
;
:
f
:
3
F:
<
A
.
A.
h

...--..,
pi
....... ~ ..... 5

Fig. 3 Grid around the slat of the NLR422

three element ahfoil (case 2)

U 1.500~
CURRENT FLOW SlMUVinON SYSTEM
REF. 1131
-
--*--
w
m
5
1.250
I

1.000

0.750

0.500

0.250

0.000
0.000
i 0.200 0.400

Fig. 4 Mach numhr distribution around the NACAOO12

0.600 0.800
WC
1.000

single aidoil (case 1)

 12-13

Fig. 5 G M view of the fine grid around the NLR422

three element airfoil (case2)

coarse grid (multi grid)

------ CW grid (multi grid)
coarse grid (single grid)
1-r (single grid)
M.R.M.E.

,o-lol work unils

t I
I
I I I I
I
I
i
0 4000 8000 12000 16000 20000 24000 28000

Fig. 6 Maximum residual of the mass equation (case 2)

12-14

COARSEGRID ( M u m GRID) -
FINE GRID (MULTI GRID) _-____
cL6.00/
5.00

0.00
, WORK UNITS
I

0 2000 4000 6000

fig. 7 Lift coefficient history (case 2)

Fig. 8 Mach number distribution around the NLR422

three element airfoil (case 2)
 -
12-15

CP - X UNSTRUCTURED COARSE GRID

CP-XPOTENllALFLOW(l1)

CP __
4
4

-2.001
I I I x/C I
-0.200 -0.100 -0.000 0.100 0.200

Fig. 9 Pressures coefficient disbibution on the slat of the NLR422

three element aidoil (case 2)

5.00-
CP - X UNSTRUCTURED COARSE QRlD
CP. x POTENTIAL n
ow (11)
-
--*--
CP
4.00--

3.00--

2.00--

1.oo--

-1.oo
WC
0.000 0.200 0.400 0.600 0.800 1.000

Fig. 10 Pressure coefficientdistribution on the wng of the NLR422

three element airfoil (case 2)
12-16

CP - X UNSTRUCTURED COARSE GRID -

---*--

3
CP-XPOrrNTlALFLW(1i)

CP

1 .oo

0.00

-1.oo

-2.00 1 WC
I I I I
I
0.800 0.900 1.000 1.100 1.200
Fig. 11 Pressure coefficient distribution on the flap of the NLR422
three element airfoil (case 2)

Fig. 12 Is0 total pressure loss lines amund the slat cove of the NLR422
three element airfoil: warse grid (case 2)
 12-17

Fig. 13 Is0 total pressure loss lines around the Wing cove of the NLR422
three element airfoil: coarse grid (case 2)

Fig. 14 Is0 total pressure loss lines around the slat cove of the NLR422
three element airfoil: fine grid (case 2)
12-18

Hg. 15 Is0 total pressure loss lines around fhe wing cove of the NLR422
three element aitfoil: fine grid ( w e 2)

I 0 EULER FREEFLOW 1
I A EULER TUNWALL I
MACH =.a0
REYN =4.00

-
EULEWMEASURED RESULTS

aa =5
ALPHA [DEGREES]

Fig. 16 cl - a diagram for the MI three element aitfoil (case 3)

 12-19

MACH =.220
REYN -4.00

EULEWMEASURED RESULTS

Act
-
=5
ALPHA [DEGREES]

Fig. 17 cd - a diagram for the M1 three elemeflt airfoil (case 3)

CM

Aa =5
4 ALPHA [DEGREES]

fig. 18 cm - a diagram for the M1 three element airfoil (case 3)

12-20

1.000 q-
1
7 UNSTRUCTUREDGRID CALCULATIONS
EXPERIMENTS I

RAKE ORIENTATION

o.200
0.000
1
I I
I
I
I
I I I
o\o

I
I
4.400 -0.200 0.000 0.200 X COORDINATE

Fii. 19 Pressure defect at the slat cove rake holes of the MI three element airfoil

0.600 -
t; - UNSTRUCTURED GRID CALCUIATIONS
fW 0 EXPERIMENTS
O
$ 0.500 --
3
v)
v)
W
0.400 --

0.300 --

0.200 --

0.100 --

0.000 ' 0
"
e

I I I
I
I
I
I 1 I
1

0.74500 0.74700 0.74900 0.75100 X COORDINATE

Fig. 20 Pressure defect at the flap cove rake holes of the M1 three element airfoil
 13-1

PREDICTION OF THE HIGH-LIFTPERFORMANCE OF

MULTI-ELEMENT AEROFOILS USING AN
UNSTRUCTURED NAVIER-STOKES SOLVER

anaV dQcas: CRS4, Vh N d o SUUD 10.

09123 cqlimi, Ildy
13-2
 13-3

dV

'XY
13-4

PI - cp P
k'
-
c

cc, - 1 . 4 4 , cc, - 1.92

 13-5

K - 0.41 , E - 9
13-6

. k&e.

(41)
 13-7

d!k
- I
k!!

(53)
13-8
13-9
13-10

13. w, DA. 'An hnkdd Anlyd. d 'haodc, s1*sd

Ahfdb', AUA Papa 74S41. 1974
13-11
13-12

Pip. 1
-
X cell centres; -
Notation for computational cella and c e l l edges
cell vertices
pip. 2 Auxiliary cell ( - - - - -)
for viscous flux computation
-

w c w c IWK
0.050 Y/C. 0.4w 0.09 XZ. 0.57:
CP
, 0. os0

J
-1.6

0.025 0. ox 0.025

0.000 0. MXI 0. ow
0.0 0.5 1.0 1.5 0.0 0.5 1.0 1.5 0.0 0.5 1.0 1.5
w c UN
. nit WUI mK Ulur
0. OM

0.025

1.4 1 0. WO
0.0 0.5 1.0 1.5 0.0 0.5 1.0 1.5 0.0 0.5 1.0 1.5
w k wu, Lvl!w

(a) Surface pressure distribution (b) Upper surface mean-velocity profiles

- 9 - Effect of near-wall mdclling

Pig. 3 RAE 2822 aerofoil
o,A experiment; - Case
one-equation model
- - - - - - wall functions
.
 13-13

Fig. 4 Computational.grid for NLR 7301 wingltrailing-edge

flap configuration

21

( a ) a = 6.0"

-12

-10

-8

-6

-2

(b) a - 13.1'

Fig. 5 Surface prcssure distributions for NLR 7301 winglflap

0 4 experiment; - computation
13-14

(a) a = 6.0'

./e - 0.050

1 WlI

(b) a - 13.1O

Fig. 6
-
Upper surface mean-velocity ptofilcs for Nu( 7301 winglfiap
o expcriment; computation
 13-15

x
X

(s) Lift vs inoidcncc

IO.

5 x

"
B.
.. 0

e,
al.
ao as 1.0 1.5 2.0 2.5 10 ss 4.0

(b) Drag vs lift (c) Pitshing m m e n t vs lift

Pig. 7 Integrated loads for NLR 7301 -ving/flap

o experiment: X computation
 13-16

Fig. 8 Geometry and inner region of computational grid for

SKP 1.1 acrofoil/flap configuration

b
2.O 1

a5

1.0

1.5 1

(a) M, = 0.600. a = 0 . 1 8 O . run 223 (b) M, - 0.650, a = O.OZD, run 229

1.5 1 I
(0) = 0.701, a - -0.1So,run 235

Fig. 9 Surface pressure distributions for SKP 1.1 winglflap

Mach number sweep a t a 1 0 '
o,A experiment;- computation a t nominal M,
______ computation including MI,
 13-17

2.0

CL

1.5

1.0

0.5

n
0.0 4

0.5 0.6 0.7 0. a

(a) Lift vs Mach number

-0. 1 I
I
0. 6 0. 7 0.8 I
n

.o. 2
X
O O
Y
x x 0

05 .o. 3 x

, 0

X
X x
.o. 4

n
-" ,I.

,oo .o. 5
0.5 0. 6 0. 7 0. a

(h) Drag VI Mach number (c) Pitching mOmcnt VI Mach number

P i g . 10 Intcgratcd loads for SKP 1.1 winglflnp

Mach number sweep a t II L oo
o experiment; X computation
13-18

(a) Mach number contours ( * indicates sonic line)

- A,-

----/
AA

-A
- 4 -

A-. A-. -_ .- .-

(h) Velocity vectors in flap covc region

Fig. 1 1 Typical computed flowfield solution for SKF 1.1 winglflap

M, = 0.701, D = -0.15', run 235
 14-1

NUPIERICN, CALCuLhTIONS OF EIGB LIFT PLOWS

USING STWCRIRED AND UNSTRUCTURED m 0 D S
R. Bailey, J.M.A. Longo
R. Radespiel. A. Ronzheimer
A. Demier, N. Kroll
C.-C. ROSSOW
Institute for Design Aerodynamics
DLR-Braunschweig. Am Plughafen
0-3300 Braunschweig. Germany

different inviscid and viscous flow regions

(see e.g. [ l l . 121). In particular. the
At the DLR Institute for Design Aerodyna- existence of significant regions of separa-
mics current research in the area of high ted flow and of strong wake/houndary layer
lift aerodynamics is directed towards the interference distinguishes the high lift
development of a computational analysis airfoil problem from the aerodynamic issues
capability for high lift systems. The heing faced at cruise conditions. The suc-
Navier-Stokes equations are solved with a cessful design of future efficient high
multigrid method based on central spatial lift devices will largely rely on a better
differencing and Runge-Kutta time stepping. understanding of the basic physics. TO
Two particular problems are addressed in achieve this, both improved experimental
this paper. The first concerns the calcula- and theoretical prediction techniques are
tion of maximum lift for a single airfoil required.
and a clean wing configuration. The accura-
cy of the basic structured flow solver is At the DLR Institute for Design Aerodyna-
investigated by comparing the results with mics a numerical procedure for analyzing
experimental data for two test problems and high lift configurations is currently heing
several flow conditions. mphasis is placed under development. The method is based on
on the sensitivity of the computed solution the solution of the mass-averaged Navier-
associated with turbulence modelling. The Stokes equations using a central spatial
second aspect dealt with in the present discretization and Runge-KUtta time step-
paper concerns the extension of the numeri- ping scheme. Rapid convergence is achieved
cal method to multi-element airfoils. Both. by the use of various acceleration tech-
the block-structured and the unstrucutred niques. in particular a multigrid algo-
grid approach are investigated in order to rithm. The basic solver has been applied to
explore their specific merits and 1 M t a - a variety of flow prohlems including finite
tions. Detailed comparisons of the struc- span wings and wing-body combinations
tured and unstructured approach are Pre- [3-51.
sented for low Reynolds' number laminar
viscous flows around a single airfoil, and In the present paper two particular aspects
for the inviscid flow around a multi- of the code development for high lift con-
element airfoil. figurations are addressed. Firstly. the
capability of the Navier-Stokes solver to
LIST OF SYMBOLS predict the maximum lift of a single air-
foil and a wing is investigated. For this
analysis a structured grid approach has
n' normal force coefficient been used. Comparisons of theoretical re-
cl,cL lift coefficient sults obtained with different turbulence
models and experimental data are presented.
cd,cD drag coefficient secondly, the extension of the basic flow
C surface pressure distribution solver to treat multi-element airfoils is
P discussed. Although substantial progress
E specific total energy
- has been made in the development of effec-
..
P
n
tensor of flux density
unit vector of outward normal
tive algorithms for solving the Navier-
stokes equations. very often the uae of
these methods for practical applications is
M Mach number limited due to deficiencies in grid genera-
tion. Today the generation of an appro-
u.v.w Cartesian velocity components priate grid around complex aerodynamic
Re Reynolds number geometries such as high lift configurations
continues to be a major obstacle. For the
V control volume
treatment of multi-element airfoils mainly
av boundary of control volume V two approaches have been used in the lit-
erature, namely the block structured grid
a angle of attack concept (see e.g. [ 6 1 ) and the unstructured
? spanwise section mesh approach (see e.g. [711. At DLR cur-
rently both approaches are being investiga-
P density ted in order to explore and assess their
merits and limits. First results of these
investigations are shown here. The studies
presented in this paper appear to be neces-
sary steps towards the developoment of an
1. INTRODUCTION accurate and efficient computational proce-
dure for high lift configurations.
In the developlllent of new aircraft the
design of high lift system for l o w speed
take-off and landing capabilities is Per-
haps one of the most complicated aerodyna-
2. DESCRIPTION OF NUMERICAL METHODS 1
mic problems. The flow around high lift The governing equations are the mass-
configurations is characterized by many
14-2

averaged Navier-Stokes equations repre- have to be added explicitly in order to

senting Conservation of mass, momentum and prevent odd-even point &coupling and os-
energy. Expressed in integral form they can cillations near shock waves. The artificial
be written as dissipation model used in the structured
code is based on the work of Jameson et al.
a
- ddv+ F < d S = O . (1)
1131. A blend of fourth and second dif-
at ferences of the conserved variables is used
V av to provide third order background dissipa-
tion in smooth flow regions, and first-
where order dissipation at shock waves. In order
W = [P. PU. PV. PW, PEIT (2) to avoid excessively large dissipation for
cells with high aspect ratios present in
is the vector of conserved variables. The the boundary layer, a variable scaling
quantities p , U, v, w and E denote the factor of the dissipative terms is em-
density, the Cartesian velocity components ployed. According to [I41 this scaling
and the specific total energy. V represents factor is a function of the spectral radii
an arbitrary control volume with boundary of the Jacobian matrices associated with
a V and outer normal ii. The definition of the three directions of the body fitted
the flux density tensor P is given in [31. coordinate system, and accounts for varying
Tha equation of state for perfect gas is cell aspect ratio.
used to calculate pressure and temperature. The spatial discretization results in a
The laminar viscosity is assumed to follow system of ordinary differential equations
an empirical power law. In the case of in time which is advanced in time by an
turbulent flows the laminar viscosity, p , explicit multistage scheme. For computa-
is replaced by p+v and the heat conducti- tional efficiency, the physical viscous
vity p/Pr is r e p t k e d by p/Pr + p /Prt, terms are computed on the first stage only
where the eddy viscosity pt, and the ‘turbu- and they are frozen for the remaining
lent Prandtl number, Prt are provided by a StageS. Since current interest is focussed
turbulence model. In the present work the on steady flow fields. various techniques
turbulence models of Baldwin-Lomax [81 and are used to accelerate the convergence to
Johnson-King 191 are used. The Baldwin- steady state. With local time stepping the
LOmaX model is an algebraic model which maximum allowable time step is used for
relates the eddy viscosity to the local each cell to advance the solution in time.
maximum of vorticity. This equilibrium Implicit smoothing of the residuals is
model is widely used for computing attached employed to extend the stability region of
and mildly separated flows. In order to the basic explicit multistage scheme. Using
allow a more accurate determination of the variable smoothing coefficients as a func-
turbulent stresses in separated boundary tion of the cell aspect ratio and the time
layer flows, Johnson and King developed an step can considerably improve the damping
non-equilibrium turbulence model in which properties of the scheme for high-aspect-
both. convection and diffusion of turbu- ratio cells L31. Finally, a multigrid
lence are taken into account. method according to [151 is implemented. A
sequence of successively coarser meshes is
2.1 Structured Method used to efficiently damp out errors
throughout the frequency spectrum simulta-
The basic flow solver (DLR-code CEVCATS) neously. Due to the less restrictive time
13.10.111 is a finite volume method based step limitations and the smaller number of
on structured grids. The numerical approxi- p i n t s on the coarse meshes. the solution
mation of ECI (1) follows the method of can be advanced much more rapidly 131.
lines, which decouples the discretization
in space and time. The physical domain The code CEVCATS allows multiblock decompo-
around the aerodynarmc body is divided into sition of the computational domain Ill1
hexahedral cells by the generation of a which is important to treat complex air-
body-fitted grid. The discrete values of craft geometries such as high lift systems.
the flow quantities are located at the Considerable effort has been made to deve-
Vertices of the cmputational mesh cells. lop efficient multigrid strategies for
The cells surrounding a vertex form a super multiblock computations 1161.
cell. The rate of change of mass, momentum
and energy associated with a cell vertex is 2.2 Unstructured Method
obtained by summing up the fluxes through
the outer boundary of the super cell. The In order to explore the capabilities and
flux through a boundary segment is obtained limits of the unstructured grid approach,
by averaging the quantities at its corner the basic solution method described above
points. It can be shown [121 that this cell has been extended to allow computations on
vertex discretization scheme is first-order unstructured meshes. Following the work of
accurate if the distribution of the seg- Mavripilis 117.181 a two-dimensional Euler/
ment-normal vector is smooth over the seg- Navier-Stokes code has been developed
ment. and if the cell faces do not degene- [19.201 which uses control volumes formed
rate to triangles. On smoothly varying by triangular elements (see Fig lb). The
meshes the discretization is second-order numerical scheme for the solution of the
accurate. The viscous fluxes contain first full mass-averaged Navier-stokes equations
derivatives of the flow variables. They are uses a spatial discretization employing a
computed using a local transformation from Galerkin finite element formulation. De-
Cartesian coordinates to the curvilinear tails are given in 1201. The process is
coordinates. Details are given in [31. In
the present CO& the thin layer approxima- equivalent to a finite volume approximation
tion is employed in which the viscous for regular triangular meshes. AS in the
fluxes take into account gradients in the structured algorithm artificially construc-
direction normal to the body and in span- ted dissipative terms have to be added to
wise direction only. provide background dissipation and to allow
shock resolution. In the case of unstruc-
Since the spatial discretization is based tured meshes the artificial dissipative
on central differencing, dissipative terms operator is constructed as a blend of an
 --

e 14-3

undivided Laplacian and a biharmonic opera- measurements is obtained for all Reynolds
tor 1171 . For Navier-Stokes calculations number and Mach number combinations. In the
the dissipation model is modified according vicinity of the normal force maximum the
to Mavripilis 1181 in order to take into theoretical results obtained with the
account the high aspect-ratio cells present Johnson-King model show discrepancies to-
in the boundary layer and wake regions. For wards higher drag coefficients as compared
advancing the solution with respect to to the experimental data. It can be said
time, an explicit five stage Runge-Kutta that considerable differences between com-
scheme is used. An in the structured scheme putations and measurements become apparent
convergence acceleration is achieved when substantial flow separation occurs.
through local time stepping, implicit resi- The best agreement between theory and ex-
dual smoothing, and a multigrid method. In periment is obtained for M, = 0.3 for the
order to improve the accuracy and efficien- low Reynolds number. This may be due to the
cy of the numerical procedure, a local grid fact that for these conditions only small
adaptation procedure has been implemented. domains of flow separation exist. The same
Important flow features are w e l l defined trend can be seen in Fig 4 where the normal
and the overall efficiency of the unstruc- force coefficient 1s plotted versus the
tured scheme is not adversely effected by angle of attack. In this figure also ex-
the adaptation procedure. The adaptation perimental data with wind tunnel correc-
technique has been originally developed for tions [221 are displayed. Fig,5 shows the
application to Euler solutions 1191 and has convergence history for some selected cal-
to be extended to be applied efficiently to culations. For low angle of attack around
viscous flows. A detailed description of 80 multigrid cycles are required to achieve
the current implementation of the 2-D un- a residual reduction of 4-5 orders of mag-
structured Navier-Stokes solver is given in nitude. Note that with respect to the lift
1201. coefficient sufficiently converged solu-
tions are obtained within 40 multigrid
cycles. In the vicinity of maximum lift the
3. COMPUTATION OF MAXIHUM LIFT efficiency of the Navier-Stokes solver
somewhat decreases.
The structured Navier-Stokes Solver des-
cribed here has been applied to a wide 3.2 Wing-Body Combination
variety of two- and three-dimensional
flows. This paper addresses the capability The second test case used in this analysis
of the Navier-Stokes solver to predict is the DLR-F4 wing-body combination con-
maximum lift. Since so far no turbulence sisting of a transonic wing of high aspect
model has been implemented in the unstruc- ratio and a fuselage of Airbus type. The
tured solver, this investigation is re- DLR-F4 wing-body configuration was used
stricted to the structured algorithm. within GARTEUR for experimental studies of
3-D transonic flow fields. An extensive
3.1 NACA 0012 Airfoil data base of surface pressure distributions
and total forces is available for a range
As a first test case the two-dimensional of transonic Mach numbers and angles of
flow around the NACA 0012 airfoil has been attack. Fig 6 shows the geometry of the
calcyated for two geynolds numbers Re = configuration [231. The computational grid
3.10 and Re = 9.10 and for Mach numbers generated by transfinite interpolation is
between M, = 0.3, and M, = 0.7 1211. For of C-H topology. that is C-type in stream-
these flow conditions experimental results wise direction and H-type in spanwise di-
of Harris 1221 exist. For the computation a rection. The grid contains a total of 259
C-type grid with 385x65 grid points has points in the streamwise direction. 59
been used. Fig 2 shows the grid structure pOints in the direction normal to the wing
in the vicinity of the airfoil. Two slight- surface and 59 points in spanwise direc-
ly different grids have been generated for tion. The clustering of the grid lines in
the two Reynolds numbers. While the distri- direction normal to the surface is suffi-
bution of the grid points along the surface cient to produce the non-dimensional wall
remains constant p. both case- the first
spac ng is 1.5.10- c and l*lO’*c for Re =
coordinate y+ to be less than unity in the
cells adjacent to the solid surface.
3.10‘ and Re = 9.106, respectively. where c
denotes the chord length of the airfoil. As This configuration has been analyzed exten-
in the experiment the transition from lami- sively using the 3-D structured Navier-
nar to turbulent has been fixed to 5% of Stokes solver, and detailed results are
chord length in all calculations. For tur- given in 151. Comparisons of computed and
bulent flow computations the Johnson-King measured total lift coefficients Versus
turbulence model has been used. angles of attack for three Mach numbers.
&= 0.6, &= 0.75, and M, = 8 are presented
in Pig 7. The Reynolds &umber based on the
Fig 3 shows the computed drag ?lass
&= 0.3 and M,=
for
0.55 at Re = 3.10 (see Fig mean chord is Re = 3.10 . Calculations have
been performed with three different turbu-
3al as w e p as for M,= 0.55 and M,= 0.7 at
Re = 9.10 (see Fig 3b). The experimental lence models. the original Baldwin-Lomax
results of Harris are also displayed. It model, a modified Baldwin-Lomax model ac-
has to be mentioned that in order to allow cording to Kays 1241, and the Johnson-King
a comparison with experimental data the model. In the computation transition was
normal force coefficient has been plotted fixed on the upper wing surface at 15% of
instead of the more usual lift coefficient. the local chord and on the lower surface at
The normal force coefficient cn is defined 25% chord. on the upper surface of the wing
as at the wing tip and the wing body junction
cn = c1 cos a + cd sin a , (2) as well as for the whole fuselage the flow
was considered fully turbulent.
denote the angle of attack,
where a’ coef cient, and the drag coeffi-
the lift The agreement is Quite satisfactory even
cient, respectively. At lower angles of beyond the onset of buffet. Up to buffet
attack good agreement of computations and onset no difference can be found between
14-4

the lift curves computed using the Johnson- Navier-Stokes method for predicting laminar
King model, the Balwin-Lorn model or the viscous flows is investigated. For a low
modified Baldwin-Lomax model. However, at Reynolds’ number viscous flow around the
very large angle of attack only results single element NACA 0012 airfoil the effect
obtained with the Johnson-King turbulence of the grid topology upon the predicted
model compare well with the experimental solution is examined. The computations have
data. Both, the original and the modified been performed for a flow at Mach 0.8 with
Baldwin-Lomax model yield considerably the airfoil at an incidence of 10.. The
higher lift values. For the case with free- Reynolds‘ number is 500. Three separate
stream Mach number M-= 0.75, Pig 7 indi- unstructured grids have been constructed
cates the effect of fully turbulent flow on for this analysis. The solutions are com-
the wing upper surface. Instead of the pared with results obtained using the
nonliriear lift C U N e which has been ob- proven structured method described above.
tained with fixed transition at 15% chord, In the following we will refer to the three
the fully turbulent calculation yields an unstructured grids as cases T1, T2 and T3.
a m s t constant lift slope. In Pig 8 the respectively. All of the meshes have been
calculated and measured drag w l a r s are derived from the same structured quadrila-
compared. Discrepancies are noticeable for teral C-mesh. The far-field boundary is
low lift coefficients. They are larger for located approximately 20 chord lengths away
the subsonic Mach number than for the from the airfoil section. The mesh has 192
transonic cases. A comparison of predicted cells defining the body, 64 cells in the
and experimental pressure distributions is wake and 64 cells normal to the airfoil
given in Pig 9 f o r the transonic Mach num- surface. The first mesh spacing normal to
ber b= 0.75 at a = 0.93.. These conditions the wall is 0.002 chords. This results in
characterize the flow just below buffet unstructured triangular meshes having a
onset (see Pig 7 ) . The shock strength and total of 20,800 points or 40.960 triangles.
location are captured fairly accurately. The unstructured mesh T1 is a symmetric
The agreement between prediction and ex- grid. To form this mesh the structured grid
periment is quite satisfactory, particulary points have been used for j odd and the
in view of the fact that this case includes midpoint of the face li,j)-li+l,j) has been
regions of separated flow of two types, taken for j even. Mesh T2 is also a symme-
bubble type at the shock foot and open tric grid but was derived by direct subdi-
separation at the wing trailing edge 151. vision of each of the structured grid qua-
The convergence history of this case is drilateral cells into two traingles. Mesh
presented in Fig 10. The residual of the T3 is an unsynrmetric grid and was also
continuity equation drops approximately formed by the same procedure used for the
three orders of magnitude within 150 multi- COnStNCtion of mesh T2. These meshes are
grid cycles (90 minutes computation time on shown in Fig 11. The computed results for
CRAY-YMP). The changes of lift coefficient the isobar contours obtained from the three
and pressure drag coefficient as a function unstructured meshes and from the structured
of required multigrid cycles show that calculation are given in Fig 12. The pre-
after 150 cycles the lift coefficient dicted values of total lift and pressure
reaches its final value, while the drag drag coefficients are given in Table 1. The
coefficient is 0.25% underpredicted. Pressure coefficient distributions of the
three cases are shown in Pig 13 where they
are directly compared with the structured
4. INVESTIGATION OF SOLUTION NETHODS TO- grid results. Observing the results for the
WARDS THB APPLICATION TO HIGH LIPT isobar contours, it can be stated that the
CONFIGURATIONS TeSUltS for the three unstructured mesh
cases compare qualitatively. Mesh 2, how-
At DLR currently two approaches are being ever, exhibits a wiggle in the isobar con-
investigated to predict the flow field tours at the mesh Symmetry line which is
around multi-element airfoils. On the one most noticeable downstream of the airfoil
hand UnstNCtUred meshes are used in order trailing edge. This is clearly seen in the
to take advantage of their flexibility enlargement of the trailing edge and wake
concerning the discretization of topologi- region. This behavior has been previously
cally complex domains. They also provide a documented by Ronzheimer 1251 and results
natural basis for the use of adaptive directly from the topological dependency of
meshes. However, the accuracy of unstruc- the calculated artificial dissipation
tured mesh solutions, especially for 3-D tellRs. The predicted lift coefficient6 show
viscous flows, is still an issue which a difference of approximately 0.5% between
needs further investigations. On the other the largest value, predicted for case T4,
hand the block-structured grid approach and the smallest value, case T1. The pre-
may be employed. With this method the dicted pressure drag coefficients show a
donain of interest is Partitioned into a difference of approximately 1.5%. Referring
number of subdomains or so called blocks. back to the COmment made above about the
Within each block body fitted grids are topological dependency of the dissipation
constructed. The resulting block-structured scheme, it would seem reasonable to con-
grid has as much regularity as possible clude that differences in the solutions for
around each component of the configuration cases T1 and T2 result partly from this
being considered. Such grids facilitate behavior. Also the grid used in T1 consists
accurate calculations of viscous flow of triangles which are essentially less
fields. Different test cases are being distorted than those in T2 and T4 as the
investigated in order to explore the merits minimum angle is larger, remembering that
and limits of both approaches with respect Baker 1261 has shown that maximizing the
to the application for high lift configura- minimum angle yields a higher mesh quality
tions. As mentioned in chapter 2 both the - at least for inviscid solutions. using
block-structured and unstructured code use this criterion, mesh T1 would appear to
the same basic solution method. provide the “best”, i.e. least distorted,
mesh for highly stretched triangular grids,
4.1 Laminar Plow around NACA 0012 Airfoil and it shows excellent agreement with the
structured grid results (see Table 1). The
Firstly, the accuracy of the unstructured asymmetry of grid T4 means that the dif-
I
ld 14-5

ferences between the predicted results for given in Figs 19a and b. For both schemes
this case and those for T1 are larger than computations have been carried out with and
those between T1 and T2. The pressure coef- without multigrid acceleration. In case of
ficient distributions shown in Fig 13 exhi- the structured method Pull Multidrid was
bit only small differences when compared to used to establish a reasonable starting
the structured grid results. This is true solution on the finest mesh. and a 4-level
for each of the meshes used by the unstruc- W-cycle has been used on the finest mesh.
tured technibue. Case T2 exhibits a jumb in In the unstructured computations, a 4-level
the pressure distribution at the trailing V-cycle was used. In the Unstructured ap-
edge of the airfoil. The reason for this is proach the point distribution on the coaser
again the unbalanced dissipation stencil on meshes is independent from the distribution
the line of symmetry for this mesh. This on the finer meshes. whereas in the struc-
analysis shows that generally the accuracy tured scheme each point on a coarser mesh
of the unstructured algorithm is comparable is represented on the finer mesh. This may
with that obtained using the structured 1 M t the Wssible number of grid levels,
grid approach, provided an unstructured but careful grid generation can circumvent
grid without distorted cells is used. these limitations. The convergence proper-
ties of both schemes are very similar. For
4.2 Inviscid Flow around Two-Element single mesh computation, about 1500 time-
Airfoil steps were neyssary to decrease the resi-
duals t o 10- . With multigrid, for the
structured method 100 cycles on the coarse
mesh and 150 cycles on the finest mesh were
needed. whereas in the unstructured scheme
about 250 cycles on the finest mesh were
used. starting directly on the finest
mesh.

5. CONCLUSION
Two particular aspects of the development
of a computational procedure for high lift
aerodynamics have been presented. Navier-
Stokes calculations on structured meshes
have been carried out to Predict maximum
lift in the case of a single airfoil and a
wing without high lift devices. Using the
Johnson-King turbulence model good agree-
ment between theoretical results and ex-
perimental data has been achieved for very
high angles of attack. Rapid convergence to
steady state has been obtained due to the
multigrid method, making this solution
method attractive for engineering applica-
tions.
The scope of the applications underlined
the flexibility of the method. and cmpari-
son with experimental data demonstrated the
capability of the basic CEVCATS code for
maximum lift flows. The second aspect of
the investigations was a direct comparison
of a structured and an unstructured algo-
rithm. Calculations of the laminar flow
around the NACA 0012 airfoil showed very
good agreement of the results obtained from
both strategies. However, computations on
different triangular meshes where always
the same structured mesh was used as basis
for the triangulation. showed a dependency
of the solution on the mesh quality. For
structured meshes the influence of the mesh
quality on the solution is a well known
fact. In case of unstructured schemes cri-
teria for the necessary mesh quality are
not that well established, and the computa-
tions carried out in this study indicate
that mesh quality is a critical issue also
for unstructured schemes. hs a more complex
test case the inviscid flow around a
Karman-Trefitz airfoil with deflected flap
was chosen. Again the solutions of both
schemes agreed w e l l with each other. For
the structured scheme. a slight deficiency
occurred at the mesh singularity located on
the airfoil surface. the effect was however
only local. More sophisticated mesh genera-
tion techniques will remedy this problem.
Convergence properties of both schemes were
also comparable. and the time-lag between
different blocks in the structured approach
caused no disadvantages with respect to the
unstructured method.
14-6

The results from the investigations carried [121 Rossow, c.-C.: Berechnung von Stro-
out in this study indicate that for both, mungsfeldern durch L6sung der Euler-
structured and unstructured schemes, mesh Gleichungen mit einer erweiterten
quality is the key issue for accurate solu- Finite-Volumen Diskretisierungs-
tions. Assuming a sufficiently sophisti- methode. DLR-FB 89-38. 1989.
cated mesh generation system, the block-
structured approach appears as feasible as [131 Jameson. A., Schmidt, W., Turkel. E.:
the unstructured approach for the applica- Numerical Solutions of the Euler Equa-
tion to complex configurations. With this tions by Finite Volume Methods using
argument and taking into account the uncer- Runge-Kutca Time-Stepping Schemes.
tainties of unstructured methods with re- AIM-81-1259, 1981.
gard to three-dimensional viscous flows, at
the present time high lift activities at [141 Martinelli. L.. Jameson. A.: Valida-
DLR are geared towards more effective and tion of a Multigrid Method for the
more flexible grid generation methods for Reynolds-Averaged Navier-Stokes Equa-
block structured flow codes. tions. AIM-88-0414, 1988.
L151 Jameson. A. : A Vertex Based Multigrid
6. REFERENCES Algorithm for Three-Dimensional Com-
pressible Flow Calculations. In:
111 Butler, D.J.: Recent Progress on Deve- "Numerical Methods for Compressible
lopment and Understanding of High Lift Flows - Finite Difference, Element and
Systems. AGARD-CP-365. pp. 1.1.-1.26. Volume Techniques". Ed. by T.E.
1984. Tezduar. T.J.R. Hughes, 1986.
[21 1161 ROssow. C.-C.: Efficient Computation
of Inviscid Flow Fields around Complex
Configurations Using a Multi-Block
Multigrid Method. Fifth Copper Moun-
tain Conference on Multigrid Methods,
Colorado, USA, 1991.
L31 Radespiel, R.. ROSSOW. C.-C., swanson,
R.C.: Efficient Cell-Vertex Multigrid [171 Mavripilis, D.: Jameson. A.: Multigrid
Scheme for the Three-Dimensional Solution of the Two-Dimensional Euler
Navier-Stokes Equations. A I M Journal, Equations on Unstructured Triangular
Vol. 28, NO. 8. pp. 1464-1472, 1990. Meshes. AIM-87-0353. 1987.
141 Longo, J.M.A., Radespiel, R . : Vortical 1181 Mavripilis, D. J., Jameson. A.,
Flow Simulation by the Solution of the Martinelli, L.: Multigrid Solution of
Navier-Stokes Equations. 8th Symposium the Navier-Stokes Equations on Trian-
on Turbulent Shear Flows, Technical gular Meshes. AIM-89-0120, 1989.
University of Munich. 1991.
1191 Dimier. A.. Ronzheimer. A.: solution
151 Longo, J.M.A.: Viscous Transonic Flow of the Euler Equations on Unstructured
Simulation around a Transport Aircraft and Adaptive Meshes. DLR-IB 129-90/40,
Configuration. DGLR-Jahrestagung, 1990.
Bremen, 1992.
[201 Bailey, R.H.: A Multigrid Algorithm
161 Sanetrik. M.D., Swanson, R.C.: A Mul- for the solution of the Navier-Stokes
tigrid Solution Procedure for Multi- Equations on Unstructured Meshes for
element Airfoils. 5th Conference on Laminar Flows. DLR-IB 129-92/19,
Domain Decomposition Methods for Par- 1992.
tial Differential Equations. Norfolk,
VA. 1991. [211 ArnSt, G.: Berechnung Yon Profilum-
stromungen und Vergleich mit Mes-
Mavripilis. D.J.. Jameson. A., sungen. Studienarbeit Nr. 89-2. TU
Martinelli, L.: Multigrid Solution of Braunschweig. 1989.
the Navier-Stokes Equations on Trian-
gular Meshes. AIM-89-0120, 1989. L221 Harris, C.D.: Two-Dimensional Aerody-
namic Characteristics of the NACA 0012
Baldwin, B.S.. Lomax, H.: Thin Layer Airfoil in the Langley 8-Foot Transo-
Approximation and Algebraic Model for nic Pressure Tunnel. NASA-TM-81927.
Separated Turbulent Flows. 1981.
AIM-78-257, 1978. 1231 Redeker, G., Moller, R., Ashill, P.R.,
Johnson, D.A., King, L.S.: A Mathema- Elsenaar, A., Schmidt. V.: Experiments
tically Simple Turbulent Closure Model on the DFVLR-F4 Wing-body Configura-
for Attached and Separated Turbulent tion in Several European Wind-Tunnels.
Boundary Layers. A I M Journal, Vol. AGARD-FDP, P.2, Symposium Naples,
23. Pp. 1684-1692, 1985. 1987.

Cl01 ROSSOW. C.-C.. Kroll. N., Radespiel. I241 Kays. W.M.: Heat Transfer to the
R.. Scherr. S.: Investigation of the Transpired Boundary Layer. Journal of
Accuracy of Finite Volume Methods for Fluid Mechanics, Vol. 30. pp. 741-773.
2- and 3-Dimensional Flows. 1967.
AGARD-CPP-437, P17:l-11, 1988.
E251 Ronzheimer, A. : LOsung der zweidimen-
(11 Atkins. H.: A Multi-Block Multigrid sionalen Euler-Gleichungen durch eine
Method for the solution of the Euler Finite Volumen Diskretisierung auf
and Navier-Stokes Equations. unstrukturiertem Netz. DPVLR FB-88-34,
AIM-91-0101. 1991. 1988.
1261 Baker. T.J.: A Three-Dimensional Mesh
Generation by Triangulation of Arbi-
trary Point Sets. AIM-87-1124. 1987.
 14-7

I . Figures

a
Fig. la Control volume in Structured code. Fig. IbCOntrOl volume in unstructured code.

Fig. 2 Scruccured g r i d around

NACA 0 0 1 2 airfoil,

I grid 384 x 64.

I .3
'- I I Cn
Cn - M,- 0.50
--------- 1_I M,- 0.55
I .a
1.2.
8.8
1.8 . 0.8
8.7
8.8 . 8.6
8.5
8.6
E.,
8.1 . 8.3
0.2
8.2 .
8.1
8.8
-8.1

l;t:r
8.88 8.03 8.E< I
8.81 0.82 8.05
cw W

I
_i[
0.4
y,- 0.55

0.1
0.3
0.2
AA exWbnmt
0.1
8.B
-8.1
0.80 0.01 0.82 0.03 0.01 8.05 c, 8.07

Fig. 3 Drag polar for NACA 0012 airfoil.

 D
I .6

1.2 -
Cn .
Cn
I. 2 1.8.
M,- 0.30 0.9
I .0

t 0.6
0,s
0.q
8.3'.
0.2
0.1

8.8
.
.
.
.
.
.
-
ImtrmStLdl
arpalment
1-
-8.1
-1 8 I 2 3 . 5 6 7 8
a["] 'I

1.8.
8.9 -
8.8 -
8.1 .
0.6 -
8.5 .
a., .
0.3 .
0.2 .
B.1 .
.
B.B

-0.1 -I
' '
8
'
I 2 3 1 5' 6 ; 8 9 I8 =[0] I2

8 1 2 1 * 5
= = '0a101'2
Fig. 4 Normal force coefficient versus angle of attack for
NACA 0012 airfoil.

Fig. 5 Convergence history

a = 5.0°
Mrn-,0.30
-
Ml-
a
Re = 3 1 1 0 ~
4.P
0.66
D = 14.50
urn-0.30
for NACA 0012 airfoil. l r
E -

:I
-3
-1

-6
-5 I ::I
-7

-8

mult4*id eyclcs

Fig. 6 DLR-F4 wing-body configuration.

 14-9

..

Fig. 7 Total lift coefficient versus

angle of attack for
DLR-F4 wing-body configuration.

-4 -3 -4 -3 - 4 - 3 - 2 - 1 0 1 2 3 4 a
M_ = 0.6
M- = 0.75

M_ = 0.8
pig. 8 Drag polar for DLR-F4
wing-body configuration.

Johmon-King -
measuremems 0

0.01 0.01 0.01 0.02 0.03 0.04

CO

O.lOrn
:\.
-cP-.
8..
1.2.
0.
1.0.
0.6 .
0.6 .
0.4 .
0.2
0.0.

102
-cP
? .2
I .O
0.e
--- Euler 10.c II \ 1
0.B
__ Navier-Stokes
0.1 A measurements
0.1
0.0
-0.2
-0.. Fig. 10 Convergence history for
-0.6 DLR-FQ wing-body configuratio-
-3.0
M, = 0.75 U = 0.93'
-1 .O
0.0
5 -
,.o
Fig. 9 Cp-distribution at three different
spanwise sections.
DLR-F4 wing-body configuration.
14-10

" W\W.'.: : ' ' .

,..,..........
'U^ cl unstruetured grid T3 (unsymmetric)

Fig. 11 Unstructured and structured grids. around

NACA 0012 airfoil.

a1 unstructured grid ~1

c) unstructured grid T3
P i g . 12 Comparison of isobar contours for NACA 002 airfoil,
Re = 5 0 0 . Iz, = 0 . 8 , a = loo.
 14-11

'I /I
- 1 .2
-1.0

-1.4
-1 .6
-1.8 CASE T1
Structured Grld
Unstructured Grid -

CASETZ -
- - -- -

1 CASET3

-2.0
.@ .2 .4 .6 .8 1 .O
I
.2 .8
I
I .O.O .2 .4 .6 .8
-
X
L
.0 .4 .6 I .0
-
X
L
-
X
L
Fig. 13 Comparison oi predicted surface pressure coefficitnts
for NACA 0012 airfoil, Re = 500, M, = 0 . 8 . a = 10 .
I I
CL

I
Grld T I 0.44822 0.14431

Grid T2 0.44881 0.14658

Grid T3 0.45057 0.14665

Structured Grid 0.44691 0.14312

1 I
Table 1 Comparison of predicted total lift and pressure drag
coefficients for various grids, NACA 0012 airfoil,
~e = 500, M, = 0 . 8 . a = 10'.

a) structured grid b) unstructured grid

Fig. 14 Structured and unstructured grid around Karman-Trefftz

airfoil with flap.
14-12
0

a) structured grid b) unstructured grid

Fig. 15 160-Mach contours for Kamn-Trefftz airfoil,

M,= 0.15, a = 0'. AM = 0.02.

a) structured grid b) unstructured grid

Fig. 16 Enlarged view of structured and unstructured grid

around Karman-Trefftz airfoil at flap region.

0 04

a) Structured grid b) unstructured grid

Fig. 17 180-Mach lines for Karman-Trefftz airfoil at flap
region. &= 0.15. Q = 0'. A M = 0.02.
 14-13

-cP T
7.5' -

m Unstructured Grid
5.0 - - Structured Grid

2.5: -

0.0 -

-2.5-
.
-0.0 0.5 1.a X

F i g . 18 Comparison of surface pressure distribution for

KaRmn-TrefftX airfoil with flap, M,= 0.15, a = 0'.

1
Res

Rese 1

-6 1 Multigrid Single grid

-7 I
' ' 250 ' ' 500 ' ' 750 ' 'lobo' '1250 ';loo 0
500 Cycles
IO80 1500
Cycles
=) structured method b) unstructured method

Fig. 19 convergence histories for Karman-Trefftz airfoil with

and without multigrid acceleration.
 15-1

NAVIER-STOKES SIMULATION OF FLOW FIELD AROUND A BLOWN-FLAP HIGH-LIFT SYSTEM

R. Chow*, K. Chu and G. Carpenter**

Corporate Research Center
Grumman Corporation
Mail Stop A08-35
Bethpage, NY 11714-3580
United States

SUMMARY When the best possible high-lift configuration for a new

Solution of the Reynolds averaged Navier-Stokes equations wing is sought, wind tunnel matrix tests for the gap-
is obtained to simulate the flow field around a 13% overlap optimization between the high-lift elements are
thickness super-critical airfoil slat/blown-flap high-lift often pursued. While this approach is reliable in obtaining
system. A stadked-C mesh topology is used in conjunction information on the best CL- gain, without understanding
with the slat and the flap trailing edge streamlines and the flow physics and their implications to the data, the
adapted to accurately locate the strong viscous flow process is not cost effective in that an isolated set of test
regions. A previously modified PARC2D implicit AD1 results for a particular wing can not readily be made useful
solver is employed whereby the multiply connected for another design. In general, the flow fields under these
boundary value problem can be treated with a single test conditions are very complicated. In most cases, they
computation zone. A modelled injection boundary are dominated by viscous effects resulting from boundary
condition was demonstrated to connect transient flap wake layer confluency that can not be modelled accurately by the
vortices downstream of the computational domain. The inviscid-boundary-layer interaction method; flow
converged surface pressures and the values of the lift separation is a common feature and can even occur in the
coefficient are compared with the wind tunnel data at M, = linear range of a lift curve.
0.17. Re = 3.4 M for momentum injection coefficients of
c, =0.01 and = 0.04,respectively. The conditions under which a Navier-Stokes solution
method is needed can be demonstrated in the following
1. INTRODUCTION example of a pair of wind tunnel tests of a three-element
The development of analytic and numerical techniques is airfoil high-lift system (Ref 5). The configuration shown
important for the design and analysis of a high-!ift system. in Fig. l a consists of a 13% thickness supercritical airfoil
An accurate and dependable method to compute the lift can with a 20% chord slat having -25 deg of extension, and a
complement the laborious and expensive experimental test 30% chord single-slotted flap with main elemenvflap gap
process to achieve a desirable configuration. Interactive and overlap sizes, both at 2% chord. Shown in Fig. l a are
methods employing an inviscid solution, coupled with a curves of CL vs a and in Fig. l b and Fig. I C are pressure
confluent boundary layer analysis, have been used for some data collected at flap angles of 6f = 20 deg and 6f = 40 deg,
time to compute the high-lift flow field. A long list of respectively, at a nominal test Mach number of 0.17 and a
references using this approach can be found, for example, chord Reynolds number of 3.4 million. Turbulent flow
in the recent review article of Brune and McMasters (Ref 1). conditions were established over the entire high-lift
These methods, with often elaborate modelling of the
boundary layer interactions. are based on the assumption system. In the linear range, CL for the 6f = 40 deg case is
that the upstream influence of the flow field is transmitted seen to be higher than for 6f = 20 deg. The sudden changes
through the inviscid process. When viscous effects are of slopc of the lift curve at 6f = 40 deg also suggest that
small and when flow separation is limited to confined different flow interactions are involved as the angle of
regions (i.e., shallow separation bubbles), these attack increases. In each case (Fig. l b and IC). the smooth
calculations show reasonable results that are primarily portions of the curves on the main foil indicate the range
corrections of boundary layer and wake effects applied to an where pressure is determined primarily by inviscid
inviscid pressure field. These codes are fast and easy to run, processes. The apparent change of slopes and the
and, when used conservatively and judiciously, they can inflection points at the rear portion of the wing box show
function reasonably well in the traditional "evolutional" the extent of the upstream influence due to the presence of
high-lift design procedure to complement an experimental
the flap. For 6f = 20 deg (Fig. lb), the flow remains
intensive process. In most of these cases, the data of the
base line design are well in hand and the new designs never attached until the point just before the stall where flap
deviate far from the original optimal ones, where, as had separation occurs. The slope of the separation pressure
been pointed out by Dillner et al. (Ref 2), strong boundary curve at a = 18.36 deg also suggests that the separated
layer confluency conditions rarely exist. A two- region is represented by a thick viscous layer with sizable
dimensional, inviscid-boundary-layer, interactive multi- flow entrainment. The pressure patterns in this case are not
element airfoil method has also been used in the high-lift unlike those on the elements of a typical multi-flap high-
wing modification study for an existing aircraft (Ref 3). lift system of a civilian transport aircraft where the flap
More recently, interactive codes were used successfully to angles between the elements remain in the 15-25 deg
validate the high-lift flight test data of a civilian transport range. The inviscid-boundary layer interactive methods
aircraft (Ref 4). In the absence of large flow separation, the such as the MCARF code (Ref 6. 7) or the Moses code (Ref
study demonstrated that at near-design conditions, the 8, 9) can treat these cases adequately, with perhaps some
interactive codes can be fine-tuned to study laminar to inaccuracies near the CL-.
turbulent transition, incipient flow separation, and other
boundary layer related effects at the flight Reynolds In an effort to reduce the complexity of the system, the
number. This knowledge is valuable for correlation and single slotted flap configuration is preferred in the modem
interpretation of the high-lift wind tunnel data used in the design approach. The flap angle is increased in an attempt
design. to gain lift, which usually leads to a more complex flow
------_--_
* Senior Staff Scientist ** Staff Scientist
15-2

I SYMBOL ALPHA
4.01 I
1

SEPARATION
AT ALPHA

6
0
....,,.eo....*'

I/
-I
0.0 0.2
WC
0.2 0.4
XIC
0.6 0.8
1
1 0.7 0.9 1
WC
I

(b) S, =20°
I""-]
-4 0 4 0 12 16
1 \ 1 I v 8.19 I
ANGLE OF ATTACK, a

(a) LIFT COEFFICIENT

s-
0-

+0.2
0.0 0 0.2' 0.4 0.6 0.8
-a
1 0.7 0.9 1
WC WC XIC
(c) 61= 40'

Fig. 1 Wind Tunnel Lift Coefficlent Data 81Surface Pressure Data for High-Lift System at Flap
Extensions 6f=20° 81+doo, Mz0.17, Re=3.4M

field. Fig. IC, at aflap angle of 6f = 40 deg, shows the solver. The trailing- edges'
- streamlines and the flan
pressure distribution from one such attempt. The flow field separation line were adapted during iterations. Thi's
is dominated by massive flap separation even at very low effectively reduced the mesh dependence of the solution.
angles of attack. The relatively small surface pressure
gradient in the flap separated region suggests a general Shown in Table 1 are the results of computations of CL and
flow structure consisting of a mostly inviscid separated CD compared with experiment (Ref 5 ) for 2% and 5% chord
core region sandwiched between the flap boundary layer and flap gaps. It is seen that very good agreement w& obtained
a complex wake structure formed by the flow emanating at these angles of attack. More recently, converged
from the main-foil cove region and the top surface main- solutions at higher angles of attack, approaching the
foil boundary layer. As the angle of attack approaches C h , were obtained. The results indicated, however, that
C h , this inviscid separated core structure begins to a turbulence model more realistic than the algebraic one
break up, forming a thick viscous separated region as used in the PARC2D method is required in this high-angle-
indicated by the flap pressure slope change. The upstream of-attack range to compute CL acourately and the vortex
influence of the wake flow field becomes so strong that the breakup'of the separated region.
flow begins to separate on the main foil just before the
stall point (see Fig. IC). The variation of the lift In this paper, our methodology is employed to investigate
coefficient in a range beyond the first CL peak as seen in solution of the Navier-Stokes equations for the flow field
Fig. l a suggests transitions of flow structures are around a blown-flap high-lift system. It has been known
occurring. The challenge of an analysis method is to (Ref 11) that lift augmentation can be achieved by properly
accurately predict U priori these complicated flow fields so placing a concentrated momentum jet in front of the
that meaningful trend studies can be made in the gap- trailing edge flap. The chordwise blowing utilizes the
overlap optimization investigations. Coanda effect to attach a thin, high momentum sheet of
fluid to the curved flap surface. The injected jet energizes
the boundary layer and keeps the flow attached through
In Ref 9, we computed a solution in the linear range of the large flap deflection angles; it increases wing circulation
6f = 40 deg case, where massive flap separation was a and entrains more of the freestream air. resulting in a
dominating feature. A novel "stacked-C" mesh system was significant increase in lift generated by the wing. The
used to map the multi-element geometry into a single blown-flap experiment and the input data used in the
computation domain. The PARC2D code (Ref lo), modified present computational simulation are described briefly in
exclusively for this application, was used as the flow the next section.
 15-3

Table 1 Lift Coefficient 81 Drag Coefficient was part of aresearch program designed to explore the
Comparison for Three-Element STOL capability of a carrier-based military aircraft (Ref 5).
High Lift '-
= 0'17' Re = 3AM
Extensive surface pressure and force measurements were
made for various combinations of slat/conventional flar,

I
and slat/blown flap settings. For the present powered lift
study, we utilize the configuration having a 20% chord slat
OAP SlZE with a -25 deg extension and a 23% chord blown flap with
43-deg extension. The wing box contains a plenum from
which high pressure air was allowed to expand through a
full span adjustable slot formed by the wing box upper skin
and the flap upper surface. The blowing slot height can be
varied by adjusting the flap position. The case we analyzed
had a slot height gap of 0.006 in. Assuming an isentropic
expansion to the freestream pressure p-. the momentum
coefficient, C,. is given by:

lilVj
C, =-,
qc
where m,the mass flow per unit span at the momentum
.-.-.-.
slot, is given by

/' lil=pjVjh,

2. TWO-DIMENSIONAL BLOWN FLAP

CONFIGURATION 8i SURFACE PRESSURE
DATA po. Po, and To are the plenum density, pressure, and
For the present Navier-Stokes numerical simulation, we temperature, respectively: q is the freestream dynamic
have chosen a 2-D,three-element, blown-flap high-lift pressure: and h and c are the momentum slot height and the
airfoil system as shown in Fig. 2. The wind tunnel test chord length, respectively. The plenum conditions were
model used for surface pressure and force balance calibrated with a given momentum slot height and
measurement is a two-dimensional wing panel with a two- maintained within certain tolerance during the
foot chord, three-foot span, whose clean section is a 13% measufement. The Cp provided by the experiment was used
thickness supercritical airfoil. The high-lift experiment to model the boundary condition for the present blown-flap

13% SUPERCRITICAL AIRFOIL

20% CHORD SLAT

MOMENTUMSLOT
HEIGHT

23% CHORD FLAP

Fig. 2 Three-Element Blown Flap High-Lift System

15-4

computation. The wind tunnel conditions were the same as

those used in Ref 9 to obtain the results of Table 1 with
Mach number M , = 0.17 and Reynolds number Re =
1.7M/ft. Turbulent flow was assumed. Details of the wind 4!
a2
tunnel test setup can be found in Ref 5. Shown in Fig. 3 are 81 e3
examples of the surface pressures from the blown-flap -J -J

experiment at an angle of attack U = 10.3 deg and

c\
momentum coefficient values of C, = 0 and C, = 0.012, I
respectively. (a) SURFACE MESH DISTRIBUTION WITH SLAT I
8 FLAP TRAILING EDGE STREAMLINES
3. COMPUTATION METHOD I
Here, the solution procedure used in Ref. 9 for solving the K=K MAX
Navier-Stokes equations for a standard 2-Dmulti-element
airfoil system is extended to compute the flow over a
blown-flap configuration as described above. In this
computation scheme, a structured, stacked C-mesh is used SLATTRAILING
EDGE
1I I
whereby the physical trailing edge streamlines from the
respective high-lift elements are employed to control C1 * L S L A T L B , ...............................
STREAMLINE. D1
KSL4T+1 ..................... .
... .. .... .. .... ............... .... .... .
.
automatically the mesh condensation for the solution ..................... ...............................
KSLAT
accuracy required. In case flap separation occurs, the flap c2 A ~ - ' S L A T ~B~ D2
separation streamline can also be used to specially treat the ; INJECTION I
separation zone (Ref 9). This step was not necessary in the
present application. Shown in Figs. 4a and 4b are the
stacked C-mesh system in the physical plane, with the K- 1
boundary mesh distributiofis, and in the computation J
plane, respectively. The lines bldl and b d z in Fig. 4a are BLOW FLAP BLOW FLAP
the adjacent coordinate lines corresponding to the slot LOWER UPPER
Railing edge streamline. Similarly, the lines e3c3 and f3d3 SURFACE SURFACE
are the wake cuts Corresponding to the flap trailing edge
streamline. Since the flap and the wing box are snugly (b) COMPUTATION PLANE
fitted except for the momentum slot gap (see Fig. 2), they
are treated as a single element. The point corresponds to Flg. 4 2-D Blown Flap System
the momentum slot injection location of the experiment.

The numerical solution for the Navier-Stokes equations is getting an accurate lift solution for the multi-element
obtained using a single computation zone as shown in Fig. system. Our version of the PARC2D code for application
4b. The flow solver is our modification of the implicit to the high-life airfoil has the general efficiency of the
finite difference code PARC2D (Ref 9). The use of this code NASNAmes ARC2D code (Ref 12) as applied to a single-
allows the boundary condition on the slat AIB1B2A2 to be element airfoil calculation. However, due to the
prescribed directly. This unique feature allows multi- complexity of the flow field in a high-lift system,
element high-lift airfoil systems to be treated conveniently considerably more CPU time is required for a converged
with a single zone, thus avoiding the residual error control solution for CL . The code uses an algebraic turbulence
problems one usually has to monitor carefully in a multi- model loosely based on the Thomas formulation (Ref 13) of
zone algorithm. This issue is particularly important in the Baldwin and Lomax model (Ref 14). Turbulent flow
over the entire high-lift system was assumed in all the
calculations made in the present work.

The experimental values of Cp for the angle-of-attack

DATA survey chosen for our simulation had fluctuations of up to f
20% from an average value of Cp = 0.0107. However, the
experimental values of lift were not overly sensitive to
these variations. We chose a nominal value of = 0.01 c,
for the numerical computations (barred quantity denotes the
values used in the computations). The actual experimental
condition near the momentum slot exit could not be
simulated easily without a controlled study of the internal
flow through the gap and its interaction with the flow

I
beyond the exit. We anticipated a fair amount of friction
loss through the gap. Furthermore, the velocity
1
0.0 0.1 02 0 02 0.4 0.0 I
0.8 1
distribution was expected to be different from that given by
a constant value of Vj (Eq (1)) across the momentum slot I

X/C WC
gap. Without any additional knowledge, we chose thc c, I
I
value as the controlling magnitude to model the boundary I
condition in the flap. The dynamic injection pressure
I
Fig. 3 Experimental Pressure Distribution for 13%
Fjv; can be computed from Eq (1) with a given q.A
Supercrltlcal Alrfoll Blown Flap System, distribution function f(s) =pi(s)[V,(s)] /p,Vf I 1 (Fig. 5 )
2 --
M, = 0.17, a = 10.3 deg, Re = 3.4M was used for establishing the boundary condition along the
 15-5

geometry. specifying the airfoil and the upper surface of

the flap, and exits at the upper trailing edge with J=375.
The boundary then proceeds along the same anticipated

4- EXPERIMENTAL
JET WIDTH
.streamline paths as for the initial segment (J=1,27) to the
far field but is displaced from the initial segment by the
local grid stepsize. The outer boundary (K=50) for the

I\
inner subface proceeds from the far field along a path
mimicking the shape of the lower wing-flap edge and
provides a normal approach to the slat's lower leading edge
0.5 at point a2 shown in Fig. 7. The boundary then continues
from J=117 along the bottom surface of the slat and exits at

0 b===J S
the trailing edge of the slat at point b2 with J=241. The
remainder of the boundary follows the anticipated slat
trailing edge streamline passing over the top of the main
foil and flap and out to the far field.

The outer subface shown in Fig. 6 is defined for K=51.121

and J=1,401. resulting in 28,471 grid points. The inner
boundary (K=51) of this subface follows the same path
Fig. 5 Numerlcal Momentum lnjectlon Distribution, shape as for K=50. but is displaced outward by the local
grid stepsize. until it attaches to the slat at point al, where
it then proceeds around the slat's upper leading edge. The
boundary exits the upper slat trailing edge at J=241 and
once again follows the slat trailing edge streamline path to
top flap surface. where s is the dimensionless running the far field but is displaced by the local grid stepsize from
surface length measured from the experimental injection K=50. The far field boundary for K=121 represents the C
point to the flap trailing edge. The injection velocity Vj(s) topology, extending four cord lengths forward and nine
was assumed to be tangential to the flap surface, from lengths behind the multielement airfoil.
which the surface velocity boundary conditions were
determined. One could think of this procedure as modelling The distribution of grid points along the individual
the complex local interaction by a thin shear layer across boundary curves is critical for insuring grid quality based
which the static pressure changes are assumed to be on orthogonality and skewness measures. GRIDGEN-2D
negligible. The function f(s) was determined through provides features for easily controlling point clustering as
numerical experiment such that the local value was the illustrated in Fig. 7, where the point distributions in the J-
minimum required to convect the flap vortices downstream. direction along the lower section of the slat are set
The effectiveness of the blowing is determined here by its essentially uniform, due to the small flow gradients in the
elimination of the separation region on the flap. The cove region as compared to nonlinear distributions along
method minimizes the details of the blowing simulation as the leading edge. Here there is significant clustering near
well as provides a mechanism for convecting transient the anticipated stagnation point required to capture the
vortices downstream of the high-lift system. The validity pressure suction peak, followed by a thinning of points
of this approach is determined by the agreement of the along the upper edge due to the fact that the same number of
calculated pressure distribution with the experimental points must span the upper and lower edges of the slat. The
measurements. distributions cluster again towards the trailing edge of the
slat in order to blend with distributions selected in the slot
4. STACKED C-MESH GENERATION region. Another area which required considerable
The "stacked-C" computational mesh is generated clustering in the J-direction is shown in Fig. 6 in the
interactively on a Silicon Graphics Iris-4D workstation region of the momentum injection slot (J=321,332).
using GRIDGEN (Ref 15). a three-dimensional, elliptic grid
generation code developed by the U.S. Air Force. The The height of the inner subface over the top of the airfoil or
process of developing the mesh begins with extracting the position of K=50 above the wing is 0.013 cord lengths
from the database describing the slat-foil-flap geometry the while the first step size in the Navier-Stokes boundary
input into GRIDGEN3D, a member of the GRIDGEN family layer is 0.0000063 cord lengths. This step size was
of codes used to develop two-dimensional surface grids.
The grid strategy used for the high-lift multi-element airfoil
is shown in Fig. 6. where the stacked C's are designated as .J = 401
the inner and the outer C-meshes. These divisions are J = 241
usually referred to as blocks or zones for three-dimensional
grids or as subfaces in GRIDGEN terminology. The
combined C-mesh topology has a total of 48,521 grid MOMENTUM INJECTIONSLOT
points, with the normal grid index K ranging from 1 to 121
and the longitudinal index J varying from 1 to 401. One of
the major challenges in developing the grid for this
problem is specifying the point distributions along the
subface boundaries such that grid orthogonality is met at K-1
the boundaries and grid skewness is gldbally minimized. K=50
OUTER BLOCK (71. x 401) K=51
The inner subface is composed of 20.050 grid points with
K=1,50 and J=1,401. The boundary for this subface with
K=l starts in the far field at J=l and follows the anticipated
flap trailing edge streamline forward, where at J=27 it
attaches to the underside trailing edge of the flap. The
boundary continues along the flap and around the subface Fig. 6 Stacked C-Mesh Topology
15-6

OUTER BLOCK

INNER BLOCK

Fig. 9 Navier-Stokes Grid in Slat & Cove Region

Fig. 7 Slat Geometry Specifies Block Boundaries

maintained along the upper edges and was increased to

0.00002 on average along the lower edges. With the edge
clustering specified for both the inner and outer subfaces.
the two-dimensional grids were generated first using the
algebraic or transfinite interpolation algorithm. Next a
Laplace solver was used to eliminate grid crossings caused
by the complex geometries, but it resulted in a grid where
the K curves are pulled away from the surface geometry,
resulting in poor orthogonality conditions. Grid quality
was restored with the use of an elliptic solver using
Thomas-Middlecoff control functions, wi,th the resulting
grid meshes for the inner and outer subfaces shown in Fig.
8. Details of the grid for exploded regions around the slat Fig. 10 Navier-Stokes Grid in Flap Region
and flap regions are given in Fig. 9 and Fig. 10.
respectively .

Fig. 8 Navler-Stokes Grid Around Multi-Element Airfoil

 15-7

5. RESULTS & DISCUSSION An additional calculation was made for a higher momentum
The experimental pressure distributions in Fig. 3 show -
coefficient C, = 0.04. The'angle of attack, a,was equal to
that, without boundary layer control, the fluid cannot
negotiate the large camber change at the flap and has 6.195 deg. Again, the surface pressure results compare well
separated. Thus the flap has lost its effectiveness in with the data, as shown in Fig. 15. The isobars for this
contributing to the lift. Data also indicated that the case are shown in Fig. 16. The streamlines are shown in
separation had also occurred at much lower angles of attack Fig. 17. The converged numerical solution for CL value
(not shown). It is seen that very little energy (C,, = 0.012) was4.22 as compared with the value from the integrated
is required to achieve a dramatic change in the whole pressure data of 4.17.
pressure field by shedding the vortices in the separated
wake through using momentum injection. The local 6. CONCLUDING REMARKS
expansion has created a negative pressure peak and allows
the flow to recompress gradually towards the trailing edge In the present paper, we have shown successful examples of
with a back-end surface pressure distribution similar to that a computation of a 2-D blown flap using the Navier-Stokes
of a typical rear-loaded airfoil. The injection also lowers equations compared'to experiment. This method can be
the overall pressure levels on the top surfaces, including employed to estimate the lift gain of the present type of a
that of the slat, making the lift gain particularly effective. powered high-lift device. Since the boundary layers
remain through a vortex shedding process, the simple
For our calculation, we used a mesh with 401 cells in the algebraic shear-layer turbulence model used in the present
streamwise direction and 121 cells in the normal direction. calculations appears to be adequate for predicting the
The Mach number, M-, was equal to 0.17 and the Reynolds surface pressures. The, present flow solver is very slow. It
number, Re, was 3 1 . 4 ~ 1 0 ~Turbulent
. flow condition was is highly desirable to develop a fast multi-grid algorithm
assumed over the entire high-lift system. The momentum for this type of application so that the methodology can be
coefficient, C,. unless otherwise mentioned was 0.01. The used as an effective engineering tool. Further sensitivity
computation started with the standard procedure using a studies are required of the model of the injection boundary
freestream initial condition with the surface boundary condition using a more carefully controlled experiment
condition blended in gradually at the early cycles of the near the injection slot, which was not sufficiently provided
iteration. After about 5000 cycles, the flow established by the present collected data.
was largely separated even with the injection boundary
condition already imposed. The time-sequenced results of 7. ACKNOWLEDGMENT
the streamlines and the surface pressure solutions at an
We gratefully acknowledge the Engineering Flight
angle of attack, a. of 6.153 deg are shown in Fig. 11. We Sciences Directorate of the Grumman Aircraft Systems
observe that, in Fig. 11a, the flow was largely separated at Division for providing the wind tunnel data for the present
the mainfoil-flap juncture with the streamlines showing a investigation.
slight hint of the effect of the injection. The shedding
process continued as shown in Fig. l l b and l l c . and, 8. REFERENCES
finally in Fig. l l d , the vortical flow of the wake
completely disappeared from the the flap surface. The 1. Burne. G.W. and McMasters, J.H.. "Computational
surface pressure solution at this stage agreed very well with Aerodynamics Applied to High-Lift Systems,"
the experimental data except for a confined local overshoot Progress in Astronautics and Aeronautics, Vol. 125,
of the result, reflecting the effect of the model of the A A , P.A. Henne.
injection boundary condition. The size of the separated
ed.. 1990. pp 389-433.
cove region decreased as the flap separation was reduced. In
Fig. l l a , we observe that the back-end separation had a
2. Dillner, B., May, F.W., and McMasters, J.H..
blocking effect on the flow in the slat channel. This is a
"Aerodynamic Issues in the Design of High-Lift
good example of the dominating nature of the back pressure
for a separated high-lift flow condition. For completeness, Systems for Transport Aircraft," AGARD CP-365,
Paper 9, May 1984.
the pressure isobars of the converged solution for this case
are shown in Fig. 12, and the velocity vector field for the
3. Waggoner, E.G., "Development of High-Lift Wing
boundary layers on the top surfaces is shown in Fig. 13.
Modifications for an Advanced Capability EA-6B
Using the same procedure, four additional cases with angle
Aircraft," J%opress in Astronautics and Aeronautics.
of attack a = 2.051 deg, 10.300 deg, 14.370 deg, and Vol. 125. b u l ied Comuutational A e r o d v n a m ,
16.390 deg were computed. The computed lift coefficient, P.A. Henne. ed.. 1990, pp 435457.
together with the force balance data and the value CL
obtained from integrating the experimental surface 4. Vijgen. P.M.H.W., Hardin. J.D., and Yip, L.P.,
pressures, areshown in Fig. 14. As we mentioned, there "Flow Prediction over a Transport Multi-Element
were Cp value fluctuations during the angle of attack High-Lift System and Comparison with Flight
survey, resulting in some scatterings of the integrated Measurements," Aerodynamic Flows, California State
University, Long Beach, January 13-15, 1992.
values of CL. whereas computations were made with a c,
value of 0.01. The discrepancies between the force balance 5 Laia, J.R., "Analysis and Test Report of 2-D Wind
CL'S and those of the integrated pressure data values are Tunnel Test of a 13% Super-critical Airfoil Section
somewhat puzzling and unresolved in Ref 5 . One Fitted with Various Blown and Mechanical Flap High
conjecture as to the cause of the differences might be an Lift System," Grumman Aerospace Corporation
incorrect tunnel wall interference correction on the values Report #XA128-900-2, July 1983.
of CL from the force balance for the blown-flap
experiments. The values of CL from the numerical results 6. Stevens, W.A., Goradia, S.H.. and Braden, J.A.,
"Mathematical Model for Two-Dimensional Multi-
compare well with the integrated pressure data in the linear
component Airfoils in Viscous Flows," NASA CR-
range of the lift curve. There is some deterioration in the 1843. 1971.
agreement at higher angles of attack near the CL-.
15-8

DATA
....... NUMERICAL-UPPER SURFACE

6/ i.- I : I__,
o EXPERIMENTAL-UPPER SURFACE
EXPERIMENTAL-LOWER SURFACE jB

............... .., . 00

....... ............
........; ',*". .e.;
0 .
0

- I , , , , , , . , , *
0.0 0.2 0 0.2 0.4 0.6
' 0.8 1
(a) 5KCYCLES iuC WC

0 : j#
1-0
0

0.2
0.0 0.2
.ob
.............

. 0

0.4 0.6 0.8 1

I:
WC WC
(b) 10 K CYCLES

1 1
va ...........
:
a
9. -
0 . Q-

-I--, 4 , I , , , , , , I ,

0.0 0.2 0 0.2 0.4 0.6 0.8 1

(c) 20 K CYCLES WC WC

I0.2
O.
-O 0.2 0.4 0.6 0.8 1
WC WC
(d) 30 K CYCLES

Fig. 11 Time Sequenced Streamline Solution 81Surface Pressure Solution

(M,=0.17, Re=3.4M, Ok6.153", c p = 0.01) I
 16-1

THE GARTEUR HIGH LIFT RESEARCH PROGRAMME

J . J . THIBERT
Frencb GARTEUR Aeroresponsable
OFFICE NATIONAL D’ETUDES ET DE RECHERCHES AEROSPATIALES
ONERA - BP 12 - 92322 CHATILLON CEDEX FRANCE
SUMMARY ML local Mach number

An extensive European wind tunnel research pro- Re,R Reynolds number

gramme on high lift systems has been carried out in
the past few years within the framework of a Re,, Reynolds number, based on retracted
GARTEUR Action Group. To provide a relevant and chord of the 59 % wing station
realistic case, permission was given by British
Aerospace to use a section from the A310 wing from Re, Reynolds number, based on aerodynamic
which a 2D airfoil could be derived. A Deutsche mean chord of half model. Reynolds
Airbus 3D half model of the A310 aircraft has been number, based on retracted chord of 2D
used for the 3D test and an airfoil representative model
of the 59 % span section has been used for the 2D c
test. The wind tunnel test programme carried out in R attachment line Reynolds number
the major European low speed wind tunnels (ONERA F1
in France, NLR HST and LST in Netherlands, RAE 5 m U, v velocity components of 3D-B.L profile in
in UK) was complemented by a full scale flight test streamline coordinate (U = streamwise
supported by Airbus Industrie. A wide range of component, V = crossflow component)
Reynolds numbers and Mach numbers has been covered V local total velocity of 3D-B.L. and wake
by the test and a very comprehensive, well profile
integrated and accurate body of data has been
generated by this research programme. After pre- v, total freestream velocity
senting GARTEUR the paper deals with the Reynolds
number and the Mach number effects as well as the X, coordinate of the airfoil
correlations between 2D and 3D data and between
wind tunnel and flight test results. 2 height of wake or B.L. probe
a angle of attack
LIST OF SYMBOLS
P angle between x-axis and local total
velocity of 3D-BL profile ; sideslip
AL angle of attack
6, displacement thickness
C chord of retracted high lift system
rl wing station
CD total drag coefficient
0 sweep angle
CD” wake drag coefficient
skin friction coefficient SUBSCRIPTS

CFLTW skin friction coefficient (Ludwing- C based on retracted chord

Tillman)
c,,,, at maximum normal force coefficient
total lift coefficient,
CL,,, at maximum lift coefficient
integrated normal force coefficient in
wing-axis-system LE leading edge
total moment coefficient, reference point max maximum
0.25

c, static presure coefficient min minimum

cw chord of main wing 2D 2D flow
D1W displacement thickness of B.L. and wake, 3D 3D flow
based on wall pressure coefficient 59 59 % wing station

KPI pressure coefficient

M,Ma freestream Mach number
16-2

1 INTRODUCTION the basis of balanced contributions.

1.1. Garteur presentation Parallel to the formation of GARTEUR, on the in-
dustry side various collaboration arrangements
GARTEUR (Group for Aeronautical Research and Tech- between companies have been developed. When GARTEUR
nology in m o p e ) was formed in 1973 by represen- was formally established under the MoU in 1981, the
tatives of the Governmental departments responsable airframe industries of the four GARTEUR countries
for aeronautical research in France, Germany and replaced an earlier body with the Industry Group
the United Kingdom. The Netherlands joined in 1977 for Collaboration on Aeronautical Research and
and Sweden in 1991. Technology in Europe (CARTE).
The GARTEUR community at the present time comprises Representatives from Swedish industry joined CARTE
more than 250 individuals from various research in 1991.
establishments, industrial companies and government
authorities who actively participate in the orga- CARTE is the formal communication link with GARTEUR
nization. and interfaces with the Executive Commitee (XC) on
both policy and research matters. Relations with
According to its MoU, the mission of GARTEUR is to CARTE have also been established at other GARTEUR
mobilize, for the mutual benefit of the GARTEUR levels. Initially Points of Contact from industry
member countries, their -scientific and technical were assigned to all Groups of Responsables, and,in
skills, human resources and facilities in the field due course some of the industrial senior
of aeronautical research and technology for the specialists themselves became Responsables. At
following purposes : Action Group level all CARTE companies are actively
participating, and in some areas are supplemented
a Strengthening collaboration between European by participants from the materials supply industry
countries with major research capabilities and and other aerospace related companies.
government funded programmes.
GARTEUR has operated about 46 Action Groups since
m Continuously stimulating advances in the aero 1981, 25 of which have successfully finished their
nautical sciences and pursuing topics of applica- activities.
tion-oriented research in order to maintain and
strengthen the competitiveness of the European 1.2. The Iiigh Lift Action Group programme
aerospace industry.
High levels of performance for transport aircraft
Concentrating existing resources in an efficient high lift systems are mandatory since they play a
manner and seeking to avoid duplication of work. crucial role for the payload/range of the aircraft
and its noise characteristics. This has been reco-
a Performing joint research work in fields suitable gnized for a long time and the development of
for collaboration and within research groups spe- mechanical high lift systems during the past has
cifically established for this purpose. provided practical systems which give good CL,,, and
L/D for today's aircraft. However the potential for
a Identifying technology gaps and facility needs further improving the performance of mechanical
and recommending effective ways for the member high lift devices is substantial but intensive
countries to jointly overcome such shortcomings. efforts are required both in theoretical and expe-
rimental aerodynamics. Concerning the experimental
rn Exchanging scientific and technical information. aspect, the lack of high quality experimental data
for 2D and 3D high lift configurations is one of
GARTEUR is organized around three main elements the reasons that preclude the development of re-
(see fig. 1) : liable and efficient theoretical methods for the
design and the analysis of such complex configura-
a The GARTEUR Council, the governing body, assisted tions.
by the Executive Committee and the GARTEUR Secre-
tary. This is why GARTEUR decided in 1984 to set up an
Action Group with the task of proposing and coor-
rn The Groups of Responsables, the scientific mana dinating a major multinational research programme.
gement body and think-tank of GARTEUR. Each of the f.our member countries FRANCE, GERMANY,
\ THE NETHERLANDS and the UNITED KINGDOM, provided
m The Action Groups which constitute the technical representatives both from their national research
expert body for programme formulation and the ex- organizations as well as from their aircraft in-
ecution of the GARTEUR research work. dustries to contribute to this group. The Action
Group therefore proposed a test programme with the
GARTEUR is an independent organization ; it has, following objectives :
however, no staff of its own or a common fund at
its disposal. The necessary resources (staff, - To study in 2D flow the Reynolds number and the
facilities, costs in kind) for the joint research Mach number effects on a selected high lift
activites are made available by the Governments of configuration.
the member countries out of their national pro-
grammes, or by the participating organizations on - To study the same configuration in 3D flow.
 16-3

- To study the effect of wing sweep on the The programme proposed to investigate the following
transition location as a function of the Reynolds high lift configuration :
number.
conf.1 Take-off configuration
- To correlate wind tunnel and flight test measu conf.2 Landing configuration
rements.
The main emphasis of the research programme in the
To provide a relevant and realistic case, permis- 2D and 3D tests is directed to the take-off confi-
sion was given by British Aerospace to use a sec- guration which has a well ordered flow and the
tion from the A310 wing from which a 2D airfoil landing configuration which has a separated flow on
could be derived. In the period between 1985-1989, the flap. For the comparison with the corresponding
the Action Group carried out an extensive wind 2D flow data 3D tests were performed with the half
tunnel measurement programme on a 3D half model of model without nacelle.
the A310 aircraft and on the local sweep normalized
59 .% span section in 2D flow. This wind tunnel However to provide correspondence with flight tests
programme was complemented by a full scale flight which became a firm part later in the research
test programme supported by Airbus-Industrie programme, tests with a through-flow nacelle moun-
(fig.2). This action Group AD (AG08) consists of ted under the wing of the half model were added for
members from ONERA, NLR, DRA, DLR, Aerospatiale, the take-off and landing configurations in the RAE-
Fokker, BAe, Deutsche Airbus and Airbus-Industrie 5M and ONERA F1 wind tunnels.
(fig.2).
2.1.1. 2D Models M1 and M3
_________________________
2. MODELS AND TEST PROGRAMME The model M1 was constructed and manufactured by
DLR-Braunschweig and the instrumentation was partly
2.1. Hodels and instrumentation carried out by DLR and completed by ONERA. The
model M3 was manufactured and instrumented by NLR.
The series of models used for high lift investiga-
tions of the GARTEUR AD (AGO81 are :

2D model M1
The main dimensions of the model are :
IModel M1 I HS;del M3
LST

(": 1
2D model M3
half model M139 span (mm)
Ichord (retracted) (mm)l:::2
as shown in fig. 3. The A310 flight tests completed aspect ratio 2.22 5.25
these tests.
The half model M139 consists of the left part of The chordwise instrumen ation of models M1 and M3
the complete MBB model M139, a 1 : 9.5 scale A310. is listed in fig.3 and briefly discussed below :
The high lift system of the half model (Fig. 44)
consists of : - pressure taps

-,a Kruger flap in the wing root fairing Both models have the following number of pressure
taps : slat 35 taps, wing 60 taps, flap 35 taps.
- a slat (divided in three parts) at the leading The diameter of the pressure holes was 0.3 mm.
edge
- a double slotted flap (vane-flap) at the trailing - Unsteady p r e s s u r e t r a n s d u c e r s (h'uli t e s )
edge of the inner wing and,
Unsteady pressure was measured at 12 locations on
- a single slotted flap (Fowler flap) at the the model M1 : 8 on the slat upper surface, two
trailing edge of the outer wing. on the upper side of the wing nose and one on the
wing shroud, one on the flap upper surface.
The model is also equipped with an all speed aile-
ron (ASA) in the engine region and a low speed - Skin f r i c t i o n d e v i c e
aileron (LSA) on the outer wing. The 59 .% wing
station is located on slat 2 between slat tracks 6 Model M1 was equipped with 4 hot film gauges of
and 7 and nearly in the middle of the Fowler flap. 6 mm diameter and depth to measure skin fric-
tion : one at the slat t,.e., one at the wing t.e.
The 2D model coordinates and settings are obtained and two at the flap upper surface. Preston tubes
I from the A310 wing coordinates of the 59 .% statSon with an outer diameter of 0.5 mm were mounted on
using local sweep normalization process. The the model surface to measure skin friction on
ordinates of the undeployed 2D section are given model M3.
then by
- Boundary l a y e r cove r a k e s
(Z/C),,=(Z/C),,/cos#, (x/c) , 6,, = 6,,/cos#,
In the slat and wing cove small rakes were ins-
where #, is the sweep of the undeployed constant talled on the models M1 and M3 to measure total
x/c lines. pressure in the separated flow region.
16-4

- Internal boundary layer traversing mechanism in order to obtain a total survey height of
22 mm.
An internal traversing mechanism could be instal-
led on the model M3 to measure the viscous flow 2.2 Wind tunnel test programs
at the slat t.e. (station 1) and at the wing
leading edge (station 2 x/c = 0.131). 2.2.1. 2D Test Programs
.......................
- Transition observation
ILR LST and EST Wind Tunnels
Model M1 was equipped with a plastic strip of
polyurethane resin on the upper surface of slat, Extensive 2D tests with model M3 were first per-
wing and flap, to measure the laminar / turbulent formed in the NLR LST and HST wind tunnels.
transition location by infrared image technique.
The model was installed vertically in the 2D test
2.1.2. Half Model M139 section of NLR LST wind tunnel between the two
...................... turntables in the ceiling and the floor, fig. 6. In
The half model M139 consists of the left wing of order to determine wake drag of the wing section,
the existing MBB A310 - complete model M139 and of a'quadruple wake-rake was mounted horizontally in
a new half fuselage manufactured by Deutsche Air- the tunnel with its tubes approximately one chord
bus/DLR. The main dimensions of the half model are distance downstream of the trailing edge. An ex-
ternal boundary layer survey device was separately
semi-wing reference area 1.2135 m* mounted on the tunnel floor.
semi span 2310 mm
aspect ratio 8.8 In the NLR-HST wind tunnel the model M3 was mounted
leading edge sweep 30.2O in the test section with the special 2D set-up,
sweep of 25%-chord 28.0° fig. 7. The wake was measured with a rake at a
trailing edge sweep 20.5O distance of 0.76 retracted chord downstream of the
aerodynamic mean chord c 614 mm trailing edge. An external boundary layer traverse
chord of 59 % wing station cs9 427 mm device, constructed on a special wind tunnel strut,
fuselage diameter / length 594 / 4830 mm was installed to measure with a dual probe the
total and static pressure at the boundary layer
The instrumentation of model is listed in fig. 3 stations.
and briefly discussed below :
The measurements in the NLR LST and HST wind tun-
- pressure taps nels consisted of pressure measurements, boundary
layer and wake surveys and flow visualization
In the high lift configuration 602 pressure holes within a Reynolds and Mach number range of Re = 1.9
with a diameter of 0.5 mm are fitted in 13 wing - 7.10' and M = 0.13 - 0 . 3 0 . In addition to the
sections. 40 holes are located on the fuselage. airfoil pressure distribution for a complete a-
The relevant 59 % wing station contains the sweep up to stall, detailed flow investigations
following number of pressure taps : slat 16 taps, were carried out. These included :
wing 30 taps, flap 15 taps.
- unsteady pressure measurements with Kulites at
- hot film gauges the slat nose and the main wing nose, and at the
trailing edges of all three components,
Three hot film gauges were glued at the slat
hook : one at the 59 % wing station, one inboard - measurements of the extent of the separation
and one out,board of the 59 % wing station. bubbles in the slat and main wing coves with
fixed rakes,
- razor blades
- skin friction measurements at the trailing edges
One razor-blade of 0.64 mm thickness on the flap of slat and main wing coves with Preston tubes,
at x/c = 0.895 and one of 0.44 mm thickness on
the wing shroud at x/c = 0.7 were installed on - total wake surveys behind the airfoil for drag
the model at 59 % wing station to measure some determination.
skin friction values in the RAE-5M wind tunnel
tests.
Flow visualization was carried out for the take-off
- internal boundary layer traverse device and landing configurations to obtain information
about flow separation and boundary layer
To measure the wing boundary layer at x/c = 0.5 transition, using oil flow and sublimation tech-
of the 59 % wing station an internal boundary niques. Boundary layer and wake surveys were car-
layer traverse device could be installed by ried out at 8 stations for the take-off configura-,
mounting a special shroud on the main wing, the tion at a = 12O and 20°, and at 9 stations for the
viscous shroud in Fig. 5. The internal boundary landing configuration at a = 12O and 21°, at
layer device, designed and built by ONERA was M s 0.22 and Re = 1.9.10' (LST), M = 0.22 and
motorized allowing a survey height of 11 mm for Re = 4.0.10' (HST). In the HST, some extra surveys
the two three-hole probes that were 11 mm apart were performed at Re = 1.9.10' and 6.0.10' for
 16-5

comparison with measurements in the LST and ONERA 0.18 - 0.34 for the take-off and landing
F1 wind tunnels respectively. configuration with and without nacelle,
investigations of the flow over the wing upper
OWERA P1 Wind Tunnel surface with particular emphasis on the 59 %semi-
span region were carried out using oil flow
2D tests with the large 2D model M1 were carried visualization techniques to obtain information
out in the ONERA F1 wind tunnel. The model was about flow separation , especially separation
mounted in the 2D test section, fig. 8. The exter- bubbles in nose regions, and china-clay
nal survey device was installed to measure the visualization for transition detection. Results
boundary layer profiles at the same station as on were obtained for the take-off and landing
model M3 and the near wake (station 8) at configurations without nacelle at M = 0.25 and two
x/c = 1.06. perpendicular to the flap chord. The far Reynolds numbers (Re, = 3.35 and 7.33.10'), for two
wake was measured with the same external survey angles of attack : a, corresponds to 0.7 C,,,, and a,
device at a distance of 1.0 retracted chord to near CL,,,. Some skin friction measurements using
perpendicular to the flap chord. The far wake was the razor-blade-technique were also made for the
measured with the same external survey device at a same model and test conditions. Hot film
distance of 1.0 retracted chord perpendicular to measurements were performed to detect the
the wing reference chord, downstream of the trai- attachment line transition on the take-off
ling edge. configuration with and without nacelle.

The measurements in the ONERA F1 wind tunnel in- ONERA P1 Wind Tunnel
cluded :
The 3D tests in the ONERA F1 wind tunnel, which
pressure measurements as follows : were complementary to the RAE tests, were carried
out over a Reynolds number range of Re, = 3.5 -
static pressure measurements on the element 11.10' at a Mach number of M = 0.25 for the take-
surf aces off and landing configuration with and without
unsteady pressure measurements with Kulites at nacelle. Balance measurements, pressure
the slat nose and the wing nose and at the trai- measurements on the entire wing and fuselage, and
ling edge of all three elements boundary layer and wake survey at the 59 % wing
total pressure measurements in the slat and wing station were performed. For the attachment line
cove, investigation hot film measurements were carried
out on slat and wing in the take-off and landing
boundary-layer and wake surveys, configurations as an aid to the interpretation of
the flight tests.
skin friction measurements, and
In both wind tunnels an additional test was carried
transition detection by thermography out at M = 0.2 and Re, = 2.58.10' at atmospheric
pressure to provide a direct comparison with
for the take-off and landing configurations within measurements made in the DNW wind tunnel on the
a Reynolds and Mach number range of Re = 6 - complete model M139.
16.4.10' and M = 0.13 - 0.30. The boundary layer
measurements were carried out at the corresponding 2.3. A310 Plight Test
stations for two angles of attack of a, = 12O and
at = 21°, at M = 0.22 and three Reynolds numbers To provide a greater understanding of the
Re = 6, 9 and 13.10'. differences between winn tunnel and flight, tests
with the A310-300 prototype MSN 378 were carried
2.2.2. 3D Test Programs out with the support of Airbus Industrie.
.......................
2.3.1. Flight Test Instrumentation
Extensive 3D tests were performed on the half model ____________________--------------
M139 in the RAE-5M wind tunnel and in the ONERA F1
wind tunnel. The installation and operation of the
instrumentation of the starboard wing of the
The half model was mounted above a peniche of 81 mm aircraft was under the responsability of a flight
height (to separate if from the tunnel boundary test group. The general arrangements is given in
layer) on a six component underfloor balance in fig. 11. At the 59 % wing station the entire wing
both wind tunnels, fig. 9. To measure the wake drag section is instrumented and at the 64.5 % station
at the 59 % wing station an external motorized wake only the slat :
survey device was installed in the ONERA F1 wind
tunnel, fig. 10.
- pressure holes
RAE-5H Wind Tunnel
Three pressure belts.were fixed around the wing
The measurements in the RAE-5M wind tunnel included box including airbrake, the flap and the slat. 2,
balance measurements, pressure measurements over containing 25 holes on the slat, 3.0 holes on the
the entire wing and fuselage, within a Reynolds and wing and 15 holes on the flap. The location of
Mach number range of Re, = 3.40 - 9.2.10' and I = the pressure holes was the same as on the half
16-6

model M139 but with additional holes in the nose 3. DATA ANALYSIS
region of the slat.
The programme has generated a very comprehensive,
- hot film gauges well integrated and accurate body of data which are
still under analysis. In this section some results
Adjacent to the 59 % and at the 64,5 % station will be presented with emphasis on Reynolds number
two rows of 7 hot film gauges were glued on the and Mach number effects on C,,,,, 2D/3D correlations,
slat belt for the attachment line transition wind tunnel/flight test comparisons and Attachment
investigation. Line Transition results.

- unsteady pressure transducers (Xndevco) 3.1. Reynolds number effect in 2D flow

For.the measurement of the unsteady pressure 5 The evolution of the maximum lift coefficient Cl,,,
Endevco pressure transducers were flush-mounted over the measured Reynolds number range from
with the slat belt surface : 4 in the slat nose 1.9.1O'up to.16.7.10' and a constant Mach number of
region and one near the slat t.e.. Also one M = 0.22 is shown in fig. 14 for the take-off and
Endevco transducer was installed at the wing t.e. the landing configurations. Favourable Reynolds
effect occurs up to 6.7.10' for both configura-
- boundary layer rakes tions. Further increasing Reynolds number has no
To measure the total pressure in the boundary layer effect on Cl,,, for the landing configuration while
one rake was installed at the slat t.e. and one at it causes a small decrease of the C,,,, for the take-
mid-chord of 59 % station on the wing. The slat off configuration. In that case the forward displa-
rake with a height of 30 mm comprised 14 tubes, the cement of the transition location mainly on the
wing rake with a height of 160 mmn 21 tubes. flap but also on the slat leads to an increase in
the viscous effets which is not compensated by the
The scanivalves and the electronic equipment needed effect of Reynolds number. Thus the overall circu-
for the slat instrumentation were installed on a lation decreases for Reynolds numbers above 6.10'
special bracket which was attached to the slat as shown in fig. 15 where the Maximum Mach number
structure in the shroud area underneath the slat. on the slat is plotted versus the Reynolds number;
In order to avoid causing early transition on the
slat attachment line the belt on slat 2 was For the landing configuration transition on the
extended to cover its full length, as shown in the flap is less sensitive to Reynolds number since a
photographs of fig. 12. ONERA had built a mockup of velocity peak occurs at the leading edge and the
the slat, wing and flap nose in order to achieve overall circulation is almost constant for Reynolds
best fit of .the belt segments. The belt surfaces numbers above 6.10'.
were very smooth and the edges were faired by means
of a filler. For low Reynolds numbers laminar separation bubbles
are present on the three elements which increase
2.3.2. Flight Test Program the viscous layers and decrease the overall circu-
.......................... lation.
Three test flights were successfully performed in A good correlation is observed between the C,
the following sequence : values obtained in the HST and F1 wind tunnels at
6.10' which is confirmed by the pressure distribu-
Flight no 289 Landing configuration tion plotted in fig. 16.
Flight no 290 Landing configuration
Flight no 291 Take-off configuration Detailed boundary layer surveys have been performed
at 8 stations on the different elements as well as
From the fliuht measurements data points with in the near wake, as shown in fig. 17 for the take-
sufficiently stabilized flight parameters were off configuration. Reynolds number effect on the
selected for further evaluation : wing boundary layer are presented in fig. 18 for
the take-off and the landing configurations near
37 data points for the take-off configuration CL,,,. As shown on the figure a Reynolds number
36 data points for the landing configuration effect occurs for Reynolds numbers lower than
6.10'. For this angle of attack merging between the
The Mach number and the Reynolds number envelopes slat wake and the wing boundary layer occurs
which have been covered by the different test are downstream of station 3 (x/c = 0.5).
shown in fig. 13. The Mach numbers and Reynolds
numbers of 2D and 3D tests are correlated by simple No significant Reynolds number effects have been
sweep theory in order to compare 2D and 3D data found on the dimensions of the separated flow areas
under equivalent freestream conditions. As shown on in the slat and wing coves as well as on the extent
the figure 2D test were performed at Reynolds of the separation on the flap for the landing
numbers as high as those in flight and Reynolds configuration. For both configurations the stall is
numbers of between 1/3 and 1/2 of those in flight dominated by a wing trailing edge separation.
were achieved in the 3D test.
 16-7

3.2. Wach number effect in 2D flow that span section the nacelle generates large 3D
effects as shown on the boundary layer profiles,
The Mach number effects were investigated at a measured at mid chord (fig. 24). Wake surveys have
Reynolds number of 4.10' in the NLR HST wind tunnel
and at a Reynolds number of 9.10' in the F1 wind
been carried out with a five-hole probe at x/c ,.
1.05 in the ONERA F1 wind tunnel. An example of the
tunnel. effect of the Reynolds number on the wake profiles
is shown in fig. 25. Increasing Reynolds number
The Mach number dependence of the maximum lift induces a decrease of the wake size at high angle
coefficient C4,,, over the measured range of attack and less mixing between the flap and the
0.13 < M < 0.3 is shown in fig.19 for Re = 9.10'. A wing wakes.
large decrease of the C,,,, occurs for Mach numbers
above 0.22 for both the take-off and landing 3.4. Wach number effect in 3D flow
configurations. Pressure measurements show that the
strongest effect of Mach number occurs on the main Mach number effect has been investigated in the RAE
element on which a trailing edge separation appears 5m tunnel for a Reynolds number of 7.4;lO' (equiva-
at an angle of attack several degrees lower than lent 2D Re = 4.10'). As in 2D, increasing the Mach
-
act,,, number induces a decrease of C,,,, (fig.26) for the
complete model as well as a decrease of the maximum
The maximum local Mach number at the slat leading normal force coefficient of the 59 % span section
edge is plotted in fig. 20 for both configurations (fig.27). The compressibility effects are greater
and the two Reynolds numbers. It increases up to for the take-off configuration. The maximum Mach
values around M = 1.3 - 1.4 which indicates that number on the slat at the 59 % span section is
shock boundary layer interactions are likely to lower in 3D than in 2D.
occur on the slat which influence the slat boundary
layer and its wake. Thus, the stall of the wing is 3.5. Comparison between wing section data and 2D
certainly influenced by transonic flow phenomena flow data
which occur on the slat at high angle of attack.
Simple sweep theory has been applied to the 3D test
These have been clearly observed through the signal data in order to compare the data with the 2D
analysis of the Kulites which were located in the values. Fig. 28 shows that this simple correlation
slat leading edge region. works quite well for the maximum normal force.
However Reynolds number effects for the landing
3.3. Reynolds number effect in 3D flow configuration are different due to a different flow
behaviour on the flap in 2D with separation and in
Reynolds number effect has been investigated for 3D without reparation.
M = 0.25 on the A310 half model 139 within a large
Reynolds number range (3.4.10' Re, 9.2.10') in For the pressure distribution the same technique
the RAE 5m wind tunnel and up to Re, = 10.8.10' in using local sweep angle provides a good correlation
the ONERA F1 wind tunnel. as shown in fig. 29. Some differences appear on the
main wing near x/c = 20 % due to different shapes
Correlation between the results from the two wind for the 2D and the 3D models in the slat trailing
tunnels have been found to be very good. However edge junction region and on the flap. A good 2D-3D
.
small differences have been observed near CL,,, The correlation is also obtained for the section wake
data obtained on the half model 139 are also in drag up to CNmax (fig.30).
good agreement with those obtained in the DNW wind
tunnel on the complete model as shown in fig. 21. 3.6. Comparison between Plight and wind tunnel data
The evolution of C,,,, with Reynolds number for the The total lift in flight is plotted versus angle of
take-off and landing configurations in both wind attack for both configurations, and compared with
tunnels is summarized in fig. 22. For the take-off wind tunnel data in fig. 31. The lift coefficients
configuration with and without the nacelle as well at flight Reynolds numbers are in good agreement
as for the landing configuration with the nacelle with the wind tunnel data for a large range of
the increase of C,,,, is about AC,,,, = 0.04 over the angle of attack. Near maximum lift the lift coeffi-
whole Reynolds number range. For the landing cient of the aircraft is a little bit lower than
configuration without the nacelle the effect is that of the 3D model.
smaller with a tendency for the CL,,, values to
decrease for Reynolds numbers above 7.10'. The
effect of the nacelle on the C,,,, is only signifi- For the 59 % span section good agreement is also
cant for the landing Configuration. observed (fig. 32) in spite of the difference in
Reynolds number which confirms the small influence
of this parameter for this configuration. The
Considering the 59 %span section which corresponds differences near maximum lift which occur for the
to the airfoil tested in 2D the evolution of the landing configuration is due to the flap which
normal force CN versus Reynolds number near C,,,, is carries more lift in the wind tunnel than in
presented in fig. 23. These evolutions correlate flight, probably due to small differences in the
-
with those obtained in 2D. with an increase of the
CN up to Re, = 7.106 (equivalent Re,, 6.10') and no
significant effect for higher Reynolds numbers. For
slot geometry.
16-8

3.7. Attachment line boundary layer investigations important Mach number effect,
important effect of the nacelle on the CL,,, and on
The state of the boundary layer along the slat the boundary layer profiles at the (1 = 0.59
attachment line was investigated in wind tunnel and station (X/C = 0.51,
in flight near the 59 % span section. This has been the local sweep normalization process used for
achieved in wind tunnel using one hot film glued in 2D-3D correlations leads to a broad agreement
the slat near the hook for three spanwise stations between the pressure distributions at any speci-
(54 %, 59 % and 66 8 ) while for the flight test the fied C,, for the take-off configuration.
whole slate was equipped with two raws of 7 hot at moderate angles of attack, the take-off confi-
films located at I) = 59 % and q = 65 %. guration presents satisfactory agreement between
2D and 3D wake drag estimates,
Some typical hot film signals are shown in fig.33 attachment line transition occurs on the slat
for the take-off configuration with increasing from Re, = 7.10'.
Reynolds number at an angle of attack near maximum
lift. These measurements in the RAE 5m tunnel Plight test
indicate that the Attachment line on the slat seems
to be-fully turbulent at Re, = 7.3.10' which leads C,, a correlations good between F.T. and W.T.
to a R value in good agreement with the usual value data :
-
of 300. A laminar separation bubble was observed
for the same conditions which implies relaminariza- - for the complete aircraft at moderate C,
tion of the boundary layer on the upper surface. - for the q = 59 % section
In flight a laminar boundary layer on the slat Pressure distribution correlation between half
attachment line at the 59 % station was observed model and flight :
for higher Reynolds numbers indicating that less
disturbances cross the 59 % station in flight that - quite good for take-off configuration
in wind tunnel. These differences can be explained - some differences on wing upper surface and flap
by the fact that the belt on the aircraft slat upper surface for landing configuration.
gives a very smooth surface while the roughness is
higher for the wind tunnel model. Furthermore, the Hot Film signal analysis on slat :
relative size of the brackets is greater in the
wind tunnel. Sweep effect on attachment line tran- - both laminar and turbulent attachment line have
sition has also been demonstrated in flight as been observed at moderate angles of attack,
shown in fig. 34 where the reduction of the air- - b.1. relaminarization has been observed
craft sideslip angle by 11O.4 leads to a laminar - at high angles of attack a laminar attachment
attachment line and a decrease of the boundary line is detected in flight at higher values of R
layer thickness at the slat trailing edge. than in the wind tunnel tests.
4 . CONCLUSION All of this substantial set of measurements has
provided what is certainly the most comprehensive,
The success of this GARTEUR high lift research well integrated, and accurate body of data dealing
programme was made possible by the excellent with muti-element aerofoils and wings operating at
cooperation between the partners from industry and high lift that is available anywhere in the world.
research establisments. Form the extensive wind The programme of data acquisition has in itself
tunnel tests performed on 2D and 3D A310 high lift already provided valuable detailed comparisons
models in the NLR, DRA and ONERA wind tunnels as between three major European wind tunnels ; these
well as from the flight test conducted by Airbus have in turn led to improvements in thenderstanding
Industrie the following main conclusions can be and the application of test and correction techni-
drawn : ques. Furthermore the all-too-rare availability of
corresponding measurements made in flight are
2D test immensely valuable even when assessed purely on the
balance sheet of advances in measurement accuracy.
- moderate Reynolds number effect from Re, = 4.10' However, more importantly, GARTEUR member nations
up to 16-10', now have a better understanding of the flow about
- much more important Mach number effect, high-lift systems, which will facilitate the deve-
- the CCut is due to the stall of the main element lopment of improved designs, together with a
which is induced by the slat wake-wing boundary comprehensive data base available for the develop-
layer interaction, ment and validation of new computational techni-
- at high angles of attack, a shock-laminar boun ques. It has consequently been possible to set up
dary layer interaction is observed on the slat a successor to the Action Group tasked with taking
and induces the thickening of the slat wake, the first steps along the road to real improvements
- the take-off configuration presents an unfavou in the aerodynamic design of high-lift systems.
rable Reynolds number effect on the C,,,, from
Re, = 7.10'.
3D test

- moderate Reynolds number effect,

 16-9

I
I
I
I
I
I
I
I
I
I
I
I Exploiatory Groups
I
I
I
European Aerospace Industry I I

Fig. 1 GARTEUR organization

ESTABUSHYeST PERSON
ONWA Mr U.TNbert (chairman)
NLR Dr. B. vbn Den Eeq (Vicechaim)
DRA Ik.SP. Fiddes (partly)
Mr. P a Earnshew
DLR a. 5 stewleuer (pamv)
Dr.QRedeker
As YaCBanet
FOKKER Mr. PJ. Laan
Mr. JW Boer
BAe Mr. B. schclield Wty)
Mr. RW. Hesd (partty)
DORNlER Mr. AD. w l t e
Deutsche Airbus Mr.QDfuQd
Mr. J. W W
Mr. H. Schnieder
Al Dr. D. Schmitt
W. J.P. Robert

Fig. 2 GARTEUR High Lift Action Group schedule and members

16-10

MODEL EQUIPMENT MANUFACTURE TEST

130 PRESSURE TAPS M 3 MODEL NLR-LST
-~ _ _
13 KULITES NLR AND
'0 6 RAZOR BUDES NLR .HST
1 INTERNALB.L. DEVICE WITH EXTERNAL
2 D MODEL
~

SURVEYDMCE
6 TUBES FOR FLOW
VISUALIZATION
PLASTIC STRIP FOR
TRANSITION MEASUREM.
WITH EXTERNAL
2 B.L. RAKES IN COVES
SURVEYDMCE
4 SKIN FRICTION GAUGES
~-
I ! ~

I
HALF MODEL 602 PRESSURE TAPS MBBIDLR RAESM
(Cl FOR 'I = 0.59) FUSELAGE AND
HOT FILMS
+ ONERA F1
M 2 c=0.43m WITH EXTERNAL
RAZOR BLADES
1 INTERNAL B.L. DEVICE SHROUD
I I MBB-MODEL139 IN THE VISCOUS SHROUD

2 ROWS OF 7 HOT FILMS MBB - MOD

.~ AI

I T 1 A310
PLASTIC BELT WITH
70 PRESSURE HOLES
2 D.L. RAKES
6 ENDEVCOS
II LEADER
( MBB. ONERA.
AS) II
Fig. 3 Models, instrumentation and wind tunnels

Tu
J - TRACK
I

JACK

Fig.4 High lift system of half model 139

X/C = 0.5

Fig. 5 Viscous shroud for the half model 139

 16-11

mouman

WIND1IIIIIILWNTMU

Fig.6 Sketch of the NLR LST M3 model installation and external

probe traversing system

DlMENYON IN mm

BLOWING SLOT

IOTATION POINT

S L O ~ K D W A L L W I T ~ ~ PUTU
~TII

+- n n v l summi

Fig.7 Test set up with model M3 in NLR HST wind tunnel

16-12

Fig.8 Model M1 in ONBRA F1 wind tunnel with external Boundary

Layer device survey system

Fig.9 Half model M139 arrangement in the RAE 5m tunnel

 16-13

Fig.10 Half model M139 in the ONERA F1 wind tunnel with the wake
survey device

Fig.] General arrangement of flight test instrumentation

16-14
e

A
FLAP BELT

Pig.12 B e l t s on A310 w i n g for f l i g h t test

M 10
10 .I3 $8 20 .?? ,25 .?E M
25

ONEMF1 .e
20 -

15 -

-10.8
- 9.2

- 5.5
- 3.6

Pig.13 T e s t envelope
 16-15

M = 0.22

P
p'Q\
- -
~ 1.2

1.0 -
I
/
/I \ 1.2

1.0 -
J
/
p" lc'

k c 106 Rec 106

0.8 I b

o( =21.4O Y = 0.22 Rec = 6 106

---- Y3 - HST
- Y 1 - F1

I TAKE OFF

A-=I
I
i "(

Fig.16 Comparison of HST and F1 wind tunnels data

16-16

Toko off conflgurotlon

I I

Fig. 17 Example of boundary layer surveys - NLR data

Fig.18 Reynolds number effect on boundary layer data near C

,,
in 2D flow

---a-----
---+..
1A CL,,.^ = 0.5

---A ----_-e--
%\\
\

\
\

\
\
\
Q\
‘0

Y
c
0:1 0:
.2 0:s c

.Fig.l9 Mach number effect on ,Cmu in 2D flow

 16-17

A
1.4-

1.2-

1.0-

0.8 -

0.6-

N Y
C C
0.2 0.) 0.2 0.a

Fig. 20 Mach number effect on maximum slat Mach number near C,,,,
in 2D flow

c,
M -
TAKE OFF CONFIGURATON
0.20 Rec - - NACELLE ON
2.88'1V

Fig.21 3D wind tunnels data comparison

A
Roc/ loe
c

Fig.22 Reynolds number effect on maximum lift in 3D

16-18

-iGK€-iii1

'
NACELLE OFF

TAKE OFF co(IFKIuRLRK)u

tQ

R=/106

s 5 7 0 11
L -. 0
Re/@

I1

Fig.23 Reynolds number effect on section normal forces

TAKE Off CONTCXJRAnoN

BOUNDARY LAYER PROFCES AT W N C STATON X / C - 0.5 Il = 59%

Re, = 10 5.106
t t
?O

15

10

Y r 12 14 UIVC
d

Fig.24 3D boundary layer profiles at q = 59 %

- w c = 19.5. a c I 11. ---- w c = 14.6. - QC = l9.T.

TAKE OFF NACELLE OFF
-Za(rn) -2.
UPPER SKI€

0.1
Rec= 7.4 x 10' Roc= 10.5 E 10'

-0.1

Fig.25 Near wake profiles at q = 59 %

 16-19

NACELLE OFF
A310 HALF YOOEL M I 3 9

W / T RAE - SM Re, = 1.4 106

Fig.26 Mach number effect on maximum lift in 3D

A310 HALF MODEL M138

R% = 7.4 106 W/T RAE -5M

TAKE OFF - NACELLE OFF

0.15 0.30 0.55 0:sO 0.35

Fig.27 Mach number effect on maximum slat Mach number at

9 = 59 % near C,,,,

- 20-L
_._
.---.3 0 NORMALISED
SECTION q = 0.59

4% m.x
TAKE OFF
- - 0
0.1 U-?--"---; -
+---- +--+.-_
SDWORY
@ - ' - X w o I T A K E W / '
,/"--*--.--A
-.#
+' n4lE-m Fl

.18 .22 .25 .20 32.94 H~ 3.8 5.2 6.4 7.4 - s P 6

I , I ,
I I 1 ) I P
16-20

il:
I
I
I
ncp = 1
I ---_2D MOOEL
I
I
I
I ACP
I
I II I I
i
FLAP
I
I
I t I
IL-------
L: --

Fig.29 Normalised section q = 0.59 in 2D flow and wing flow -

Comparison of the pressure distributions near C,,,,

W / T ONERA F1

1 A CD = 0.02
Y = 0.25
ROso= 8.4 106

/ M=0.22
ROC = 6 106
o(
I I I
-5 0 5 10 15 20 25
Fig.30 Normalised section q = 0.59 in 2D flow and wing flow
comparison of the drag

Fig.31 Wind tunnel / Flight test comparison for the total lift
 16-21

I TAKEOFF I

0.2s f Y s0.27

Fig.32 Wind tunnel / Flight test comparison for the q = 0.59

normal forces

o( = 1 9 . 4 O M = 0.25

TAKE OFF CONFIGURATION - NACELLE OFF

W/T RAE 5M HALF MODEL M139

7 = 0.59 M = 0.21 Re = 15 lo6

- - A
0.0
A ---- 11.4

R=3.7 R=204 R=4.4 R=223

I 1
R=6.6 R=272 R=7.3 fi=285

I- GAUGE 1
q=54% 1-
I
1-
I
R=8.0 n=300
1
GAUGE3
I(=66%
.
1-
I

R = 10.1 = 337
I

REYNOLDS NUYBER R = Rec/ 106 0.6 0.7 0.8 0.0

Fig.33 Slat hot film signals. Fig.34 Flight test. Sideslip effect
RAE 5m tunnel. 3D model on attachment line boundary
layer and slat trailing edge
boundary layer
 17-1 I

AN EXPERIMENTAL INVESTIGATION OF THE OPTIMUM SLAT SETTING ON A

COMBAT AIRCRAFT MODEL

1. R. M. Moir

Defence Research Agency

Farnborough, Hampshire, GU14 6TD, United Kingdom

SUMMARY location of such an optimum is obviously important, and

is the result of two competing effects arising from the
I Tests have been carried out, on a combat aircraft model relative movement of the wing elements. Firstly, as the
I with high-lift devices, in the DRA Farnborough 5 metre two elements are moved together, load is increasingly
Dressurised low-sDeed wind tunnel. The deflection shared between the two surfaces, and the pressure
ingle and position' of a leading-edge slat were varied distribution on both elements is modified. This
and optimum settings established. The separate increases the load on the upstream element and
effects of Reynolds number and Mach number on overall decreases the load on the downstream element,
lift coefficient and on the optimum slat setting were reducing the likelihood of flow separation on the latter.
investigated. The results show that optimum
performance is achieved at a very high slat deflection In addition to the possibility of flow separation from the
angle and the performance is strongly influenced by upstream element due to increasing load, the wake from
compressibility effects. this element can merge with the boundary-layer on the
downstream element, thickening it and increasing the
LIST OF SYMBOLS likelihood of flow separation.
chord
lift coefficient It is apparent, therefore, that the high-lift device may be
maximum lift coefficient positioned in order to achieve a balance between the
adverse and beneficial effects, leading to the concept
pressure coefficient of an optimum setting.
pressure coefficient corresponding to local
Mach unity In addition to their application to civil transport aircraft,
free-stream Mach number high-lift devices are also used on some combat aircraft,
peak local Mach number not only for improved take-off and landing performance,
but also for enhanced manoeuvrability. However,
Reynolds number combat aircraft wings are usually of higher sweep than
free-stream velocity 'civil' wings and have lower aspect ratio as well as
slat overlap thinner aerofoil sections, this latter characteristic
slat gap leading to potential practical problems of
slat deflection angle accommodating mounting brackets for high-lift devices.
angle of incidence These features of the wing geometry result in highly
three-dimensional flows which are a severe test of the
wing leading-edgesweep two-dimensional concepts used to describe the flow
over civil wings.
1 INTRODUCTION
In general, increased sweep shifts the spanwise loading
The flow around multi-element high-lift aerofoil sections outboard, and also tends to shift the chordwise loading
is now relatively well understood. Extensive forward. The pressure gradients are made more
experimental studies, such as those in Refs 1 and 2, adverse near the tip as a result. Together, these
together with theoretical work involving CFD modelling effects would tend to encourage the stall to occur first
as in Ref 3, have identified the important features of near the tip, with a leading-edge type stall, which is also
such flows, at least in two-dimensions. The conclusions more likely with the thinner aerofoil sections used on
have been applied with some success3, in conjunction combat aircraft.
with concepts related to infinite swept-wing flows, to
describe the flow over wings of high aspect ratio and It would appear then that this markedly three-
moderate sweep, such as civil transport aircraft dimensional flow around combat-aircraft wings means
configurations, where root and tip effects are that earlier work on low-sweep, high-aspect-ratio wings
comparatively small in the context of the overall flow. may not give a good guide to the optimum positioning of
high-lift devices, and probably of leading-edge slats in
I Extensive research has also been carried out into the particular.
I mechanisms which govern the optimum relative
positions of the components of a high-lift wing (eg slat, Thus a need was perceived for fundamental research
wing, flap), but rather less on the effects of Mach into the genyal area of high-lift devices for combat
number and Reynolds number on this optimum. The aircraft wings, and with this in mind a model with a low
aspect-ratio, high sweep and high taper-ratio wing was
Copyrbht 0, Controller HMSO, London 7992 manufactured at DRA Farnborough, U.K. In the present
17-2

tests, the deflection angle of the slat and its position encompassed the range of Reynolds number and Mach
relative to the wing were varied in order to establish the number listed in Table 2; this was close to the widest
deflection and location, necessary to optimise CL= ,. range allowed by the operating envelope of. the tunnel,
The freestream Mach number and Reynolds number as shown in Fig 5, and by the design stress limit of the
were also varied independently in order to examine how model. Force measurements were made at small
these parameters affected the overall CLmax and increments of angle of incidence to beyond the stall,
optimum slat position and deflection. while surface pressures were monitored at 57.8% span
on the port wing and 82.6% span on the starboard wing,
2 DESCRIPTION OF MODEL as shown in Fig 1. These pressures were recorded at
a 16", 20" and 24" and at 3.2", 1.2" and 0.2" below the
5

The model, designated D R A Model 4954, is stall.

representative of a low-level strike-fighter. It was
designed for general research into high-lift devices on For the slat optimisation process, the slat was set at
low aspect-ratio swept wings. The general layout of the each of a number of positions in turn, and
model is shown in Fig 1, in which the principal measurements of forces and pressures were made over
dimensions are indicated; these are also given in therangeof a and hn, and Re.
Table 1. The high mounted wing has an 8% thick
supercritical section, and has provision for an 18% 4 RESULTS AND DISCUSSION
chord leading-edge slat and a 33% chord trailing-edge
flap, or a plain undeflected trailing-edge. As originally Figs 6a and 6b show an example of the variation of lift
designed, the slat deflection angle could be set at 25" coefficient CL with angle of incidence over the range of
and 35O, but for reasons discussed later, it was Reynolds number and for the range of Mach number
subsequently found necessary to make modifications to covered in the tests, respectively.
increase the available deflection to 45". The slat
position relative to the wing can be adjusted also, over The Figures show that a fairly high CLmax is achieved,
the range shown in Fig 2. This range was also increased albeit at a high angle of incidence, as would be expected
during the present tests. The slat brackets were of a for this low aspect-ratio configuration.
special profiled design (illustrated in Fig 3) which is used
to minimise the wake shed from the brackets. Such Fig 6a shows that CLmm is Reynolds number dependent
wakes are known to be entrained through the slat gap at and Fig 6b shows that it is very strongly Mach number
moderate to high a's and can induce premature flow dependent.
separation on the wing.
The foregoing suggests headings under which the
The flap also is adjustable in overlap and gap, as well as results may be discussed, ie:
having a number of deflection angles available. (i) Geometry effects
(ii) Reynolds number effects
The model has a fixed fin and an adjustable tailplane, (iii) Mach number effects
but the latter was not fitted for the present tests.
and these three aspects will be considered in turn.
The port wing, slat and flap have provision for
measurement of surface pressures by means of 4.1 Geometry Effects
spanwise tubes which have tappings drilled in them at
several spanwise stations, the unwanted holes being As mentioned in the Introduction, the low aspect-ratio
blanked off with adhesive tape, so that pressures may and high sweep of the present configuration should
be measured at one discrete station. The starboard typically result in a high loading towards the tip. Fig 7a
wing has tubes and tappings at one station only, near shows the pressure distributions at the two tapping
the tip. The pressures are scanned by on-board stations, just below Qrnax, and these do show much
Scanivalves, and converted to electrical signals by higher suction peaks at the outboard station.
pressure transducers. Comparison with results from a typical civil transport
model in Fig 7b, at similar spanwise locations, also just
For the present tests, the model was mounted on an below shows that the latter has a much smaller
underfloor six-component mechanical balance by way of variation on the slat between the tip region and the
a single main strut plus a jacking strut at the tail to station near mid semi-span. In both Figures the
provide pitch variation, as shown in Fig 4. The mounting pressure coefficient corresponding to local sonic
allowed a range of angle of incidence of -8" to +40", the velocity is indicated. . In the case of the fighter
angle being measured by on-board accelerometers. configuration the slat suction peak exceeds this value
in the tip region, and the implications of this local
3 TEST CONFIGURATION AND supersonic flow will be discussed in a later section.
CONDITIONS
4.1.1 Effect of Slat Deflection and Position
The model was tested with both the plain wing trailing-
edge and with the single-slotted flap set at 20" The original design of the model assumed that the
deflection, this being the highest angle at which optimum slat deflection angle and position would be
attached flow could be expected. The overlap and gap similar to that established on high aspect ratio, lower
were set at 0% and 3% respectively, relative to the sweep 'civil' wings. This resulted in the initial deflection
shroud trailing-edge. The model was tested with the slat angle of 25", although just prior to the first tests,
set at 25", 35" and 45" with the plain trailing-edge, and modifications were made to enable the slat to be set at
35" and 45" only with the slotted flap. an alternative deflection of 35". The first tests
commenced with the lower deflection angle (in
The tests were conducted in the DRA Farnborough conjunction with the plain trailing-edge),. but, as
5 metre pressurised low-speed wind-tunne15 and suggested by Fig 8a which shows the variation of CLma
 17-3
I
with slat position, it was quickly discovered that the 3 5 O slat may pose problems in this respect.
CLma was still increasing when the limit of the available Furthermore, it is also noticeable in Fig 9 that the slat at
slat adjustment range was reached. The rapid increase 45O deflection is rather more sensitive to position than it
in CLmax, seen in the Figure, as the slat was moved is at 35" and the lower slat deflection could be used at
away from the wing thus reducing the load carried by the an off-optimum position without incurring much penalty
slat, indicated that it was very heavily loaded, and this in % m a -
probably resulted in an early flow separation, originating
on the slat. This result prompted the abandonment of In order to determine the precise positions of the slat
the 25O slat in favour of the 35" deflection which which correspond to optimum CL^^^, values of overlap
hopefully would be sufficiently less highly loaded and and gap for a range of values of CL,= were read off the
would enable a peak CL^^ to be achieved much closer earlier plots of CLmax against slat position and plotted in
to the wing. the form of constant CLmax 'contours', as illustrated in
Fig 10, which is derived from Fig 9. Plots of this type
An example of the variation of CL^^ with slat position enabled the optimum slat location to be determined
for the 35O slat is also shown in Fig 8a. It may be seen within about f0.2% of overlap and gap. These data
that was now nearly 0.2 higher than for the lower enabled the variation of the optimum positions over the
slat deflection, but the position corresponding to peak range of MO and Re to be examined - this aspect is
CL^^ was still obviously further ahead of the wing. At discussed in the following sections.
this stage, further modifications were made to the
model, to increase the range of adjustment of the slat, 4.2 Reynolds Number Effects
enabling it to be moved further forward of the wing and
also, since it was obvious that the slat was still very 4.2.1 Overall Effects
highly loaded at 35O provision was made for an additional
increase in slat deflection to 4 5 O . The overall effect of Re on the CLmax for a fixed slat
position is illustrated in Fig 11, which also shows results
Due to insufficient tunnel time, a full range of slat for a typical civil transport model at a similar MO. It may
positions using the extended range for the 3 5 O slat, in be seen that the two configurations show a similar
conjunction with the plain trailing-edge, was not degree of sensitivity to Reynolds number.
completed, but tests on the 4 5 O slat covered a sufficient
range of positions to enable the position giving the 4.2.2 Effect on Slat Optlmum
highest value of CLma to be located as shown in Fig 8b.
It may be seen however that this maximum value is now As suggested earlier, the optimum position of the slat
lower than that for the 35" slat, suggesting that the may vary with flow conditions. In Fig 12, the variation of
optimum deflection is lower than 4 5 O , although the x, and z, is plotted against Re for the two slat
corresponding optimum slat position is at a fairly small deflection angles at MOo 0.22 . The Figure shows some
underlap relative to the wing leading-edge at the higher apparent movement of the optima with Reynolds
deflection. number, although the actual changes in x, and zg are
of the same order as the accuracy with which the
A feature of the 45O slat deflection is the large drop in optimum positions were determined. It is reasonable to
CL^^^ (about 0.4 relative to the optimum value) as 3% suppose that the effect of increasing Reynolds number
underlap and 1.5% gap is approached. This would be to generate thinner boundary-layers, which in
corresponds to a reduction in stall incidence of about 6 O turn would affect the degree of,wake/boundary-layer
for this case and is presumably associated with a more interaction. This would suggest that the optimum slat
rapid change to the pressure distribution, and hence the position should move closer to the wing as Reynolds
proportion of the overall load carried by the wing as the number was increased. However, the results do not
slat is moved away from it than was the case for the 35O give clear support for this hypothesis.
deflection. The configuration with a slotted flap shows a
I
much smaller loss in at this setting as will be Although the 3 5 O slat appears to show the larger
seen in the next section. movement of optimum position with Re , the sensitivity
of CL^^^ to slat position has been shown to be rather
I 4.1.2 Effect of Slat Posltlon less than that for the 4 5 O deflection, so the actual
variation of optimum CL^^^ with Re has to be
1

In order to explore fully the optimisation of the 3 5 O and considered in order to assess the relative loss of CL^^
4 5 O slats in conjunction with the single-slotted flap, associated with fixing the slat position for the full range
additional modifications were made to the model to of Re , and this is shown in Fig 13. The Figure shows
increase the range of slat adjustment in both x, and the increment in for the slat fixed in its optimum
z, , as the earlier measurements showed that optimum position for the lowest Reynolds number, compared with
CL^^^ tended to occur at more negative values of x, the corresponding increment when the slat is moved to
and lower values of z, than could be achieved on the the position giving the highest attainable CLmax as Re
model originally. is increased. The increment in CLmax for 8, E 35"
I relative to 8, a 4 5 O is also included.
Fig 9 shows typical examples of the variation of CLmax
I with slat position at the two slat deflection angles for the It may be seen that the potential loss in CLmax for a

1i slotted flap configuration. As with the plain trailing-edge

configuration the 35" slat achieves the higher CLmax,
but at the expense of an underlap greater than 4% -this
amount of underlap may not be possible on a real
aircraft for structural reasons. Similarly, it may not be
'fixed' 3 5 O slat is about 0.03, compared with a
're-optimised' slat. Also, the 'fixed' slat has a lower
CL^^ than the 4 5 O slat at the higher Reynolds numbers.
The 4 5 O shows very little loss in CLm= when fixed at the
same position through the Re range, although its CLmm
possible to incorporate practical slat brackets to is lower than that for the 're-optimised' 35O slat through
provide a 45"deflection in a thin fighter wing. Even the the whole range.
17-4

4.3 Mach Number Effects Configuration at the highest MO appears to coincide

with Cpmin reaching the limiting value suggested by this
4.3.1 Overall Effects criterion. It is postulated that the Mayer criterion may
describe a 'primary' inviscid effect in which local
The overall effect of the variation of MO on is velocity is limited by compressibility, while the more
shown in Fig 14 which also shows results for the civil general case of loss of CL^^ with increasing MO is a
transport model; these latter are at a rather lower Re. 'secondary' effect in which the boundary-layer is
The plot for the combat aircraft configuration is thickened by compressibility and results in earlier flow
characterised by a large reduction in CL^^^ between separation.
M O= 0.22 and 0.28, while the transport aircraft
configuration appears to exhibit a more gradual drop off It would appear then that the sudden loss of
in CLmax. This compressibility effect is presumably between the two highest freestream Mach numbers for
associated with high local subsonic and supersonic the combat aircraft is associated both with the spread of
flows, as suggested by Fig 7a. This is further illustrated local supersonic flow inboard on the slat and wing which
in Figs 15a and 15b which show the variation with MO of could give rise to compressibility-induced boundary-
peak local C, just below CLma on the slat and wing layer thickening resulting in an earlier stall, and also with
leading-edges for the fighter and transport a direct limit imposed on the maximum value of Cpmin,
configurations, at similar stations near the tip and near as suggested by the Mayer criterion.
mid semi-span. It is seen that for MO= 0.16 (0.15 for
the transport aircraft) the flow is wholly subsonic for 4.3.2 Effect on Optimum Slat Deflection
both configurations. At MO= 0.22 the flow attains
supersonic velocity on the slat near the wing tip for the Fig 17 shows the variatipn of CLma with slat deflection
combat aircraft, while the wing leading-edge is just angle over the range of M O ,for the plain trailing-edge
sonic, and the inboard station is entirely subsonic, as configuration.
are both stations for the transport aircraft (MO= 0.21) on
slat and wing. At MO= 0.28 there is supersonic flow on The CL,,,^^ data for the 25" and 35" are the highest
the slat and wing leading-edge at both the tip and achieved values, while those for the 45" slat are the
inboard station of the combat aircraft; in contrast the optimum values. Compared with the 35" slat CLmaxis
transport aircraft (MO= 0.27) just achieves supersonic lowerat 45" deflection for MO= 0.16 and 0.22, even with
flow at both stations, and that for the slat only. the 3 5 O slat in a non-optimumposition. At MO= 0.28 the
45" develops the slightly higher CLmax -this is possibly
because the 35" slat, being more highly loaded, suffers
It is interesting to note that although the 35" slat, being more from compressibility limitation at this higher MO.
more highly loaded had a much higher peak suction than
the 45" slat at the lower Mach numbers, at MO- 0.28 the 4.3.3 Effect on Slat Optlmum
peak suctions for the two deflections have very similar
values, at least on the outboard wing. This suggests In Fig 18, the optimum values of xg and z, are plotted
that there is an effective limit imposed on Cpmin , against MO for the 35" and 4 5 O slat deflections, at one
presumably by compressibility effects. This feature is value of Re . As with the variation with Reynolds
illustrated in Fig 16 which shows the overall Cpmin number, the movements are fairly small compared with
plotted against MO for both slat deflection angles at two the accuracy of the optimum position data.
slat positions for each. Also shown on this plot is the
variation of a maximum C, as derived by M a y d . It is to be expected that any effects due to
Mayer's empirical criterion states that compressibility depend on the relative loading of the slat
and the wing. Hence the movement of the 3 5 O slat
-CpM2 <1 optimum away from the wing initially and then back
towards it as MO is increased may be due to the
compared to the theoretical limit of 1.4. appearance of supersonic flow first on the slat and then
on the wing. The 45" slat may behave differently as
The Mayer expression gives C, = -12.75 at MOE 0.28 supersonic flow appears on the slat and wing roughly
which is very close to the apparent limiting value at this simultaneously. This will be discussed in more depth in
MO from the present tests, as seen in Fig 16, although the following section.
Mayer's result was obtained from tests on a simple
aerofoil at rather higher Mach numbers. This effective Fig 19 shows the overall effect of fixing the slat position
limit on Cpmin has implications for other aspects of the for the full range of MO,compared with re-optimisingthe
results, as will be seen later. slat at each MO. The relative effects for the two slat
deflections are shown. It is seen that the potential loss
It has been shown that supersonic local velocities exist in CL^^^ for the 35" is about 0.05 at the highest MO,
on the slat and wing leading-edges at the higher while the 45" shows an appreciably smaller loss of about
freestream Mach numbers. Using the expression given 0.01, and even the 're-optimised' 35" slat has a lower
in the Annex, the combat aircraft configuration can be CLma than the re-optimised 45" slat.
seen to generate local Mach numbers of up to about 1.5.
4.3.4 Variation of Local Mach Number
Comparison with the results for the lower sweep, high
aspect-ratio case also shown in Fig 16 shows that for In section 4.3.1 it was shown that the high levels of
this latter the peak local suctions are appreciably lower, peak suction on the slat and wing leading-edges were
with correspondingly lower Mlocal, and that in this case associated with local supersonic velocities, and in the
the Cpmin is not high enough to assess whether it too discussion on slat optimisation in section 4.3.3 it was
conforms to the Mayer criterion. On the other hand, the suggested that the optimum slat position could be
large loss of CLmax experienced by the fighter
 17-5 I

influenced by the variation of Mlocal on these This concept of a limiting value of Cpmin,and hence
components. Mlod adds a further parameter to the variation of the
optimum slat position with MO . As the highest
Figs 20a to 20f shows the variation of the peak Mlmd on freestream Mach number is approached the peak Mloal
the wing and slat for both slat deflections at on the slat and wing tend to have very similar values at
MO= 0.16, 0.22 and 0.28, at the outboard pressure the optimum slat setting, and this value is close to the
plotting station. Also shown on the plots are the limit suggested by Mayer. This effect is most noticeable
optimum slat locations as discussed in section 4.1.2. It for the 35O slat which has large differences in peak
will be seen that several of these plots have a rather Mlocal between slat and wing at the lower MO's. At
'confused' appearance, while others show a well ordered MO= 0.28 the load on this slat therefore appears to be
variation of peak Mioca! with slat position. The most limited to a greater degree than for the 45" deflection.
likely explanation for this feature is that due to the This is consistent with the earlier results for the
sharpness of the suction peaks near CL^^^ , it is quite variation of CLma with slat deflection and MO.
possible that there would not have been a pressure
tapping coincident with the maximum suction, and the Referring to the discussion of the movement of the slat
apparent value of Cpmin may well be in error by a optimum with MO in section 4.3.3, the increase of MO
significant amount. For example, if at MO= 0.22 the from 0.22 to 0.28 results in a much larger shift of the
true Cpmin is -15, but the maximum recorded value is optimum position towards the wing than for the increase
-14, the resulting error in peak1, M
I is about 0.05. It from 0.16 to 0.22. This, as shown in Fig 20f is where the
will be seen that adjustment of some of the points in the peak values of Mlocal on the wing start to reach the
peak Mlocal plots by increments of this order could 'Mayer limit', and the optimum is determined solely by
result in a much more ordered appearance. the maximum load on the slat, which itself is imposed by
a balance of the limiting value of Cpmin and the slat
Bearing this point in mind, the plots in Fig 20 do show wakdwing boundary-layer interaction.
some interesting features which reflect the overall
variation of Firstly, it is seen that peak Mlocal It will be noted that the foregoing only considers data
increases as expected with MO, from totally subsonic from the pressure plotting station near the wing tip. The
values at MOE 0.16, to a mix of subsonic and results from the mid semi-span station have not been
supersonic at MO= 0.22, to almost totally supersonic fully analysed, but due to the higher peak suctions in
values at MOP 0.28, as seen in section 4.3.1. At the the tip region, the limiting values suggested by Mayer
are not reached further inboard.
highest MO (Figs 20e&f) the maximum peak Mlocal
approaches 1.5 which corresponds to a value of Cpmin
which comes close to satisfying Mayer's criterion as 5 SYNOPSIS
discussed in section 4.3.1. As MO is increased it is
Fig 21 shows the overall variation of CL^^^ with MO
also obvious that the total variation of peak M l o d over
and Re for one slat deflection and position for the
the range of slat movement also increases. This is fighter configuration, with the corresponding plot for the
consistent with the expression for Mlocal given in the transport aircraft. This summarises the relative effects
Annex which shows that the calculated value becomes of Reynolds number and Mach number as well as the
more sensitive to Cp as MO is increased. This effect is fundamental differences between the two
probably partially responsible for the less well-ordered configurations. It must be borne in mind that any
examples of the variation of peak Mlocal with slat differences in absolute values of CL^^ between the
position. It is possible that it also has some influence two configurations can be attributed at least in part to
on the location of the optimum slat position, at the differences in flap size, type and deflection. The rate of
highest MO at least, as will be discussed shortly. change of CLmax with Reynolds number is remarkably
similar for the two configurations, and the only
It may also be seen that the variation of peakI , M
I for significant difference would seem to be the sensitivity to
the two slat deflections differs significantly, at MO.
M O= 0.28 in particular. At this MO the 35" slat
(Fig 208) shows quite a small range of peak Mio& but The increments in CL^^^ associated with slat re-
fairly large on the wing, while the 45" slat has a very optimisation (ie up to 0.05),discussed in sections 4.2.2
large range of peak MIA and the wing has a very small and 4.3.3, are seen to be significant in terms of total
range. It will be noted that in both cases the maximum Chaw
value of peak local Mach number is generally close to
1.5, which, as has been seen, approximately satisfies 6 CONCLUSIONS
Mayer's criterion, and this appears to impose a limit on
the amount of load the slat or wing can carry. The 35" Tests have been carried out in the DRA Farnborough UK
slat, being already highly loaded, quickly reaches the 5 metre wind tunnel on a model which represents a
'Mayer limit' as it is moved towards the wing, while the fighter configuration with a low aspect-ratio, high
less highly loaded 45" slat is able to increase its peak mounted swept wing with a supercritical section profile.
Mlwd a large amount before the this limit is reached. In
The model had two basic configurations, one with a
the latter case the wing has a peak M l o d near 1.5 over plain, undeflected wing trailing-edge, and the other with
the whole slat adjustment range, and so is near the limit a single-slotted flap. Both configurations had a leading-
suggested by Mayer, whatever the slat position. The edge slat with three available deflection angles, 25",35"
effect is further demonstrated as MO is reduced and 45", although only the two higher deflections were
(Fig 20d), where we see a much more 'balanced' tested with the slotted flap.
variation of peak Mloml on the slat and wing, and at
these lower freestream Mach numbers the 'Mayer limit'
is no longer being reached on either component.
17-6

The purpose of the tests was: edge, as established from the measurement of surface
pressures. The values of peak Cpmin measured at the
(a) to examine the overall characteristics of the highest MO were close to those predicted by the
configuration and how these were influenced by empirical relationship of Mayer.
scale and compressibility effects.
(5) The positions of the slat optima show small
(b) to determine the optimum deflection angle and movements with Mach number and Reynolds number.
location of the slat, and The increase in for the 35" slat if it is re-
optimised as MO and Re are varied is fairly large, with
(c) to investigate how these optima varied with Mach much smaller corresponding gains for the 45" slat.
number and Reynolds number.

To achieve these aims, overall forces and moments

(6) Consideration of the variation of the peak local
Mach number on the slat and wing leading-edges near
were measured, and surface pressures were recorded the tip shows that at freestream Mach numbers
at two spanwise locations. approaching 0.28,the optimum slat position coincides
with an apparent limiting value of peak MI,,^^ . This limit
The main conclusions obtained from the results are:
appears to be close to the value predicted by Mayer's
empirical relationship.
(1) The configuration develops a high CLmax but the
lift-curve slope is low as would be expected for the low
aspect-ratio, resulting in a very high angle of incidence Table 1
at CLmax .
Model Data
(2) The slat deflection angle for optimum is
between 10" and 20" higher than the typical value for a Quantity Symbol Value
transport aircraft.

(3) The 25O and 35O slat deflections result in optimum Span 2.700 m
positions which are at large values of underlap. The Area 2.1465m2
optimum for the 35" slat in particular is about 2% wing Aspect ratio 3.4
chord further forward than that for the 45" deflection. Centre-line chord 1.200m
For an actual aircraft design, the slat brackets
necessary to achieve either this large underlap or the Tip chord 0.390m
large deflection angles involved might be difficult to Mean chord 0.795 m
incorporate in a relatively thin fighter wing. However, 0.864 m
since the 35" slat achieves similar values of CLmax to
the 45" deflection even when at a non-optimum position, Taper ratio 0.325
it could be used at a similar value of underlap to that at Leading-edgesweep 40"
the 45" optimum with only a small penalty in . Slat chord 18%
(4) CL^^^ shows a similar sensitivity to Reynolds Flap chord 33%
number as for a typical transport aircraft. Mach number Thickness/chord ratio 8%
effects were quite large, particularly for values of MO
above 0.22. This appeared to be associated with the
appearance of local supersonic flow over the leading-

Table 2
Test Conditions and Slat Positions
e, = 35" e, = 45"

M Rx lo4 x,% z,% x,% Z,%

0.11 5.87 -2 1.5 -1 -1

0.16 5.87 -3 -1 -1 0
0.1 6 8.81 -3 0 -1 0.75
0.22 5.87 -3 0.75 -1 1.5
0.22 8.81 -3 1.5 -2 -1
0.22 11.23 -4 -1 -2 -0.5
0.28 5.87 -4 -0.5 -2 0
0.28 8.81 -4 0 -2 0.75
0.28 11.23 -4 0.75 -2 1.5
-4 1.5 -3 -1
-5 -1 -3 -0.5
R is based on E = 0.864m -5 0 -3 0
-5 0.75 -3 0.75
-5 1.5 -3 1.5
 e 17-7

REFERENCES

Smith, A.M.O., "Aerodynamics of High-Lift

Aerofoil Systems". in "Fluid Dynamics of Aircraft
Stalling", AGARD CP 102, April 1972

Foster. D.N., Irwin, H.P.A.H. and Williams, B.R..

"The Two-Dimensional Flow Around a Slotted Flap",
RAE Technical Report 70164 (1970)

Hardy, B.C., "An Investigation of a Proposal lo

Extend the Use of MAVIS to Three-Dimensional
Wings", RAE Technical Memorandum Aero 1840
(1980)

Moir. I.R.M., "A Description of a Typical Strike-

Fighter Model for the RAE 5 metre Wind-Tunnel
(model 495) and its Associated Half Model (model
21 13)". RAE Technical Report 81022 (1981)

Spence. A., Woodward, D.S.. Caiger. M.T.,

Sadler. A.J., Jeffery. R.W.. "The RAE 5 metre
Pressurised Low-Speed Wind Tunnel", ICAS Paper
B3-05. June 1978

6 Smith. A.M.O.. "Remarks on Fluid Mechanics of

the Stall", in "Aircraft Stalling and Buffeting".
AGARD Lecture Series No.74, March 1975

Annex
EFFECT OF WING SWEEP ON CRITICAL PRESSURE COEFFICIENT

A useful indicator of likely compressibility effects is the Taking y = 1.4, this gives:
local Mach number based upon the component of
velocity normal to a given sweep line. To obtain a value
for the local Mach number, it is necessary to make the
assumption that the flow at the spanwise station where
C; = - -+
0.6739 2
0.5283 m s A + 4
o ( 4 c o s A) .
the pressure is measured locally approximates to that 4
on an infinite sheared wing, ie the spanwise component ...... (A-4)
of velocity is constant and equal to V sin A . With this
assumption the local Mach number normal lo the sweep Thus, for low values of MO (a maximum of 0.28 here) it
line is given by isthe first term that is dominant and rapidly varying and
Cp is only made more negative by a relatively m a i l
amount by increasing sweep as indicated by the typical
values for the two major terms in (A-4).

Putting MI- - 1 we obtain

-
...... (A-2)
Expanding this for small MO; Hence the maximum error due to ignoring the term in A
is about 6%.

+ (72+ )1 " y ' cos% + O(Mi cos4A) . (A-3)

17-8

p i p - pressure plotting

Fig 1 Combat aircraft model (M495)

..

Fig 2 Slat adjustment ranges

L---

Model 495 in DRA 5 metre wind

tunnel
 r
17-9

15.c

Rx

11.2 -
-- -k---r---
I
I
I

10.1

8.8 -
I I
I I

5.8 -
5.1

Fig 5 Tunnel envelope and M495 test

conditions

2.8

2.6

CL

2.4

2.;

2. I

1. I I I I I I I I 1 1.8I
20.0 220 2L.0 26.0 28.0 30.0 32.0 34.0 36.0 38.0 1 0
LO 220 2.4.0 26.0 28.0 30.0 32.0 34.0 36.0 38.0 U-
a'

Fig 6a Variation of lift coefficient with angle of Fig 6b Variation of lift coefficient with angle of
incidence and Reynolds number incidence and Mach number
17-10

-2(

-lt

35deg S l a t
-It MO= 0.22
6
Re = 5 . 8 7 ~ 1 0 ,
Alpha = 33.0deg
-14
CP*

-12

P
-10
--
- T Inboard Station
Outboard S t a t i o n
-8

-6

-4

-2

L
I I 1 1 1 1 1 1 1 1 1 1 1 1 1
3
x/c

Fig 7a Comparison of chordwise variation of pressure

coefficient for inboard and outboard stations of combat
aircraft configuration

-2C

-lE 25deg S l a t
MO= 0.21
Re = 3 . 7 ~ 1 0 ~

-
-1Z Alpha = 19.5deg

Cp*
-14

-12

-10
e
-8

-6

-4

-2

0
.--------------
2 I 1 , 1 1 1 1 1 1 1 1 1

! -0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 I 4
x/c

Fig 7b Comparison of chordwise variation of pressure

coefficient for inboard and outboard stations of
transport aircraft configuration
 17-11

2.25 1 I

2.20

2.15 I
354.9 slat

2.10 j 8 xa
0 xs
.+ x*
--
-
-1)
-2*
-3*
254- slat

2.05

0’
r ’

2.00 I I I I , I I I ! I I I
0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0
-xstzs%

Fig 8 a Variation of CLmawith slat position, for

slat angles 25O and 35O, M0=0.22,
Re=5.87 x 106, plain trailing edge
configuration

2.20
I

I
2.15 -

2.10

2.05
-

-
2.70

2.75

2.65
-

.
35 deq
I 8ht

--

“:I
15 deq Ilat 0 Xs -I$35deg
2.00 - 2.60 . A Xs -5*
I

j 1.95 -

1.90 -

1.85 - 2.65

- 2.60 .
1.80

1.75 - 2.55

Fig 8b Variation of CLmawith slat position, for Fig 9 Variation of CLmawith slat position,
slat angles 25O and 35O, M0=0.22, 45Oslat, M0=0.22,R e 6 8 7 x 106, slotted
Re-5.87 x 106, plain trailing edge flap configuration
configuration
17-12

-1 -2 -
x.% -3
8, = 35" e, = 45"
Fig 10 Variation of CLmawith slat position,M,=0.22, R e 4 . 8 7 x 106

2.70 . 7
+ Tranaport Configuration

1.1

0.:

2.60
2
'3 4 5 6 7
Rsynolda Nvmber I
0
lo4
9 1 0 1 1 - 0.c

-0.:

Fig 11 Comparison of variation of CLma -1.m

with Reynolds number for combat
aircraft and transport aircraft -1.5

-2.0

-2.5

-3.0

I-3.5

-4.0

-4.5

-5.0

-5.5

-6.0
5 6 7 8 9 10 11 12 13
R.yno1da amber x 10-6

Fig 12 Variation of optimum slat position

with Reynolds number
 17-13

0.12

0.08
X 3 5 d q I l a t non rs-optimbed

+0 345deg
M . g a l a t re-optimiaed
a l a t non re-optimised

,
J
a
O.O'
,

I 0.01

2.75

I -0.0.
3 4 5 6 1 8 9 1 0 1 1 1 2 1 3
2.10
\
0 Fighter Configuration
Raynolds Number I lo-'
+ Transport configuration
2.65
Fig 13 Gain in CLmmobtained by re-
optimising slat position at each
Reynolds number 2.60

2.10
0.10
L '
0.12 0.14 0.16 0.10 0.20 0.22 0.24 0.26 0.28 30
Freeatream Mach Nmnbor

Fig 14 Comparison of variation of CLmm

with Mach number for combat
aircraft and transport aircraft

-24 -22
7 \
\ \
\
\
\ \
-22 \ -2 0
\
\

\
\ -18
-20
\ + 4Mq slat,inboard
0 4 5 d q slatroutboard
+ 4%- 8lat.inboard
0 4- ale.t,outboard
0 Transport slat,inboard 0 Transport wing,inbosrd

-18 -16

-16 -14

-14 -12

-12 -11

-11 I

I
....
4 0.16 0.10 0.20 0.22 0.24 0.26 0.28 IO
I I 0.16 0.18 0.20 0.22 0.24 0.26 0.28
Freestrem Wscb N d r meeotrmm uacb n&r

Fig 15a Variation of CPminon slat with Fig 15b Variation of CPminon wing with
freestream Mach number freestream Mach number
 17-14
e
-21 2.3
\ \
I \
\ \
\*CP. \
-22 2.3
\ -I/+\
\ \
\ \ 2.2
-2c
I
I
2.2 I
-18 a \ I
\ I
I I
I I
\ 2.1 I I
\ I I
-16 a \
a
\ a

8 \
\
J 2.;1
" I
-14 Fighter Configuration
0 Tramport Configuration
\
, I
I

2.01

-12
2.0(

-10
\
\
1.9s
\
\
\
-8 \
\ 1.90

-6 , .
0.12 0.14 0.16 0.18 0.20 0.22 0.24 0.26 0.28 , 1.85
0 35 40 45 50 55 60 65 10 15
f r s c s t r s m Mach NU&I e.. + iooy

Fig 16 Variation of peak suction with Fig 17 Variation of CLmawith slat

freestream Mach number, deflection and Mach number.
indicating 'Mayer Limit' Plain trailing edge configuration

0.01

0.01

-1.0 O.O(

-0.04

-2.5 -
B
-3.0 -0.08

c
:-3.5
-4.0 1- I -0.12

-0.16

-6.0
0.10
I 0.12 0.14 0.16 0.18 0.20 0.22 0.24 0.26 0.28 0.10
I -0.20
0.14 0.16 0.18 0.20 0.2) 0.24 0.26 0.28 0.30
frssatrsm Mach N&r freeatream Mach N U & =

Fig 18 Variation of optimum slat position Fig 19 Gain in CLmaobtained by re-

with Mach number optimising slat position at each
Mach number
 17-15

R -0.8
1O B

Wing J0.S

1
-6 -5
I I
-1
I
-3
I
-2
I
-1
I
0
t

1
.+ 1
-1 -3
%s + I s %
I
-2
I
-1
Slat
I
0
I
1
A

xs +zs%

Fig 20a Variation of peak local Mach Fig 20b Variation of peak local Mach
number with slat position, 8,=35O, number with slat position, 8,=45O,
Mo=0.16,Re=11.23 x l o 6 Mo=O.l6, Re=ll.23 x 1O6

11.3

Mlocal
- 1.2

- 1.1
Miocai
- 0.9
- 1.0

- 0.8 Z S 4
Wing
- 0.9

- 0.7

Fig 20c Variation of peak local Mach Fig 20d Variation of peak local Mach
number with slat position, 0,=35O, number with slat position, 8,=45O,
M0=0.22,Re-1 1.23 x 1O6 M0=0.22, Re=ll.23 x 1O6
 17-16

- 1.5
Mlocal

- 1.4
Wing
- 1.3 11.3

- 1.2 -1.5

hlocal
- 1.1

\
Wing -1.4

X
zs.1.5
d1.0 - 1.3

- 1.2
- 1.1

-1.0

- 0.9
"A
-
"7
X S ' 4
I

I I I A I I I I I I 4
-6 -5 -I -3 -2 -I 0 -6 -3 -2 -1 0 1
K,+Z,% xs +zs%

Fig 20e Variation of peak local Mach Fig 20f Variation of peak local Mach
number with at position, 8,=35O, number with slat position, 8,=45O,
M0=0.28,Re=ll.23 x 1 O6 M0=0.28, Re=ll.23 x 1 O6

2.9

2.8

j
2.81

2.11
+--
t
-//p\\\\ --- \

2.n
Fighter CoAfipuration \\ \ , \
\
\ \
Transport Configuration \
I

'2.6!

2.6(
0 MO- 0.22
A n r 0.28
0 M.- 0.15
2.5!
0 n.- 0.10

2.5C
11 13 15 11 19 21
4on.r Re I 10-6

Fig 21 Comparison of variation of CLma

with MO& Re for combat and
transport aircraft configurations
 18-1

AN EXPERIMENTAL INVESTIGATION OF ATTACHMENT-LINE TRANSITION

ON THE SLAT OF A COMBAT AIRCRAR MODEL

by

B. C. Hardy

Defence Research Agency

Farnborough Hampshire GU14 6TD
England

SUMMARY 1 INTRODUCTION

An experimental investigation into scale effect at low The introduction into service of facilities such as the DRA
speed and high lift has been carried out in the DRA 5 m, Farnborough 5 m Low-Speed, Pressurised Wind Tunnel has
pressurised wind tunnel on a subsonic strike-fighter model allowed the effects of Reynolds number and Mach number
equipped with slotted high-lift devices. The attachment- on aerodynamic performance to be evaluated separately. At
line boundary layer on the leading-edge slat was found to be the 5 m tunnel, particular attention has been paid to inves-
turbulent on the outboard part of the wing near maximum tigating the effects of Reynolds number on the performance
lift for a range of unit Reynolds number. An adverse of high-lift wings. The impetus for this research comes
Reynolds number effect on maximum lift was measured from the fact that, despite the ability to achieve high
which correlated quite well with the onset of attachment- Reynolds number through pressurisation, models of large
line transition. The conditions for onset of transition were transport aircraft can still only be tested to about VI full-
not consistent with the assumption of gross contamination scale. There is therefore a need to understand the
by the fuselage boundary layer, the attachment-line mechanisms that cause both the Reynolds number effects
boundary layer remaining laminar on the inboard slat to that occur within the tunnel operating range, and those
more than double the expected free-stream Reynolds which might cause similar effects at higher values of
number. It is suggested that this result is due to spanwise Reynolds number - up to full-scale. This paper presents the
variation in attachment-line position, which results in results of an investigation into one of these mechanisms,
suppression of the disturbances emanating from the root and its effects, conducted in the 5 m wind tunnel on a swept-
region of the high-lift wing. It is concluded that attach- wing strike-fighter model, Model 495 l . The model was
ment-line transition is a potentially significant factor in equipped with a leadingdge slat and single-slotted
wind-tunnel testing of high-lift wings equipped with trailing-edge flap, and the objective of the tests was to
leading-edge slats. investigate the effects of Reynolds number on the slat
attachment-line boundary layer, and to evaluate the
SYMBOLS corresponding overall effects on the high-lift performance.
wing chord given by csW cos $
streamwise wing chord The position of transition on an aircraft wing can have a
maximum lift coefficient significant effect on performance, the laminar-flow wing
trip wire diameter being an extreme example. The type and location of
acceleration parameter transition is scale dependent and consequently the correct
free-stream velocity simulation of full-scale aerodynamics in a wind tunnel
hot-film power dissipation requires that these dependencies are understood and
unit Reynolds number accounted for. Broadly speaking there are three distinct
attachment-line Reynolds number types of transition that may be significant in relation to
surface distance - Fig 1 swept wings at high lift:-
velocity normal to leading edge - Fig 1
velocity normal to attachment line (1) Transition following the development of
attachment-line velocity gradient instabilities of the Tollmien-Schlichting type. either in an
spanwise component of velocity - Fig 1 attached or separated shear layer (eg a laminar bubble).
distance normal to leading edge
model incidence (2) Transition following instability of the cross-flow
additional spanwise velocity velocity profile.
non-dimensional spanwise distance
kinematic viscosity (3) Transition following the development of
instabilities in a swept, attachment-line boundary layer.
sweep angle
attachment-line boundary-layer length scale It is the third of these, attachment-line transition, that is
the focus of the present investigation. In low-speed, high-
Copyright @, Controller HMSO, London 1992 lift testing it is usually assumed that the predominant scale
18-2

effect will be a favourable one, related to a general thinning The attachment-Line boundary layer is characterised by a
of the viscous shear layers as the Reynolds number is Reynolds number, R , which is a function of the spanwise
increased. This assumption implies that the variation of velocity at the edge of the layer, V , the corresponding
transition position is small over the Reynolds number kinematic viscosity, o , and a characteristic length scale,
range of interest. This is generally borne out by two- sotha hat R = V y / u . Aconvenientformfor yr.
dimensional calculations and experiments on high-lift representative of the boundary-layer thickness, is given by
configurations, where the characteristically sharp suction
peak close to the leading edge of each component tends to
‘anchor’ transition of the Tollmien-Schlichting type close
to the start of the pressure rise. However. this is not where (du/ds)#& is the velocity gradient normal to the
necessarily the case for threedimensional wings, and some attachment l i e at the edge of the boundary layer. For an
time ago Woodward2 carried out boundary-layer calculations infinite swept wing the velocity at the edge of the attach-
for swept, high-lift wings which indicated that attachment- ment-line boundary layer is given by V = V, = Q, sin , +
line transition could supplant Tollmien-Schlichting (where Qm is the free-stream velocity and 4 is the sweep
instability as the primary transition mechanism as angle) and the Reynolds number, R ,can be written
Reynolds number was increased. The movement of tran-
sition from close to the suction peak to the attachment line
with increase in Reynolds number was predicted to result in
a significant adverse scale effect on maximum lift due to the Here, Rsw is the free-stream Reynolds number based on a
increased boundary-layer thickness implied by the forward streamwise length, c,, , and U, is the nondimensional
movement of transition. attachment-line velocity gradient given by

These results had important implications for wind-tunnel

testing because the transition Reynolds numbers involved
were found to be much lower than those relevant to high-
speed flow, actually falling within the range covered by
where U, = a + +
cos and c = c,, cos . Equation (2) is a
convenient form for R and. although derived on the
modem pressurised wind tunnels. As a consequence of these assumption of infinite swept-wing flow, it seems reason-
predictions, a detailed experimental investigation was able to use it for.three-dmensional wings of moderate
carried out in the DRA 5 m tunnel using an existing model, aspect ratio, except probably in the root and tip regions.
Model 495. In summary, this investigation showed that
attachment-line transition occurred on the slat and that Transition becomes more likely as R is increased and
there was an *verse scale effect on maximum lift. Equation (2) shows that this can occur through an increase
Transition on the outer wing developed broadly as predicted in the free-stream Reynolds number or wing sweep angle or
by the existing criterion and correlated with the adverse through a reduction in the velocity gradient, U1 . The latter
scale effect. On the inner wing, however, transition was implies a reduction in the rate at which fluid is carried away
delayed to much higher Reynolds numbers, suggesting that from the attachment line and hence a thickening of the
contamination of the attachment line by the fuselage-side boundary layer there. The value of U1 depends on the
boundary layer did not occur as expected from previous chordwise velocity distribution near the attachment line
high-speed experiments. The hot-film instrumentation which is largely determined by the local surface shape.
indicated sudden switching between laminar and turbulent Fig 2 shows schematically the importance of this fact in
states with change in model incidence, which was difficult distinguishing between the high-speed or cruise condition
to explain, but development of transition with increase in and the low-speed, high-lift condition. For the high-speed
Reynolds number at fixed incidence was found to be case the figure shows the attachment line near the aerofoil
monotonic, as expected, and engenders confidence that the nose where the high curvature results in U1 values of
unexpected variations with incidence were not spurious. around 100, whereas for the low-speed case the attachment
l i e is shown lying further aft on the lower surface where
2 ATTACHMENT-LINE TRANSITION the curvature is much lower and U1 values of around 5 (for
the main wing) to 20 (for the slat) are more typical. The
The flow in the vicinity of the attachment line on a swept significance of these values is that, according to the
wing is illustrated in Fig 1, which is taken from Ref 3. The existing criterion for infinite swept wings, they imply
component of the free-stream velocity along the attach- transition at the attachment line for some typical high-lift
ment line gives rise to the attachment-line boundary layer. models in wind tunnels such as the DRA 5 m.
On an S i t e swept wing, the properties of this layer are
invariant with distance along the attachment line. How- 3 OTHER INVESTIGATIONS
ever, disturbances can propagate spanwise and, depending
on the conditions, instability and transition can occur. The mechanism of attachment-line transition was first
Furthermore, because the boundary layers over the wing recognised as such during attempts by aircraft manufacturers
originate at the attachment-line, transition to turbulence (notably Northrop and Handley-Page) to construct laminar-
there can result in turbulent flow over the whole wing. flow swept wings in the early part of the 1960s. Much less
 18-3

laminar flow than expected was found in flight and tunnel leading edge, relaminarisation was eliminated and a 15%
tests and it was discovered that disturbances generated in adverse scale effect on maximum lift was found. Garner
the wing-body junction triggered attachment-line observed that such large effects represent undesirable risks
transition and hence ensured that whole wing was covered in scaling wind-tunnel data to flight Reynolds number.
in a turbulent boundary layer. Subsequently, more detailed
investigationsbs were made in an attempt to clarify the 4 MODEL AND INSTRUMENTATION
conditions necessary for the onset of attachment-line
transition. This work ended in the early 1970s and little Although it was anticipated that investigation of the slat
further was done until P o I I ~published
.~ the results of an attachment-lie flow would present some problems because
extensive series of experiments on attachment-line of the proximity of the sharp slat heel ( Figs 2 and 3 ), it
transition and cross-flow instability in 1978. was considered that the predominance of slats on modem
aircraft made this a configuration of considerable interest.
These experiments were carried out using a sweptcylinder The choice of model was dictated by the requirement that
model which was mounted so as to eliminate the large attachment-line transition should occur within the
disturbances to the boundary layer which normally arise operating envelope of the 5 m tunnel. Twodmensional
from the root or wing-body junction. Poll was thus able to calculations coupled with simple sweep theory indicated
investigate the effects of artificially tripping the boundary that, on the basis of the existing criterion. Model 495
layer. Whilst largely c o n f i i g the earlier results of would satisfy this requirement. A full description of this
Cumpsty and Head7 and Caste#, which suggested that model has been given by Moir'. It is a large, complete
attachment-line transition following gross contamination model repres'entative of a subsonic strike-fighter aircraft,
could be characterised by a single value of R for all sweep built to approximately one third scale. The wing planform
angles and tunnel speeds, Poll was able to discover detailed is shown in Fig 3. A useful feature of the model for the
results for the variation of the critical R to produce present tests was the special slat-bracket design which, by
turbulence at a particular position on the attachment line aligning the bracket camber surface with the local flow
with varying trip size and location. This work was direction at high lift, greatly reduced the flow disturbances
extended by Paisley and Poll9 using a tapered, swept which usually arise from the separated flows found on
cylinder so that the three-dimensional effects of spanwise conventional slat brackets. For the tests reported here, the
taper were included in the experiment. This showed that the high-lift devices were deployed in a typical take-off
earlier 'infinite swept wing' results could be applied to the condition with the single-slotted flap at 20 degrees
the tapered cylinder, although there was some change in the deflection, and the .18% chord slat at 35 degrees deflection,
critical trip sizes. measured normal to the slat quarter-chord line. The slat was
positioned close to the optimum location relative to the
More recently, an experimental investigationlo into wing for maximum lift at this angle.
attachment-line transition on a model representative of a
transport aircraft has been carried out in the DRA 5 m The wings were equipped for pressure measurement at the
tunnel. The development of attachment-line transition on three spanwise locations shown in Fig 3. The distribution
the unslatted leading edge of this model was clearly seen of the pressure taps around the slat profile is also shown in
and significant differences from previous, infinite-swept this figure. The pressures were measured using rotary
results were found, but no adverse scale effect on maximum scanning switches in conjunction with Druck pressure
lift was observed. Relaminarisation in the favourable transducers, and overall forces were measured by the under-
pressure gradient between the attachment line and the peak floor. six component mechanical balance. The primary
suction location was suggested as the probable cause of the instrumentation consisted of three hot-film gauges located
latter result. This suggestion was based on evaluation of as indicated in Fig 3. These gauges (Dantec 55R47 glue-on
the acceleration parameter, K , which, in terms of velocity probes) were positioned as close as possible to the heel of
and distance along a streamline, is given by u(du/ds)/u2. the slat and so indicated the state of the lower surface
Strong favourable pressure gradients combied with low boundary layer. This had the dual advantage of eliminating
velocities lead to high value of K and a tendency to any disturbance to the upper surface boundary layer and
relaminarise. According to the existing riter ria^*'^*^^ the ensuring that the gauge signals reflected the state of the
values of K found in Ref 10 indicated that relaminarisation attachment-line boundary layer, at least at high lift. The
was likely. It was conjectured that the expected adverse latter follows from the low pressure gradients between the
effects would be postponed by relaminarisation to higher, attachment line and the gauge location. The gauges were
but less than full-scaleReynolds numbers, and it was glued into shallow recesses to minimise interference
pointed out that this could make extrapolation to full scale effects, but were in any case very thin (0.05 mm). Because
uncertain in some cases. Garner13 discusses the results of only a qualitative indication of the boundary layer state was
similar tests also carried out in the 5 m tunnel, but on larger sought, the hot-film gauges were operated in a constant-
models, which showed very large adverse scale effects on current mode when observing velocity fluctuations, though
maximum lift. On a simple wing a 7% loss was observed a useful additional indication of.boundary layer state was
and evidence was found that relaminarisation occurred over provided by the mean values of heat dissipation in the
the whole span at lower Reynolds numbers but only gauges. This quantity was determined by measuring the
outboard at higher values. Using a specially designed
18-4

heating cments required to maintain the gauges at the same the infite-swept assumption by defming a small
mean temperatures for all measurement conditions. additional spanwise flow component, so that
V=Q .. (1 +E) sin 0 . Thevalueof E varied with the
Measurements were made for the single model configuration assumed attachment-line position but exhibited a minimum
described above. at a free-stream Mach number of 0.22 and for a particular location and this was taken to be the appro-
for nine values of Reynolds number between 5.5 and 14.5 priate value. This modification resulted in a smooth
miUion per metre. At each Reynolds number the signals velocity distributions normal to the attachment line and
from the hot-film gauges, the pressure transducers and the allowed values of U1 to be extracted from the measured
mechanical balance were recorded for model incidences from pressures in a systematic way.
zero up to the stall.
6 RESULTS
5 ANALYSIS OF RESULTS
6.1 Attachment-line velocity gradient
Analysis of the measurements presented two main
problems, determination of the state of the slat attachment- Values of the attachment-line velocity gradient, U1 , were
line boundary layer from the hot-film results and estimation estimated from the measured pressure distributions, as
of the wrresponding values of U1 ,and hence R ,from the described as above. This was a somewhat imprecise
pressure distributions. As noted above, both mean and procedure because of the relative sparsity of chordwise
fluctuating signals were recorded from the hot-film gauges pressure holes but data was recorded for several Reynolds
and these allowed the results to be classified with some numbers, so that average values of U1 could be estimated
confidence as laminar, transitional or turbulent with with some confidence. In the case of an infinite swept
subsidiary distinctions between laminar-transitional and wing, U1 depends only on the chordwise position of the
turbulent-transitional usually possible- The characteristics attachment line. Therefore, to remove the primary effects
of the unsteady signals varied greatly with incidence. as of spanwise variation in lift coefficient and hence attach-
illustrated by the examples shown in Fig 4. These results ment-lie location, the estimated U1 values are plotted
were obtained at a unit Reynolds number of 10.5 million against chordwise position in Fig 6. The range covered by
and at moderate incidences of around U)degrees. where the the results is indicated by a shaded region - part of the
slat attachment line was close to the leading edge. Over the variation within this band is due to uncertainty in the
1 degree change in incidence covered by Fig 4, the estimated values and part is due to the fact that the flow
following interpretation was put on the traces; differs significantly from that on an infiite swept wing.
The total viujation is, in any case, only equivalent to an
Incidence Station 1 Station 2 Station 3 uncertainty in R of around f6% at the higher incidences
which are of most interest. Values are not plotted for the
21.5 degrees turbulent/ turbulent/ laminar region close to the maximum because of much increased
transitional transitional scatter. To the left of the maximum, the attachment l i e
22degrees laminar/ laminar1 laminar lies on the geometric uppa surface of the slat, while to the
transitional transitional right it lies on the lower surface. The results of a two-
22.5 degrees turbulent turbulent turbulent dimensional inviscid calculation are also shown in Fig 6,
and fall within the measured band, except for the highest
Typical variations of the mean signal with incidence are incidences where the attachment l i e is close to the slat
shown in Fig 5 for two unit Reynolds numbers. A histo- heel. There is some uncertainty in both results in this
gram style of presentation is used as this better reflects the region - the calculations were made with a smooth fairing
very sudden changes with incidence which were a feature of in the cove region and the experimental results rely on a
the results. The points which correspond to the instan- rather one-sided distribution of data, with the attachment
taneous signals shown in Fig 4 are indicated. It should be l i e close to the last pressure hole. However, on balance, it
emphasised that although the hot-film signals changed is felt that more weight should be given to the experimental
character rapidly with variation in incidence, the changes results because of the consistent trend shown over a wide
were precisely repeatable. range of experimental conditions.

In order to evaluate the attachment-lime velocity gradient., 6.2 Attachment-lie Reynolds number
U1 ,and Reynolds number, R ,using equations (2) and (3).
the flow in the vicinity of the attachment line was initially The estimated U1 values for each hot-film station were used
assumed to behave like that on an infinite swept wing with to determine the corresponding values of R from equation
the same sweep angle as the attachment line at the point of (2). The results are presented in Fig 7 in the form of the
interest. This assumption gave a discontinuous variation in free-stream unit Reynolds number needed to give an R of
velocity normal to the attachment line, kom which it was 300. This is the minimum value of R , according to Poll3,
difficult to estimate U1 accurately. In view of the highly for which complete turbulence may be present at some
three-dimensional nature of the €low on this wing it was distance kom a large disturbance on an Mite swept wing.
considered reasonable to allow for some divergence from Such a disturbance is teamed gross contamination and is
 18-5

usually regarded as arising from the turbulent boundary Two or three values were obtained for most Reynolds
layer at the wing-body junction. Fig 7 shows that, for numbers and these are plotted in the figure to indicate the
incidences near the stall, the critical Reynolds numbers are repeatability of the measurements. The data has been fully
within the operating envelope of the 5 m tunnel for all corrected for lift constraint and blockage effects. At low
three spanwise stations. For incidences higher than Reynolds number conventional scale effect is apparent,
25 degrees, transition should occur first at the inboard with the maximum lift increasing with Reynolds number.
station and progress outboard with increase in Reynolds However, a maximum occurs at around 9 million and the lift
number, due to the spanwise variation in chord. At slightly subsequently falls as Reynolds number is increased further.
lower incidence however, the effects of spanwise variation This adverse scale effect is not large - the maximum lift
in attachment-line location are large enough to overpower only falls by around 1%- but it is potentially significant.
the chord variation and reverse this trend. It is clear from If measurements had only been made for a range of
Fig 7 that, on the basis of Poll’s criterion for large disturb- Reynolds numbers at the left-hand part of Fig 11 (by
ances, attachment-line transition would be expected to varying tunnel speed in an atmospheric tunnel), then only
occur at high incidence in the present experiment at all the favourable trend with increasing Reynolds number
three spanwise stations and with an increasing tendency to would have been identified. Extrapolation of those results
turbulence inboard. to full scale could lead to a sigrufcantly optimistic
estimate of maximum lift. Because of the way in which
6.3 Hot film gauge results attachment-lie transition developed on the slat, with rapid
switching between laminar and turbulent states. it is
The varying state of the slat lower surface boundary layer difficult to link it categorically to the adverse scale effect.
inferred from the hot-film gauge signals is summarised in However, it is possible to identify Reynolds numbers
Figs 8-10. In a few instances the distinction between fully beyond which the flow at the mid and outboard stations is
turbulent and fully laminar states and the adjacent tran- predominantly turbulent at the highest incidences. These
sitional states was not clear-cut, but the difference between Reynolds numbers are marked on Fig 11 and indicate that
laminar (light l i e s ) and turbulent (heavy lines) states was the adverse scale effect on maximum lift is consistent with
marked. Turbulent flow occurs at low incidences because development of attachment-line transition, bearing in mind
the attachment line lies on the upper surface and there is an that the maximum sectional lift coefficient ( which might
adverse pressure gradient ahead of the gauge positions be expected to coincide with the position of onset of flow
which gives rise to T-S instability and transition of the separation ) occurs at around 78% semi-span.
chordwise flow. The incidence at which this gradient
changes sign is indicated in the figures. The rapid changes 7 DISCUSSION
in boundary-layer state with incidence are apparent, though
there is a clear tendency to increasing turbulence at higher In discussing the results of this investigation it is neces-
incidence. The trend with increasing Reynolds number at sary to consider whether phenomena other than attachment-
fixed incidence is more nearly monotonic, ie laminar, l i e transition could have affected the measurements.
transitional and fially fully turbulent, as expected. The R Therefore, values of the cross-flow Reynolds number and
contours in Figs 8-10 show that attachment-line transition acceleration parameter have been evaluated from the
did not develop in the way expected from Poll’s criterion. measured pressure distributions. Comparison with
There was little sign of turbulence up to R values of 450 on published critical values for these parameter^^.'^.^*
the inboard station and up to around 350 on the mid-span indicates that neither cross-flow instability nor relaminari-
station. Only on the outboard station does it appear that sation is l i e l y to have had any effect on the hot-film gauge
R = 300 was a reasonable transition criterion. However, it measurements at the high incidences which are of most
must be remembered that the flow is three-dimensional and interest. This was as expected, because the measurements
discussing the boundary-layer states in terms of the were made in the lower surface boundary layer so that, at
corresponding R values is only strictly justified for high lift, the pressure gradients between the attachment
infinite-swept conditions - where R is constant across the line and hot-film gauges were small. However, the possi-
span. On this wing the flow is strongly three-dimensional, bility of relaminarisation in the upper surface boundary
and thus R varies significantly across the span; it would be layer cannot be ruled out and this could delay the adverse
reasonable to expect this to affect the development of scale effect of attachment-lie transition on maximum lift
disturbances within the attachment-lie boundary layer. to higher Reynolds numbers. On the basis of the measured
Nevertheless. the delay in development of transition on the transition values of R and accepted critical values of the
inboard wing to such high values of R and the progression acceleration parameter, it is considered that relaminari-
from outboard to inboard are striking differences from the sation is unlikely to have greatly effected the present
expected behaviour which cannot be explained by such measurements. This conclusion is supported, of course, by
consider ations . the fact that the adverse scale effect observed was broadly
consistent with the onset of attachment-lie transition.
6.4 Variation of maximum lift with Reynolds number
It is with some confidence therefore, that attachment-line
The measured variation of maximum lift coefficient ( C k J transition has been identified as the dominant mechanism
with free-stream unit Reynolds number is shown in Fig 11.
18-6

on the slat at high lift. In addition to the evidence for this The conclusion that attachment-line transition was
already discussed, useful corroboration was provided by the responsible for the observed adverse Reynolds number
pressure distributions, which were consistent with the effect on maximum lift has important implications for
existence of a laminar separation bubble on the outboard wind-tunnel testing. The most significant of these is the
wing at low Reynolds number but not at higher values. additional uncertainty in extrapolating measurements to
However, bansition did not develop as expected according full-scale, as mentioned previously in discussing the results
to the accepted criterion for infiite-swept wings and of Fig 11. If the Ut values of around 20 found in this
possible reasons for this must be considered. Firstly, the experiment are typical, the attachment-lie flow on the
absence of transition on the inboard wing for values of R inboard region of the slat of all but the smallest transport
in excess of 450 shows that gross contamination from the aircraft is certain to be turbulent, because the resulting R
fuselage-side boundary layer did not occur. This represents values are high enough for natural transition to occur.
an important difference between the present high-lift Thus, artificially tripping transition at the attachment line
configuration and the accepted picture for cruise configur- in wind-tunnel tests might produce more representative
ations (Ref 3 for example), which suggests that such results in cases where the test Reynolds number is
contamination will always be present. A possible sufficiently high for this to be possible. However, this
explanation for this difference is that the reduced slat lift technique would only result in a fully representative flow in
near rhe wing root results in an attachment l i e position cases where the boundary layer did not relaminarise down-
close to the nose so that low values of R occur and disturb- stream of the attachment line, as this process is most
ances from the root are suppressed. Ref 6 shows that very unlikely to occur at full scale.
rapid decay of turbulence in a region of low R is certainly
possible. For cruise configurations the attachment line is 8 CONCLUSIONS
likely to lie close to the section nose over most of the span
so that large reductions in R over the inner wing are The attachment-line boundary layer on the slat of a high-
unlikely. lift strike-fighter model, Model M495. has been found to be
turbulent on the outboard part of the wing near maximum
Secondly, although transition on the inner wing is delayed, lift for a range of unit Reynolds number. An adverse scale
as described above, transition on the outer wing occurs effect on maximum lift has been measured which is consist-
more or less in line with the R = 300 criterion. According ent with the onset of attachment-line transition. However,
to Poll3, for low levels of free-stream turbulence or other it was found that the conditions for onset of transition were
disturbance, the attachment-line boundary layer should not consistent with the assumption of gross contamination
remain laminar up to R values in excess of 700. Free- by the fuselage boundary layer, the attachment-line
stream turbulence obviously cannot explain the present boundary layer remaining laminar on the inboard slat to
results, but surface irregularities and the slat support more than double the expected kee-stream Reynolds
brackets, which intrude slightly into the slat heel, are number. This is in agreement with other results'" for a
possible sources of disturbance on the model. Poll's simpler, unslatted high-lift wing, but contrasts with results
results3s4provide detailed data on the permissible size of previously obtained for the flow past non-lifting and cruise
two-dimensional roughness elements for infiiite-swept configurations. where it has been demonstrated that the
cylinders, and Paisley and Poll9 found that substantially fuselage side boundary layer invariably acts as a source of
smaller values were appropriate in the case of a tapered gross contamination. Suppression of disturbances in a
cylinder. The spanwise gradient of R in the present case region of low R near the root of the high-lift wing is
was found to be close to the value quoted in Ref 9 and, using suggested as the reason for this result. Transition on the
these results, it is estimated that a 0.06 mm diameter wire outer wing was found to occur at lower Reynolds numbers,
would be required to trip transition at the outboard station more or less in accordance with the accepted, R = 300 ,
for R = 300. The model surface was smoother than this but criterion. It is concluded that attachment-line transition is
it is conceivable that disturbances of sufficient magnitude a potentially significant factor in wind-tunnel testing of
could have originated from the pressure holes, tubing high-lift wings equipped with leading-edge slats. It is
installation or slat brackets. Such disturbances also suggested that artificial fixing of transition on the
provide the only obvious explanation for the sudden, but attachment line may sometimes improve the accuracy with
highly repeatable, changes in boundary-layer state which which full-scale flows can be represented, but that the
were observed. A quantitative argument cannot be made for utility of this technique will depend on an improved
this suggestion but the estimated trip sizes deduced from understanding of the effects of relaminarisation.
Poll's results and illustrated in Fig 12. are interesting.
According to Ref 9, lower values than these would be The present investigation has been of an exploratory nature
appropriate for the present model. particularly at higher and there is considerable scope for further work. For
incidence. The figure shows that the trip sizes vary non- example, an improved understanding is necessary of the
monotonically with incidence and it seems possible that flow in the vicinity of the attachment line when it lies
this could result in the type of behaviour observed. if close to the sharp slat heel. Similarly, more detailed
sufficiently large sources of disturbance were present. measurements of the attachment-line region close to the
wing root might help to explain the observed absence of
gross contamination. It is also important to understand the
 18-7

effects of free-stream Mach number and of slat position and 13 Garner. P.L.; Meredith, P.T.; Stoner, R.C., “Areas for
deflection angle, all of which strongly influence the future CFD development as illustrated by transport
performance of a slatted wing. The first of these might be aircraft applications”. AIAA Paper 91-1527 (1991)
particularly interesting because supersonic flows and
shock-wave boundary-layer interactions can occur on slats
and so it is conceivable that attachment-line transition
could alter the nature of this interaction from a laminar to a
turbulent type. perhaps resulting in a favourable Reynolds
number effect, in contrast to the adverse effect found at the
lower Mach number of the present tests.

REFERENCES

1 Moir, I.R.M., “A description of a typical strike-

fighter model for the RAE 5 metre wind tunnel (Model
495) and its associated half-model (Model 2113)”.
RAE Technical Report 81022 (1981)

2 Woodward. Dr D.S., “Unpublished RAE work”.

3 Poll, D.I.A., ‘Transition in the infinite swept

attachment-line boundary layer”. The Aeronautical
Quarterly, Vol XXX, November 1979, pp 607-629

4 Poll. D.I.A.. “Some aspects of the flow near a swept

attachment line with particular reference to boundary-
layer transition”. College of Aeronautics Report
7805 (1978)

5 Pfenninger. W.. “Laminar flow control -

Laminarization”. Special course on concepts for drag
reduction. AGARD Report 654 (1977)

6 Gregory, N.; Love, E.M.. “Laminar flow on a swept

leading edge”. NPL Aero Memo 26 (1965)

7 Cumpsty, N.A.; Head, M.R., ‘The calculation of the

three-dimensional turbulent boundary layer. Part III
Comparison of attachment-line calculations with
experiment”. The Aeronautical Quarterly, Vol XX,
May 1969, pp 99-113

8 Gaster, M., “On the flow along a swept leading edge”.

The Aeronautical Quarterly, Vol XVIII, May 1967, pp
165-184

9 Poll, D.I.A.; Paisley, D.J., “On the effect of wing

taper and sweep direction on leadmg-edge transition”.
College of Aeronautics Report 8435 (1984)

10 Hardy, B.C., “Experimental investigation of

attachment-line transition in low-speed, high-lift
wind-tunnel testing”. AGARD CP 438 (1988)

11 Beasley, J.A., “Calculation of the laminar boundary

layer and prediction of transition on a sheared wing”.
RAE Technical Report 73156 (1974)

12 Launder. B.E.; Jones. W.P., “Onthe prediction of

laminarisation”. ARC CP 1036 (1969)
18-8

External streamline

Fig 1 Flow in the vicinity of the

attachment line on a swept
wing (from Ref. 4).

A-A attachment tine

High speed

-
Fig 2 Attachment Line locations Low speed, high lift
for cruise and high - lift
configurations.

h Wing lower surface

Hot-film
Not t o scale

r) = 0.31

-
B slat brackets

--
-
Slat heel sweep o 4 1 deg Fig 3 Model 495 Geometric details.
Aspect ratio 3.4
Taper ratio 0.325

Slat pressure plotting and hot-film gauge locations

Pressure plotting positions

Gauge locations
Station 1
-0.04 Station 2
 18-9

a - 21.5deg
- Outboard station (3)
w - Mid station ( 2 )
w - Inboard station (1)

a - 22deg
R l m = 10.5.106 -
Fig 4 Typical hot film gauge
unsteady signals.
U 10-2s

[ 0.2 " 0 , t

- a - 22.5deg

Fig 5 Typical variation with

incidence of hot - film mean
power dissipation.

Fig 5 Typical variation with incidence of hot - film mean power dissipation.

Fig 6 Range of calculated

attachment line velocity
gradients for all stations for
several Reynolds numbers.

5
18-10

Fig 7 Variation with incidence of

the unit Reynolds number
required to give h = 300.

I--_-_ \ \ ------.-
l 1 '5-);urbuient - transit ionill
10.5- ---- -I---*--- - __-
Fig 8 Variation of Boundary - Turbulent

Layer type with incidence 9s----

and Reynolds number Laminar
8.5--------
at inboard station 1.
IS-- --
6.5----
Chordwlse pressure gradient

5.5
I I I I
I
5 10 15 20 25
30 a deg 35

Fig 9 Variation of Boundary -

Layer type with incidence
and Reynolds number
at mid station 2.

65-- -- --Chordwlse pressure gradient - - - - - ----- --- - --_

- ---
I I , I
5 10 15 20 25 35
30 a deg
 18-11

1 \as350 /

9.5---->

8.5
Turbulent
- ----- ---- bL--
- _ _ _ - - -- - - ---_-----
----
-- -
Fig 10 Variation of Boundary -
Layer type with incidence
and Reynolds number
-)Laminar - transitional at outboard station 3.

I 1 I 1 1
5 10 15 20 25 a deg 35

Fig 11 Variation of maximum lift

with Reynolds number.
0.991 1
Attachment -line
turbulent for ~ 3 0 . 8 2 7 3 0 . 6 1
0

0 0

0.98
I ’ I I
I I
I I I
1
I
4 6 8 10 12 14 16
Rtm

Fig 12 Variation of trip diameter

with incidence for large trip
- detector spacing.
 19-1

VISCO JS PHENO IENA AFFECTING IIGH-LIFT SYSTEMS AND SUGGESTIONS FOR

FUTURE CFD DEVELOPMENT

bY
RT. Meredith, Senior Specialist Engineer
Boeing Commercial Airplane Group
PO. BOX3707, Mail Stop OF-AR .
Seattle, WA 98124-2207
United States
SUMMARY
This paper describes a number of viscous phenomena 3) A 1% increasein take-off UTI is equivalent to a 2800
which affect the aerodynamic performance of high- Ib increase in payload or a 150 nm increase in range.
lift systems typical of commmialjet tiansports. The nature
of thesephenomena suggest a courseof action regardingthe Whilenecessary,high-lift systemsincreaseairplaneweight,

-
continuing development of computational fluid dynamics cosfandcomplexitysignificantly.Therefore,thegoalofthe
(CFD): in addition to the ongoing work of grid generation high-lift system designer is to design a high-lift system
and algorithm development, increased attention to funda- which minimizes these penalties while providing the r e
mental fluid mechanics is called for. quired airplane take-off and landing performance.

COUS
G- P HIGH-LIFT
CL lift coefficient
Chax maximum lift coefficient Eg. 1 shows some of the viscous flow features affecting a
CP pressure coefficient typical multielement lifting system:
dUldx local velocity gradient
K =(v/G)(d~/dx), relaminarization attachmentline transition from laminar to turbulent
parameter relaminarizationof turbulent boundary layers
Rbar =(U&n(A)/v)J(v/(dU/dx)), attachment transition of boundary layers from laminar to turbulent
line Reynolds number shock/boundary layer interactions
R, Reynolds number refetenced to chord viscous wake interactions
U velocity at outer edge of boundary layer confluent wakes and boundary layers
U, freestmm velocity separated flows
x/C non-dimensionalchord location
a angle of attack AU of these features play an important role in high-lift
6* boundary layer displacement thickness aerodynamicsand all are affectedby Reynoldsnumber. Of
rl semi-span location the above flow features, transition and development of
A sweep angle or attachment line sweep surface boundary layers have received the most attention,
angle though much work remains to be done in these fields. This
"c/4 quarter chord sweep angle is appropriate given that the maximum lift of a single
V kinematic viscosity element of a multielement high-lift system,as well as the
maximum lift of the system, are strongly dependant on
I " boundary layer state and development, as discussed by
High-lift systems are used on commercialjet transports to Smith [I]. However, understandingand modelingof surface
provide adequate low speed performance in teams of take boundary layers, while necessary, is not sufficientto enable
off and landing field lengths, a p p c h speed,and commu- thecalculationofmaximumlift,theoptimizationofpressure
Nty noise. The importance of the high-lift system is illus- distributions,and the optimization of geometricpositioning
trated by the following trade factors derived for a geneaic of high-lift systems in general.
large twin engine transport.
While it is commonly thought that increasing Reynolds
1) A 0.10 inmase in lift coeficient at constant angle of numberfieldsinmashglift,atleastthreeof thephenomena
attack is equivalent to reducing the approach attitude by listed above can result in decreasing lift with increasing
aboutone de-. For a given aft body-to-groundclearance Reynolds number, sometimes termed an adverse Reynolds
angle, the landing gear may be shortened resulting in a number effect, as discussedby Gamer, [21. The three
weight savingsof 1400 Ib. phenomena addressed in the remainder of this paper are:

2) A 1.5% increasein maximum liftcoefficientis equiva- viscous wake interactions

lent to a 6600 lb increase in payload at a fmed approach relaminarization
speed. attachment line transition
19-2

Shocffboundaty
layer lnteractknr

rbulent attachment
llnes and relaminarlzatlon

Fig. I Viscousjlow features Gecting high-lift systems

-W illustrated in Fig. 2 which shows a lift curve for a landing

To illustrate the interactionof a viscous wake with a lifting configurationhaving double slotted trailing edge flaps and
surface, the interaction between a wing wake and trailing the correspondingflap pressure distriiutions at two angles
edge €lapwill be considered. There are two major interac- of attack. The flap loading is greatly reduced at high angle
tionsbetween a wing wake and a flap. First.the wing wake of attack relative to the loading at low angle of attack.
may become confluent with the flap boundary layer. Sec- Contrary to conventionalwisdom,thepFessuredisvibutions
ond, and the subject of the remainder of this section, the and surfaceflow visualization(not shown)demonstratethat
proximity of the wing wake to the flap upper surfaceresults the decrease in flap loading is not associated with flap
in a damping, or supression, of the flap upper surface separation. In fact, thereductionin flap loadingresultsfrom
pressure distriiution, even when the viscous layers are not the growth of the wing wake displacement thickness with
confluent. increasing angle of attack.

The wing wake above the flap has a displacementeffect on Another way to change the thickness of the wing wake is to
the flow field which tends to suppress the flap pressure changetheReynoldsnumber. Assumeaflapisdesignedand
distribution. Generally,the bigger the wing wake, the more optimized at low Reynolds number. The €lap design and
suppressed the flap pressure distribution becomes. This is orientation will reflect the presence of the wing wake exist-
ing at low Reynolds number. As Reynolds number is
increased,the wing wake thins and the flap load increases.
'Ihismaycauseorincreaseseparationontheflap,decreasing
SlatMlinglMainlAft lift in the process. The adverse effect is more likely to
Mach = 0.20 happenforflaps designedto maximizelift at a fmed angle of
Rn I 9 million attack rather than for flaps designed to maximizeC b a x of
the system.
'IhisphemmenonisillustratedbyFigs. 3and4. Fig. 3 shows
lift curves at two Reynolds numbers for a 2D fourelement
( s 4 wing, vandmain flap) conf@mtionobtained during a
wind tunnel test at the NASA Langley Low Turbulence
Pressure Tunnel (LTPT). The higher Reynolds number
l ,High angle of
results in a large loss of lift at low to moderate angles of
attack. Also shown are calculated lift levels using a 2D
analysis code by Kusunose, [3] which employs a full
potentialsolvercoupledwithaviscousmodelpattemedafter
that used in the Eulet code of Drela [4]. While the lift levels
a~ high relative to the Wind.tunne1data, the trend with
Reynolds number is correctly predicted. The mechanism
/ 1
Walnflap Altflap involved is illust*tted in Fig. 4 which shows the calculated
/ t wing wake displacementthickness and flap pressure disd-
butions at the two Reynolds numbers at moderate angle of
attack At the lower Reynolds number, the wing wake
becomesvery thickoverthemain flap. AsReynoldsnumber
Fig. 2 Wing wakeflap interaction
is increased,the wing wake thinsandthemain flap loading
 19-3

increasesuntil the boundary layer separates, resulting in the thepreviouslymentionedmulti4ement fullpotentialsolver

high Reynolds number case in Fig. 4,where the Wing wake coupled to an integralboundary layer and wake model. The
is thin, the main flap is largely separated,and lift is greatly wake model contains several empirically deermined pa-
reduced. At high angles of attack, the thick wing wake rameters,one of which, termed Fw, influences the response
surpressesthe flap loading which alleviates flap separation of the wake to adverse pressure gradients. As Fw is in-
atthehigherReynoldsnumbersuchthatnoadverseReynolds creasedfrom0.20to0.45,thewingwakegrowsmcfferapidly
number effect is evident at C h a x . in the presence of the adverse pnxsure gradient imposed by
the flap. The thicker wake in turn suppressesthe loading on
theflap. Available wind funneldata suggestthe lowervalue
of Fw but the integral wake model may be too simple to be
Code Result.
Rn-2.8 mllllon--to generally valid.
Rn=12 mllllon \
e /
/
In the course of the above work, the literaturewas reviewed
toidentify expehental resultswhich would aid in improv-
NASAllangley ing the viscous modeling. It was found that the available
LTPT Test Rewlts wind tunnel data are limited and tests in which the viscous
Rn=2.8 mlllbn
Rn=12 mllllon
wakes were measured,in termsof totalpressureandvelocity
profiles, and turbulence quantities, more limited still. The
CFDdeveloperis thus forced to adjust the viscous modeling
to match surface pressure data rather than the viscous fea-
tures the code is attempting to capture. To remedy this

/
t
4-Element Alrfoll situation, much experimental work must be done to obtain
SlatMllngNanelMaln the type of data necessaty for continued development of '

CFD for high-lift applications.

-
a dog. As discussed by Smith [l],it is very beneficial, in terms of
developingmaximumlift, for an airfoilor elementof a high-
Fig. 3 Adverse Reynolds number of d e c t on lifr lift system to have a laminar boundary layer up to the point
of pressure recovery, the idea beiig to begin the recovery
A final example of the wing wak- interaction is p m with 8s thin a turbulent boundary layer as possible. There
vided in Fig. 5 which shows the influence of the wake fore, all phenomena which affect h e state of the upper
modeling on the flap pmsure distributionsas calculatedby surfaceboundary layer can be expected to play a role in the
Calculated Results

/
VanelMaln Geometry Wing Wake
Displacement 1k%Rnz2.8 mllllon
Thlckness
8'
Rn=12 rnllllon

-
T n Vane & Maln Flap Pressures

t 4

XIC
Fig. 4 Wing wakeflap interaction at low and high Reynolds number
19-4

Full Potentlal Code wfth Integral Viscous Modellng

0- -
f-
' Vane & Maln
Cp Dlstrlbutlon
Vane Flap T

cP
Fwz.45

Maln Flap' \\

=.20

xlc
Fig. 5 Effect of wing wake modeling on wake growth andjlap pressure distributwn

maximum lift developed. The influence of boundary layer aaachment line boundary layer. However, under the right
stateon C h a x is shown schematidy in Fig. 6. Refening conditions,the turbulent boundary layer downstream of the
to Fig.6, C b a x rises with increasing Reynolds number attachment line may become laminar, a process termed
until the initial upper surface boundary layer begins relaminarization. In subsonic flow, relaminarizatiOn may
aansitioning from laminar to turbulent causing a loss in occur under the influence of favorable pssure @en& at
C m with firrchex increases in Reynolds number. After sufficiently low freestream Reynolds numbers.
thetransitionprocess is completed,C~axonceagainrises
with increasing Reynolds number. Two phenomena which
cancausesuchbehavior~re~onand~hment
line transition. W e the primary subject of this section is
-er
. . ' n, attachment line transition, the subject of SlatMnnglFlap (Wlng shown)

thenext section, will be touched on in the following discus- Landing Flaps, a E 22"
sion

-
TURBULENT

Atl.dmen(
un

-
CL
I..
FICTOR

tar W O
G: Fig. 7 Typical potential flow surjhce streamlines
I I l l 1 1 1 1 1 I I II'JJJ

Fig. 8 shows the trend of C h a x with Reynolds number for

a simple, swept wing tested jointly by W i n g and Japan
Airaaft Development Corporation (JADC) in the RAE
Farnborough 5m pressure tunnel. It is seen that C b s
increaseswithincreasingReynoldsnum~r.reachesamaxi-
Fig. 7 shows the potential flow surface streamlines for a mum, and then declines. dropping 7% from the peak value
sweptwing. Ihe attachment line. a lineof spanwiseflow on at the highest Reynolds number achievable. Decreasing
the lower surface well M i n d the wing leading edge, is C b s with increasing Reynolds number has been ob-
clearly shown.. For a 3D swept wing. the upper surface served before, as discussed by Woodward, [SI but the
boundary layer state is usually determined by thestate of the magnitide of the observed loss was surprising.
 19-5

calculated spanwise distribution of K 85 a function of

Reynoldsnumber. It isseenthat r e l a m i t i o n . asdefined
by -3E46, is likely at Reynolds numbers less than about
12 million. Also. due to the spanwise distribution of pres-
sure gradients. r e l a m i o n should first cease inboard
with increasingReynoldsnumber, aresultin agreementwith

/
the experimentaldata.

I 121

Reynolds No.

An investigation of the phenomenon proved inteFesting. 0 S ' 1 . 8 ' 1

0.0 0.2 0.4 0.6 0.8 1.0

Surface oil flow visualization showed a laminar separation rl
bubble on the upper surface over the entire span at low
Reynolds number. The laminar bubble began to disappear Fig. 9 spamvk variatwn of the rdamharizath
parameter K for a simple swept wing
on the inboard section of the wing as Reynoldsnumbex was
inrreased At the highest Reynolds number, the laminar
bubble was evident on only the outboard third of the wing. Additional data confirming relamharimionwere acquired
duringa subsequenttest and arebriefly discussed in the next
One mechanismconsideredwas the transition of the attach- section.
mentlinefrom laminarto turbulentwithincreasingReynolds
number. This mechanism was ruled out for h e ~easons. Much work remains to be done to understand and be able to
First,the observation of laminar bubbles outboard but not model relamharkation. For instance, the relaminarization
inboardwouldimplyaturbulentattachmentlineinboardand parameter K does not reflect boundary layer characteristics
a laminar attachment line outboard, an unlikely situation. nor the chordwiseextent of the favorablepressure gradient.
Second,a Wire trip was placed on the inboard lower suface, Also, cross flow instabilities,a function of wing sweep,can
normal to the attachment line, to cause attachment line beexpectedtoinfluencethepropensityfor relaminarization.
transition but no dif€emws in flow charactexistics were Finally, the rapidity with which the boundary layer
obsenfed.Finally,calCulatianof theattachmentlineReynolds retransitions to turbulent flow certainly has an impact on
number Rbar, after Poll [a, gave R-240 at the lowest maximumm
Reynolds number suggesting that the attachment line was
turbulent for all Reynolds numbers tested.
As previously mentioned, the state of the attachment line
The most likely mechanism is now thought to be usually determines the boundary layer state on the upper
relaminarization. At low Reynolds number, the turbulent surface. The attachment line on many high-lift devices is
boundary layer flowing away from the attachment line fat behind the leading edge in a region of low curvature and
becomeslamiduetotheactionof theexmely favorable subsequentlysmall velocity gradients which favors devel-
. 'on opment
pressure gradient around the leading edge. As.Reynolds
numberincmses,thespanwiseextentofthere~
of a turbulent attachment line. In the absence of
sufficiently large pressure gradients and at high enough
decreases. The change from amostly laminar initialupper €ieemamReynoldsnumbers,re~onasdiscussed
surface boundary layex to a mostly turbulent one accounts above is unlikely to occur. Therefore, in the absence of
for the Observed 10s in C m relamharimion. a turbulent attachment line results in a
turbulent boundary layer over the entire upper surface.
There have been severalparameteas proposed as indicators
of when relaminarizationis likely. One parameter,K, after After the experiment discussed above, a new leading edge
Launder and Jones m, wmlates reasonably well with the shape was designed for the model as shown in Fig. 10. The
experimentunder discussion. A value of m3E-06 is some primary desire was to avoid rehmhrhtion. or at least
timespropsedasarelaminarizationaiteria.Fig.9 showsthe confine it to lower Reynolds numbers, so that the trend of
19-6

C b , with Reynolds number could be established with a Fig. 12 shows the calculated spanwise distribution of the
fully turbulent upper surface boundary layer. To this end, relaminarizationparameterKforbothH2Oand~O.5atlow
the curvaturearound the leading edge was reduced in ordw and high Reynolds numbers. H20.5 has greatly reduced
to lessen the favorable pressure gradients promoting valuesofKre~tivetoII20withKgenerallylessthan3E-06,
relaminarization. However, the resulting shape had in- indicatingthatrelaminarimtionis much less likelyforH20.5
creased curvature in the vicinity of the attachment line, than for H20.
increasing the likelihood of a l a m i i attachment line.

10 -
8-
Kr(04 *
6-

4-

2-

0.0 0.2 0.4 0.6 0.8 1.0

Fig. 12 Effect of leading edge shape d i j i i a t w n

Fig. 10 Leading edge shape d @ a t w n to reduce on the relaminarimtionparameter K
velocity gradients

The model was again tested by Boeing/JADCat the RAE 5m Fig. 13showstheattachmentlineReynoldsnumberRbarfor
pressure tunnel. Some results are given in Fig. 11 which both H20 and H20.5, calculated at about 20% span, as a
showsC b , asa functionof Reynoldsnumber for both the function of freestream Reynolds number. H20.5 exhibits a
initial design WO) and the new design (II20.5). H20.5 175 unit reduction in Rbar relative to H20 at all fieestream
exhibited characteristicssimilar to H20 in that C h , first Reynolds numbers. Using an Rbar of 240 as an indicatorof
increases with Reynolds number, reaches a maximum, and a turbulent attachment line, it is Seen that H20 should have
then decreases. However. relative to €I20 characteristics, a turbulent attachment line over the range of Reynolds
the Reynolds number for peak C b , for H20.5 was much numbers W i g considered while H20.5 should have a lami-
lower and the decrease in peak C h a x was much greater, nar attachmentline at low Reynoldsnumber and a turbulent
15%asopposedto7%,althoughthedropinCLmaxforH20 attachment line at high Reynolds number. These calcula-
was limited by the maximumReynolds number attainable. tions are in reasonablygood agreement with the experimen-
Also,H20.5 showdaresumptionofincreasingChax with taldata.
Reynolds number at the higher Reynolds numbers tested.
While the adverseReynolds number effect on H20 is due to
cessation of relamhukm'on.for H20.5 it is believed to be Inboard S.cclon, (1 I 0.20

due to attachment line transition.

U.sh=O.zO
(-don at Rdunlnrlndon) 4 6
Reynolds no.
8
- mllllons
10 12 14

Roynoldr No.

Fig. I1 Effect of leading edge shape d i j k a t w n Fig. I3 Effat of leading edge shape d i j i i a t w n
on C b vmiclrion with Reynolds number on the attachment line Reynolds number Rbar
 19-7

A final bit of evidence was provided by arrays of surface a high-lift system of fmed size as the system becomes
mountedhotfilmsensorsinstalledatthreespanwisestations more point designed.
around the leading edges of both H20 and H20.5. The hot
film data, derailed discussion of which is beyond the scope 4) Current CFD tools do not model the physics with
of this paper, support the above conclusions. The hot film enough accuracy and/or in enough detail to reliably
signals for H20.5 showed a laminar attachment line at low predict Reynolds number effectson CL and CL^^ of
Reynolds number and a turbulent attachment lie, with no highly loaded high-lift systems.
relaminarizationof the upper surfaceboundary layer, at high
Reynolds number. The hot film signals for H20 confirmed 5) The existing experimental database, in terms of the
that the attachment line was turbulent over the range of number of revelant experiments and the type of data
Reynolds numbers tested and that the adverse Reynolds obtained, is inadequatefor the purpose of improvingthe
number effect was due to cessation of relaminarization. ability of high-lift CFD tools to predict Reynolds num-
Also,the&tashowedthatC~m~isaffectednotonlybythe ber effects.
presenceor absence of relaminarimtionbut by thechordwise
extent of the relaminarized flow which, l i e most Viscous To advance the state of the art in high-lift aerodynamics,
phenomena, is strongly a function of Reynolds number. the following recommendations are made:

In summary, relaminarizationand attachment line transition 1) Continueddevelopment of CFD tools is called for as the
have a very largeeffect on high-lift system performanceand likelihood of developing wind tunnels for jet transport
greatly reduce the confidence with which wind tunnel data configuration development testing at flight Reynolds
are scaled to flight Reynolds numbers. While unlikely to numbers is remote.
replace the wind tunnel, the development of CFD tools
which model these phenomena will enable the design of 2) The funher development of CFD requires that the
more efficient high-lift systems and mitigate the risk of &reatest weaknesses of current CFD tools recieve the
beiig surprised during flight test greatest attention. This means that increased work in
fundamental fluid mechanics. rather than in grid gen-
"OC eration or algorithm development. is warranted. Such
The performance of high-lift systems depends greatly on a research should be directed towards the understanding
variety of viscous flow phenomena which are strongly a and modeling of:
function of Reynolds number. Three of the many phenom-
ena havebeenaddressed in this papex turbulence
boundary layer and wake development
viscous wake interactions viscous wake interactions
relaminarization transition
attachment line transition relaminarization
s e w flows
Thesephenomenaare of more than academicinterestaseach confluent flows
can lead to advetse Reynolds numbex effects where lift at
constant angle of attack (CL)and/ormaximum lift ( C h a ~ > 3) To accomplish the above recommendations, additional
decrease with inczeasing Reynolds number. experimental data must be obtained. The type of data
needed will require experimental reSeacchezs to focus
Based on the material presented in this paper. and the less on testing ahplane configurations and components
research on which it is based.the following conclusionsare and more on experiments designed to answer specific
dram: questions in fluid mechanics.

1) Adverse Reynolds number effects on CL and C h a x High-lift aerodynamics remainsone of the most interesting,
are not uncommon. challenging,and rewarding of engineering disciplines. It is
hopedthatresearcherswillembracetherecommendationsin
2) Advetse Reynolds number effects on CL and C u this papw,advances in the understanding and modeling of
canbelarge. the viscous phenomena discussedwill benefit not only CFD
for high-lift aerodynamics but CFD and fluid mechanics in
3) Designing and testing at less than flight Reynolds general.
number is not in general consemtive and may result in
expensive airplane modifications during flight testing ACKNOWLEDGEMENTS
and/or an airplane with paom than expected perfar- me author wishes toexpress his gratitude to his colleguesin
mance throughout its economic lie. This becomes the Advanced Technology & Design group of the Boeing
morelikely ifincreasedperformanceisdemandedfrom Commercial Airplane Group. The author also thanks the
19-8

engineers and staffs of the NASA/Langley Low Turbulence Aircraft Applications”, AIM-91-1527. 1991.
PressureTunnelandRAE/Famborough 5mPressureTunnel 3. Kusunose. K., Wigton, L., Meredith, P.T., “A Rapidly
for their assistance in acquiring the high Reynolds number ConvergingViscous/Inviscid Coupling Code for Multi-
data discussed in this paper. Finally, the author wishes to ElementAirfoil COnf~gUratiOnS”,AIAA-9 1-0177.199 1.
thank Mr.Amano, Mr.Nakamura, and others of the Japan 4. Giles, M., Drela, M.. Thompkins Jr., W.T., “Newton
Aircraft Development Corporation and its member compa- Solution of Direct and Inverse Transonic Euler Equa-
nies (Fuji Heavy Industries, Kawasaki Heavy Industries, tions”,AIAA Paper 85-1530.1985.
Mitsubishi Heavy Industries) for their assistanceand partici- 5. Woodward, D.S., Hardy, B.C., Ashill. P.R., “Some
pation in the RAE testing discussed in this paper. Types of ScaleEffect in Low-Speed, High-LiftFlows”,
ICAS Paper 4.9.3. Jerusalem, 1988.
REFERENCES 6. Poll, D.I.A.. “Transition in the Infinite Swept Attach-
1. Smith, A.M.O., “High-Lift Aerodynamics”, Wright ment Line Boundary Layer”, Aeronautical Quarterly
Brothers Lecture, AIM-74-939.1974. Vol XXX pt. 4, pp. 607429,1979.
2. Gamer, P.L.. Meredith. P.T.. Stoner, R.C.. “Areas for 7. Launder, B.E.. and Jones, WS., “Onthe Prediction of
Future CFD Development as Illustrated by Transport Laminarisation”, ARC CP 1036,1969.

* * * * *
 20-1

A STUDY OF THE USE OF HALF-MODELS IN HIGH-LIFT WIND-TUNNEL TESTING

by
P. B. Earnshaw, A. R. Green, B. C. Hardy and A. H. Jelly

Defence Research Agency

Farnborough Hampshire GU14 6TD, United Kingdom

SUMMARY The results suggested that there were indeed in some

circumstances clear discrepancies between the two.
An experimental investigation into the use of half-model The more comprehensive programme reported here was
testing techniques specifically aimed at high-lift testing therefore instituted in an attempt to clarify the issue.
has been carried out in the 5 Metre Pressurised Wind
Tunnel at the DRA, Farnborough. The aim of the 2 APPROACHES TO HALF-MODEL
programme was to provide an assessment of the extent TESTING
to which the measured characteristics of a high-lift
model might be compromised by, in particular, the Many different arrangements of half-model support
existence of a boundary layer on the reflection plane schemes have been adopted in different wind tunnels,
and how any penalties might be minimised. all offering particular advantages and disadvantages.
However, in those cases where a particular effort has
The results suggest that, provided care is taken with been made to minimise the influence of the boundary
experimental technique, good agreement is possible on layer growth on the reflection plane, they appear to
stall incidence as well as the absolute values of lift, drag divide into two approaches:
and pitching moment.
(a) to reduce the boundary-layer thickness on the
1 INTRODUCTION reflection plane,
The use of half-models in wind-tunnel testing can offer (b) to reduce the influence of the latter on the
several advantages. Among these are the engineering aerodynamics of the model.
advantages of design simplification, speed of manu-
facture and reduction of model cost together with the A very convenient and economical method of achieving
operational advantages in test of the removal of support (a) is to mount the model above a short reflection plane
interference and the increase of model Reynolds so that the distance over which a boundary layer can
number. However, these benefits are only offered at the grow is restricted. However, at the 5 m wind tunnel, it
expense of possible interaction between the flow over was felt that an essential consideration should be the
the model and the boundary layer over the reflection need to maintain as far as possible the same uniformity
plane. Because of this, many wind tunnel engineers of the flow and known calibration over the half-model as
take the view that, while useful results may be achieved that over a complete model. Only when this was the
for lift and pitching moment, this is not necessarily true case could there be reasonable hope that the behaviour
in the case of drag. An alternative view, also widely up to tho stall would then be common between the two.
held, is that the half-model technique offers a means of This view led to a rejection of a short supported
establishing incremental changes to forces and reflection plane or of an endplate where the outer wing
moments resulting from modifications to model build but would be able to 'see' the edges of the reflection plane.
cannot provide the absolute values of these forces. On the other hand, if a larger reflection plane were to be
used, the design should be such that the stream
Despite these reservations, the undoubted velocities on each side of the plane should be broadly
convenience of half-models provides a strong incentive similar; if this were not the case, the oncoming stream
-
to their use for example in attempting to define would respond as if to an additional contraction which
optimum slat and flap settings in the development of would require a significant streamwise distance before
high lift devices. However, in this particular application the flow were uniform again. To ensure that this was the
where the interest centres on maximum lift and the case, the plane would have to be be mounted reason-
nature of the stall, or on drag at high lift, it seems much ably far from the test section floor and the model size
more probable that a strong interaction between the would inevitably be reduced.
model and the wall boundary-layerflow would be evident
in the test results on the half-model. Initially, the technique favoured followed the approach
(b) above in attempting to reduce the interaction of the
At the 5 m wind tunnel at DRA, Farnborough, this type of floor boundary layer with the aerodynamics of the
test is carried out fairly frequently; it was important model. Thus, the model was supported with minimum
therefore to explore more carefully what limitations there clearance directly above an earthed plinth, or peniche,
might be on using these test techniques and to what having the same outline as the fuselage. However,
extent the speculations discussed above were results from the early stages of the test programme
accurate, in particular as far as the assessment of high- suggested that, whether or not a plinth had been
lift characteristics was concerned. installed, it would be desirable to reduce the effective
boundary-layer thickness.
A preliminary exploratory test programme comparing the
results from a half-model test and the corresponding Being a pressurised wind tunnel, space around the test
results from a complete model was carried out by BAe. section at the 5 rn tunnel is at a premium and it would be
very inconvenient to install large-diameter suction
pipework. However, high-pressure pipework presents
Copyright 0, Controller HMSO, London 7992 less of a problem and it has been possible to introduce a
20-2

boundary-layer re-energisation system using a nozzle made as small as was conveniently possible in
extending across the great proportion of the width of the order to reduce airflow between inside and
test section at a point around a metre upstream of the outside the test section,
nose of the model in use.
as close as possible to an earthed plinth, or
A calibration of blowing pressure and nozzle width was peniche, of 75 mm thickness and having the
carried out against tunnel pressure and Mach number same outline as that of the fuselage centre-
using boundary-layer rakes attached to the test section section. The plinth was attached to the turntable
floor at three positions across the width and slightly which carried the balance and model and
upstream of the model nose. The choice of what was consequently rotated with the latter as incidence
considered to be an acceptable total head distribution was changed (see Fig 2),
was of course largely arbitrary. However, plainly one of
the main questions to be answered was whether, and as close as possible to a deeper plinth of 100 mm
under what circumstances, there might be any thickness in order to assess the influence of
sensitivity of the model behaviour to the boundary-layer plinth height.
characteristics at the reflection plane.
Given the typical size of model used in the 5 m tunnel
The boundary layer profile shown in Fig 1 is typical of and the overhang of the extremes of a corresponding
what was taken to be the most suitable at one particular fuselage and making due allowance for movement of the
tunnel operating condition. Evidently, the total live balance platform before equilibrium was achieved, a
boundary layer thickness changes throughout the convenient clearance between the live part of the model
tunnel working envelope but, given that an appropriate and the tunnel floor in the absence of a plinth was
criterion for blowing pressure is applied, the character of around 5 mm. The semi-span of the half-model used in
the re-energised layer does not change substantially. this instance, obviously rather smaller than would
In fact comparative tests at much higher energisation normally be the case in a tunnel of this size, was
levels show no detectable changes in stall incidence 1.750 m.
and only small changes in forces at very much higher
levels of blowing. The half-model is shown mounted in the test section in
Fig 3a and since, at a later stage in the programme, it
3 DESCRIPTION OF MODELS became clear that an effectively reduced floor
clearance was necessary, details of the improved
In making assessments of the comparative effects of sealing are indicated in Fig 3b.
different test techniques on high-lift stall
characteristics, it has to be recognised that great care It must be recognised that, because of its fairly small
must be taken to avoid even minor changes in model size relative to the tunnel, the use of this half-model
build. These inevitably create a risk of introducing adjacent to an unrepresentatively thick boundary layer
significant flow changes in the neighbourhood of stall may well have exaggerated differences with the
which in turn serve to confuse the comparisons; complete model or at the very least made such
differences between models, support systems and differences easier to identify.
possibly even tunnels offer still greater risk.
When used with a plinth, the latter was earthed and
In order to minimise this complication, an existing 1:13 attached firmly to the turntable. Again, apart from the
scale model of the A300B Airbus aircraft which had height needed to accommodate the plinth, an additional
already been extensively tested at the 5 m tunnel was clearance similar to that when used in the absence of a
used. This model has been used as a general plinth had to be introduced between it and the live
workhorse for technique development since 1978 and in fuselage.
particular has provided the basis for a lengthy
programme of research into support interference. It Since the maximum thickness of the floor boundary
happened also that BAe Hatfield had been interested in layer at the lower end of the tunnel operational envelope
using this same model in a calibration exercise in their was about 100 mm reducing to around 70 mm at the high
own tunnels and had manufactured a suitable half end, the plinth height that was thought to be the most
fuselage so that it could be tested in half-model form in a appropriate was 75 mm. This height therefore was used
smaller tunnel. They agreed to its use by DRA for its most extensively in the comparative programme.
own research programme. Nevertheless, it was always evident that, at incidence,
the floor boundary layer would be swept up over the
Measurements for a very wide array of different fuselage regardless of plinth height. Consequently, in
configurations of the complete model were already an effort to assess the influence of plinth height, some
available, in most cases using a conventional three- tests were carried out using a second plinth, 100 mm in
strut support system from an underfloor balance height .
although, as part of the support interference
programme, it had also been tested using a sting 4 DISCUSSION OF RESULTS
support with an internal strain-gauge balance. Apart
from check tests therefore, there was no need to carry The manner in which different configurations react to
out any further testing on the complete model. changes in half-model test procedure is complicated
and may well explain why, following success .in a
In half-model build, the model was mounted from the particular instance, some laboratories adopt a pro-
same underfloor balance in three positions : cedure which has proved unsuccessful for others.
Furthermore, when in particular the interest centres on
(a) as close as possible to the earthed test section stall development. especially where some effort has
floor. The model support therefore passed been spent on optimisation of the high-lift character-
through a hole in the latter. No attempt was made istics of the wing and consequently all parts of the wing
to provide a labyrinth seal but clearance was are working hard, it can be more than usually difficutt to
 20-3

reproduce the forces and moments identically following At both Reynolds numbers, the 100 mm plinth retained
a model rebuild; even in consecutive test runs, an inboard stall.
movements of sealing materials or tape edges can
manifest themselves through significant changes to the The impact of BLRE is shown at the lower Reynolds
stall development but still be quite difficult to locate number of 5 x lo6 on lift in Fig 8 and on the corre-
physically. Nevertheless, in the present test sponding pitching moment in Fig 9. In this case, the
programme, it was felt that that the controlling stall on both the 75 mm plinth and the simple half-model
mechanisms could be identified fairly clearly despite was delayed but now the stall on the latter was higher
occasional lapses when the lack of reproducibility was than both the 75 mm plinth and the complete model.
evident.
Consider first the stall behaviour of the three half-model
In the course of these present tests, a large number of configurations. The introduction of a plinth, earthed or
different configurations for both landing and take-off not, produced a fuselage which was effectively wider
have been examined at various Reynolds numbers with and which therefore increased the effective incidence of
and without the presence of a plinth, and with and the inboard wing. Provided therefore that the outboard
without the use of boundary-layer re-energisation wing did not stall first, the highest plinth case should
(BLRE). However, for the purposes of this paper, the have stalled inboard first of the three. This, therefore,
data has been limited to that necessary to illustrate the accounts for the fact that, at both Reynolds numbers
various phenomena that have been identified as and regardless of BLRE, the 100 mm plinth case stalled
affecting the results. Fortunately, it has been possible inboard and early.
to choose a single configuration to demonstrate all of
the major effects: the A30064 in take-off configuration Without BLRE, the simple half-model may suffer from an
(16O:8"/3"). early inboard stall triggered by the thick reflection-plane
boundary layer. With BLRE, it has already been pointed
Fig 4a&b compare the lift curves for the complete model, out that stall on the half-model was delayed by around a
the half-model without plinth, with 75 mm plinth and with half degree. Now corrections to incidence are usually
100 mm plinth; there was no application of BLRE in this calculated at the 5 m \unnel as a suitably weighted mean
instance. If, for the moment, we choose to regard the of an upwash which varies across the span. In the case
results for the complete model as being correct, various of the half-model, which was in fact quite small for this
distinct features are apparent: tunnel, the wall-induced upwash varied by no more than
0.05" across the span and can therefore be regarded as
(a) the simple half-model had a liftcurve which was effectively constant. However, the complete model had
too low and stalled early by a little less than 2", incidence corrections both for larger wall corrections
which varied significantly along the span and as a result
(b) the half-model mounted on a 75 mm plinth had a of the presence of strut guards.
lift-curve slope which was only marginally too
high and began to stall very slightly early, The wall-induced upwash corrections have been applied
to the data as a weighted mean incidence correction
(c) the half-model mounted on a 100 mm plinth had a which in the case of the complete A300 model
liftcurve slope still higher and stalled at around corresponded to the upwash angle calculated at about
the same incidence as the simple half-model. 0.4 span; the contribution of wall-induced upwash
relative to the corrected incidence is zero therefore at
This behaviour seems at first sight to be erratic whilst 40% span. On the other hand, the procedure used at
also providing us with a strong predisposition to use - in the 5 m wind tunnel for applying strut and guard
this tunnel at least - a 75 mm plinth. In fact, the various corrections is such that these are applied together
interacting mechanisms are quite clear and will be using semi-empirical techniques; it happens to be
discussed in the following sections. convenient to apply the corrections to lift, drag and
pitching moment rather than to incidence. While the
4.1 Stall procedure works well at incidences below stall, it does
not result in a significant correction to stall incidence.
If guidance is sought from the pitching moment The distribution of upwash due to the strut guards has
behaviour, shown in Fig 5, for the same configurations been extracted from this data and is plotted together
discussed above, it is immediately clear that all the half- with the wall-induced corrections in Fig 10. It is evident
models stalled dramatically nose-down whereas the that the outboard 20% of the wing on the complete
complete model stalled less severely and nose-up. The model was at an incidence some 0 . 5 O higher than the
implication is that there was a very abrupt loss of lift corrected incidence used in the plots. Plausibly
inboard on the half-model whereas on the complete therefore, given that both complete and half-models
model, either the outboard wing or possibly the inboard were disposed to stall outboard, the complete model
flap stalled first. In fact, examination of the rolling would be expected to stall some 0.5" lower than the half-
moment data made it clear that it was an outboard stall model.
in this instance.
The same simple argument ought also to apply to the
However similar the full-model and half-model lift curves 75 mm plinth case. Unfortunately, while it does do so at
might appear in the case of the 75 mm plinth, the stall the higher Reynolds number, shown in Fig 6, it plainly
development was plainly totally unrepresentative. does not do so with BLRE in Figs 8 and 9. In fact, the
slight improvement of stall incidence following the
At a slightly higher Reynolds number, when the introduction of BLRE to the plinth-mounted model
reflection plane boundary layer was somewhat thinner, a results from a movement of separation from inboard to
magnified view of lift near stall shows (Fig 6) that now the neighbourhood of the pylon rather than to the
the simple half-model stalled at close to the same outboard wing. While it is quite possible that the
incidence as the 75 mm plinth case. Interestingly, the introduction of a plinth could so modify the lift
change in pitching moment through stall on both, Fig 7, distribution and upwash field that the mid-wing near the
had now altered completely from nose-down to nose-up.
20-4

pylon was triggered into an early separation that did not agreement. Evidently, although ‘spillage’ of lift carried
occur at the higher Reynolds number, it may seem more at the ends of the fuselage through inefficient sealing
reasonable to recall the difficulties mentioned at the did not lead to a detectable change in overall lift, it did
beginning of this section of repeating stall behaviour on make a significant contribution to pitching moment.
a wing that has been carefully optimised. Since a substantial length of upswept fuselage
remained unsealed at the rear, it is likely that mmpletion
4.2 Lift-Curve Slope -
- and possibly improvement also of the seals would I
lead to better agreement. However, given the quite
A deficit in lift-curve slope can be seen on the simple respectable agreement that has already been
half-model in Fig 4a. Initially, this had been thought to established, it seems fair to speculate whether an error
be due to the thick reflection-plane boundary layer of this magnitude could result also from inadequacies of
operating on the fuselage and inboard wing particularly the strut and guard corrections given that a pitch strut
in view of the much closer agreement when using a was used to support the rear fuselage for the complete-
75 mm plinth but calculations made using a panel model mounting.
method (SPARV) showed that this latter improvement of
lift-curve slope above that of the simple half-model 4.4 Drag
would be expected from the change in geometry
resulting from the introduction of a plinth. Furthermore It is convenient to illustrate differences in drag between
the operation of BLRE, as shown already in Fig 8, diff erent configurations by reference to the reduced
provided little improvement in slope. drag, CDR, where: CDRa CO- CfhA and where A is
In fact, significant improvement in agreement between the aspect ratio of the full-model in all cases.
half- and complete-model lift-curve slopes (Fig 11)
Fig 13 shows that pre-stall, the simple half-model with
resulted from the installation of a baffle plate which
BLRE had a reduced drag some 50 drag counts (0.0050)
reduced the gap between the model support and the
reflection plane from around 10 mm to nearer 2 mm, too high whereas, on a 75 mm plinth, CDRwas as much
together with additional sealing around the outline of the as 150 counts too low. This latter change was largely
fuselage in an effort to reduce the flow of air from one generated by changes to induced drag since the
side of the fuselage to the other (see Fig 3b); this was inviscid panel method calculations referred to earlier in
achieved without otherwise changing the model fact tend to overestimate this change; the calculated
geometry. The implication therefore is that errors in the drag changes due to the introduction of an earthed
measured lift resulted more from deficiencies in the plinth are included in this same Figure at three
design of the support system rather than from an incidences.
inability to reproduce the correct aerodynamics of the
The use of improved sealing shown in Fig 14 slightly
wing. Evidently, there still remained a significant lift
improves the similarity of the simple half-model to the
deficit which would probably require a fundamental
full model but although the comparison is now quite
redesign,of the model support and sealing system to
reduce to an acceptable level on a model of this size. close, the differences are still clear. However, it is
However, as suggested above, there seems to be no worth pointing out that the reduced drag includes an
input from the measured lift which is itself in error. If this
indication that the aerodynamics of the wing and high-lift
system are compromised by lift losses at the fuselage discrepancy is removed by referring the reduced drag
and support system. on the half-model to the lift at the same incidence on the
complete model, Fig 14 shows that now agreement has
4.3 Pltchlng Moment been brought to within 15 drag counts. In effect this is
equivalent to establishing that the total drag agreed to
The influence of BLRE and plinth installation on pitching- within 15 drag counts. At this level of agreement, it has
moment behaviour in the neighbourhood of stall has to be pointed out that an error in tunnel upwash on the
already been discussed. In addition to this particular complete model, or of sidewash on the half-model, or
aspect however, it is evident that all three half-models indeed a combination of the two, of 0.04O would give rise
had close to the same pitching moment at zero lift to this level of discrepancy at a CL of 2.
(Fig 5), that the slope of the pitching-moment curve
measured on the simple half-model agreed quite well 5 CONCLUSIONS
with that for the complete model but that the
aerodynamic centre moved perceptibly forward as the A substantial programme of comparative tests of a
plinth height increased. complete strut-mounted A300 model and a half-model of
the same aircraft using the port wing of the complete
model has been carried out in the 5 m Low-Speed I
The effect on pitching moment of the modifications to
fuselage/reflection plane sealing discussed earlier in Pressurised Wind Tunnel at DRA Farnborough. Since
relation to the lift-curve slope are shown in Fig 12. In the complete model was sized for the tunnel, the half-
fact, the modifications were carried out in two stages, model was evidently somewhat small and consequently
the results for both of which are shown.
The first stage applied sealing simply along the parallel
section of the fuselage while the second extended
these to the edge of the turntable; since the fuselage
overhung the turntable at both ends, an extension
around the complete fuselage was not easy to contrive.
(1) Unless efforts are made to reduce the effective
It is clear in Fig 12 that, while very good agreement was boundary-layer thickness on the reflection plane, the
achieved between full- and half-models by using short use of a simple half-model mounted close to the
seals, ‘improvements’ to the seals by extending them reflection plane may possibly - although not necessarily
towards the front and rear led to a deterioration in - yield wholly unsatisfactory predictions of the stall
I

i
behaviour.
 (2) Whem the results are indeed unsatisfactory. the
technique cannot be used even for the assessment of
incremental chacges of model configuration at least as
far as stall and post-stall behaviour is concerned.
(3) In some instances, the use of an earthed plinth
(or penlche) to lift the hd-model high in the boundary
layer on the reflsctbn plane may improve reproduction
of stall incidence. However, thb technique introduces
consequential errors in liftcurve slope, drag and
pitching moment. Furthermore. although the inboard
wing may have reduced sensitivity to the boundary layer
on the reflection plane. it is now dint to a fuselage
which is effedively wider; it therefore suffers from an
increased upwash around this wider fuselage and
consequently a tendency to early stall due to this
alternative mechanism.
(4) When the effective thickness of the boundary
layer on the reflection plane has been reduced
sufficiently. good reproduction of stall incidence may be
achieved. Differences between the two are likely to
I favour the half-modelgiven thd the full model is open to
disturbance from its support system.
(5) Some care must be be taken to ensure that flow
from one side of the model to the other, or through the
gap around the support frame, should be restricted. The
immediate eflect of inadequate sealing is shown as a
significant loss of lift-curve slope. This particular
problem has proved to be the most intractable with the
present halfmodel design where the discrepancy
between full and half-models amounted to around 1.5%
on rift. However, there seemed to be no indication that
this leakage interfered with the aerodynamics of the
wing or ofits hgh-lin system but was mom restricted to
the fbw about the fuselage. Wnh a model sized more
appropriately for the tunnel and designed from the start
as a half-model. it migM be reasonable to expect that
thb discmpancy could be reduced.
(6) Just as in the case of Mt, pitching moment also
responds to changes in sealing around the fuselage but
again giving no indkation that the aerodynamb of the
wing and its high-lifl system are compromised in any
way. In the present instananca, the total absence of
sealing at the extremes of the fuselage appeared to give
significant changes to the pitching moment coefficient
with a maximum e m d amund 0.01.
(7) Perhaps most surprisingly, drag appears to be
the most satisfactory in that disagreement between
complete and half-models over a large part of the
incidence range was not wone than 15 drag munts. An
improvement in this level of agreement may not be
possible wiihout greater confidence in the knowledge of
the mean values of upwash and sidewash vebcaiss
existing in the empty test rection; a total discrepancy of
O.Mo in incidence between complete and half-modeis
would be sutfident to generate this difference in drag at
a lilt &dent of 2.
206
e e
1.'

I .I

0.5

0.1

.-zm0.:
U
L

2a 0.f
..
-3
% 0.E
.-
-.,
n
E
6 0.4
z
0.3

0.2

0.1

0
- 1

Fig 1 Comparison of velocity profiles Fig 2 Half-model mounting geometry

with and without blowing

Fig 3a Half-model mounted without Fig 3b Detail of improved sealing

plinth in test section around fuselage
 20-7

Alpha Alpha

-
Fg 48 Effect of plinth height on C,
M 0.20, Re -5.0 x 1@ -
Fig 4b Effect of plinth he@! on C,
M 020, Re- 5.0X lb

Alpha
-3

Fig 5
- -
Effect of plinth height on C,
M 0.20, RE 5.0 x 10'
Fig 6
-
Effect of plinth height on C,
M 0.20. Re = 6.45 x ib
20-8

Alpha
'
3

Fig 7
- -
Effect of plinth height on C,,,
M 0.20. Re 6.45 x 108
Fig 8
- -
Effect of plinth height on C,
M 0.20, Re 5.0 x 1 6

Fig 9 Effect of plinth height on C,,, Fig 10 Comparison of upwash due to test section walls
M - 0.20. Re- 5.0 x 106 with that due to strut guards
 20-9

2.20
a NUMODEL
+ % MODEL, BLOWlffi ON,No SEALS-
a-15 0 % MODEL, 0LOWlNGON,W H SEALS
I

Alpha
‘3

-
I Effect of internal and external seals on C,
M 0.20,Re = 5.0 x 106 -
Fig 12 Effectof internal and external seals on C,,,
M = 0.20,Re 5.0 x loE

ON. WlTHouT SEMS.

G ON, WlTH SEALS.

E
0

Alpha
AlDha

Fig 13 Effectof plinth height on CD, Fig 14 Comparison of full and halfmodel drag data
M = 0.20,Re = 5.0 x loE
 e e 21-1

IN-FUGAF PRESSURE DISTRIBUTIONS AND SKIN-FRICTION MEASUREMENTS ON SUBSONIC

TRANSPORT HIGH-LIFT WING SECTION

L m g P. Yip
NASA M e V Relauch Centu
Hampton. Vir&& USA 23681-0001
Pad M.H.W.Vijgen
High Technology Corporation
Hampton. Virginia, USA 23681-0001
Jay D. HBdin
b k h d Engineuing md S c i a e u Company
Hampton. Virginia, USA 23681-0001
C. P. van D m
Universitv of Cdifornir Davis

SUMMARY
-C m u n u m d y n m k chord 11.20 h
Plight cxpuimenrs u e bcing conducted M p p t of a multi- d PMtGn-tute diuneta. in
phased subsonic transport high-lift research program for h prusure altinuk R
m e l a t i o n with wind-humel and computational results. K reluninrrization p.ramem. eq. (2)
The NASA Langley Tranapnt System Research Vehicle M Mach number
(B737-100 aircraft) is used to obtain in-flight flow P l o 4 static puunue. p f
chulctuistica at full-scale Reynolds numbers to wntribute P= freestrum static pressur.% psf
to the understanding of 3-D high-lift multi-ehent flows Pt Local total pesnnc. p f
including attachment-line transition and reluninarizUion. bceaaun dynamic pcdsw p-VzL pf
confluent boundary-layer development and flow separation B.
characteristics. &
- Reymlds n u m k bwd on V&
R attachment-ljnc Reynolds number, eq. (1)
Flight test results of praswe distributions and skin- Re momentum-thickness Reynold8 numhex
friction measurements werc obtained for a htll-chord wing
wetion including the slat, main-wing. and triple-slotted, S refercam wing m a , 980 f?
Fowler flap e h e n t s . Tcst wnditions included a range of S d a c e . smamwisc cmrdinrte.ft
flap deflections. chord Reynolds numbers (IO to 21 V blu dnpeed. ft/we
million). and Mach numbers (0.16 to 0.40). Resrurc vi i n d i d .irspeed knots
distributions were obtained nt 144 chordwise locations of a XlC non-dmnsional chordwise wordinnte
wing section (53-percent wing span) using thin pressure Y spmwise coordiruta. R
bclts over the slat. main-wing, and flap elements. Flow d C non-dimensional thickness coordinate
characteristics observed in the chordwise pressure a aircraft angle of UuCk deg
distributions included leading-edge regions of high- Qzn wing-section angle of attack, deg
subonic flows. leding-edge attachment-line locations. AP differential Reston-lubc presnne, p, - p. pf
slat and mun-wing cove-flow separation and re- h p deflcctim, &g
attdunent, and trailing-edge flap separation. In addition Sf
to the pressure distributions. limited skin-friction NE IeddingSage s w q angle. deg
me.surements werc mndc wing Preston-Nbc probes. rl non-dimensional spmwise IocUion, y/(bfZ)
Reston-tube measurements on the ala1 upper surface V kinematic viscosity. fi2/sec
suggested reluninuiwion of the nnbulent flow inttuduccd
by the pressure belt on the slat lendingulge surface when Po freestream density, slugs/d
the slat attachment line was laminar. e momentum thickness, a
r surface shear stress, p f
Computational analysis of the in-flight pressure
measurements wing two-dimensional. viscous multi-
element methods modified with simple-sweep theory ESP electronic scanning pasure
showed re.lsoMble ngIecment. However. overprediction of KlAs l m o t s i n d i d ~
the pessures on the flap elements suggcsts a mal for bettea LE leddingdge
&tailed musuremenb and impovcd modeling of confluent TSRV Transport System Rcsureh Vehicle
boundary layers M well M inclusion of three-dimensional 2-D two-dimensional
viscour effects in the UUlysiS. 3-D three-dimensional
LIST O F SYMBOLS 1. INTRODUCTION
b wingspm.ft High-lift system umdynamics can significantly impact
Cf skin-friction coeffiient, Vq- the overall design of transport airnln in terms of sizing.
Rest~n-tubc-mesnned Cf usmg eq. (3) performance. system complexity. and certification for
C; safety and community-noise acceptance .spects.l
CL aired lift coefficient,WqA However. the design of subsonic high-lift s y s t e m remains
cn nomul-fora coefiicient from integrated p r u m a technical challenge mostly due to the limited
pesslm coefficient,@P=Uq- understanding of the wmplex flow physics .ssociated wifh
cP
C chord length. R high-lift flows. Multiclement, high-lift flows are very
21-2

sensitive to Reynolds- and Mach-number effects; and measurements of pressure disuibutions and boundary-layer
therefore, the aerodynamic performance of high-lift flow paramerns at flight Reynolds and Mach numbers is
systems is generally difficult to extrapolate from wind- critical to the evaluation of computltional methods and to
m e 1 M to predict in wmputational fluid dynamics (CFD) the modeling of turbulence for closure of the goveming
studies. In order to improve the design methodology for flow equations5*6 Detailed meosurnmnts and laplyma of
high-lift systems. additional experimental data with the multi-element flow field have generally been limited to
sufficient flow details on transport-type swept wings are two-dimensional studies. Previous detailed flow
needed at fight Reynolds and Mach numbers to better measuremenu in wind-tunnel investigations have included
understand high-lift flows. Thus far. only a very limited Reynolds-stress components. however, at suh-scale
number of flight investigations have been conducted with Reynolds numbers and only in two dimension^.^.^*^^^^
sufficiently detailed flow measurements on a high-lift These results have been applied towards the development
system for wrrelation with ground-based investigations and validation of 2-D multi-element numerical codes.
and to guide 3-D CFD modeling. One such investigation
was reported by G d f 2 on an Airbus A310-300 aircraft. Two-dimensional multi-element flow issues include the
following:
As part of a multi-phased research program at NASA
Langley, flight tests are being conducted on the Transport
f.1)
lay& inWacti& on the slat;
-
. comuressibiliw effects includina shocllboundw-
Systems Research Vehicle (TSRV), a B737-100 aircraft, to (2) laminar separation-induced transition along the upper
obtain detailed full-scale flow measurements on a multi- surfaces;
element high-lift system at various flight wnditions. In (3) wnfluent turbulent boundary layer(s) - the merging
Phase I of the program. experiments were focused on flow and inkrachg of waked from upstream elements with
visualization, pressure distributions. and flow-separation the boundaq layas of downstream elements;
measurements on the triple-slotted Fowler flap system of (4) wve separation and reattachment; and,
the research aircraft. Phase I activities have been (5) massive flow separation on the wing/flap upper
completed, and the flight test results were reported by surfaces.
Vijgen et 113 In Phase It, more detailed flow measurements
have been planned including the main wing and leading- Availability of detailed flow data in three dimensions at
edge slat components. Initial Phase-I1 flight experiments full-scale (flight) Reynolds numbers has been much more
have recently been completed on a wing section at one limited. To wntribute to the undcrstandig and eonelation
spanwise station to measure pressure distributions on the of high-lift research between wind tunnel CFD. and flight,
full chord, and these results were reported by Yip et al4. additional experiments are needed at full-scale Reynolds
Additional Phase II flight experiments are planned to number and for 3-D swept wings. F& understrmding of
provide additional full-chord pressure measurements at scale effects in both 2-D and 3-D is required to accurnlely
several spanwise s t a t i m and detailed measurements of extrapolate to 3-D, full-scale. flight ~ o n d i t i o n s . ~ ~ . ~ ~ * ~ ~
leading-edge transition and relaminarization. In Phase III.
extensive off-surface insbumentation for detailed turbulent Threedimenaional multiclement flow issues include the
boundary-layer and wake flow-field characteristics are following:
planned. This paper summarizes the test results to date of (1) leading-edge attachment-line transition;
the high-lift flight research program on the TSRV as well (2) reluniruoization of turbulent flow in the leadingcdge
as planned further flight experiments. region;
(3) crossflow instability transition downstream of the
attachment line;
Possible shocW (4) sweep effects on confluent boundary-layer
B.L. interaction development, turbulent boundary-layer separation.
Confluent boundary layers and separated wve flows; and,
( 5 ) highly three-dimensional. local flow modifications;
e.g.. vortex generators; flap side-edge-separated
flows and tip effects; and flow i n r n ~ t i ~with
n ~ slat
brackets and flapaack fairings. engine pylons, and
landing-gear smts.
Relaminarization
The three-dimensional multi-element issues of leadingcdge
‘-Cove separation J attachment-line transition and the potential for
relaminarization are discussed in more detail in the
following section.
Figure 1. Muki-element flow characteristics.
2.2 Attachment-Llne Transltlon and
2. BACKGROUND FOR HIGH-LIFT FLOWS Relamlnarlzatlon Issues
In three-dimensional, swept-wing flows, the flow along the
2.1 Multi-Element Flow Issues attnchment line (a locus of points near the leading edge
dividing upper. and lower-surface flows) can be lminar.
The flow field around a multi-element transport wing with transitional. or turbulent, depending on the pressure
sweep is characterized by several aerodynamic phenomena distribution. the leading-edge sweep angle. and the
which are highly interrelated, complex in nature, and Reynolds If attachment-line transition
generally not fully understood (see Fig. 1). Accurate occurs, the resulting changes in the development of
prediction of surface-pressure distributions. merging boundary layer flows can significantly influence the
boundary-layers. and separated-flow regions over multi- downstream turbulent flow field (i.e.. confluent boundary
element high-lift airfoils is an essential requirement in the layers and onset of separation). Relambrimtion of the
design of advanced high-lift system for efficient subsonic flow downstream of a turbulent attachment line can occur if
transport aircr8ft.l The availability of detailed the streamwise flow acceleration is sufficiently
 21-3

strong.17.18.19 H the flow of a steep ~dvcrsc chuactaristic length whae U‘ represents the inviscikJ
presnne grndient along the upper d k x of the elemam is velocity grndient lt the a h t n m t lins in the direction
l b . nn additionnl Reynolds-numba effect can occur norm4 to the aaacimmt line.14.15.16 POI a -I
due to the pr- of a 1.minU-scpantionbubble md its l t w e n t - l i n e boundary Iaya d- nn infits swapt
effect on subsequent turbulent-flow bch.vior. cylinder, themomennun thickness c m b a &own to be 0 =
0.404 I( , nnd consequently. Re = 0.404 E2I Reviolu
The issues of ludingcdge transition md mlrminuization.
illwunted in figure 2 for n singleelement, swept wing, sadiesh.vslhoamthufmR<245,theltuclnnentlias
baeome very important in the extrapolation of sub-scnlc, boundary h y a w i l l rcnd to r d luninn,ad turbulent
three-dimensionnl. wind-tunnel results to flight. conumirution introduced in the bounday layer by
T y p i d y , wind-tunnel dntn used to utrnpolate maximum sipifionnt surface r u u ~ e s smd interswting turbulent
lift in fight are obtained nt Reynolds-numbers conditions E
shear layon will decay. Far > 245. the turhlenca wiU
where wing sull is dominated by wnventionnl sule effects d-Wbh, e the ~ h t l M 3 l t - hflow Y W d Y the
(see wo~dwnrdet d.20). Conventionnl d e effects refa flow downstrcun of the utachment line to become
to the incrcnse of muimum liR with Reynolds numbs due turbulent. For incompressible conditions, in the ahance
to the thirming of the turbulent boundary-lnyer in the wing of roughness and intermcting turbulent shea layers.
trailing Cage region ad the s u w aft movanQLfof the uuchmmt-line instability followed by transition will
hailingcdge flow separation point. At higher flight ocnnontyifTiD5so?2
Reynolds numbas. nttachment-linc transition can occur
causing turbulent flow to stnrt from the lluehment line. An important pametar when examining nhinaizuion
By shifting forwnrd the starting point of the turbulent of turbulent bomduy-laya flows is the invone Reynolds
boundary layer, the trailing-edge separation locntion cm number, K, which is defined Y follows:
.Is0 shift forward due to the incnrsed growth of the
turbulent boundary layer. Becwe of the incMscd extent
of trdingcdge sepnrntion. a signitcant reduction in lift
cnn occur. However, beewe of steep fnvorable pressure
gradients nssociated with high-lift flows, relpminaizntion
is possible for some sections of the wing and would w h the chunctcristicvelocity is r e p s e n d by the local
k e b y deviate m e of the lift loss expected ns n result
of the attachment-line trmsition. In a high-Reynolds- inviscid velocity. U. md the characteristic length. 6. M
number wind-tunnel investigation of a swept-wing &tined by 6 = UN: with U; rcpesmting the velocity
configuration without slnts, maximum-lift losses of the
order of 15 F o n t have been measured when transition . ~K~ . ~ ~
gradient dong the inviscid s t r e u n l i n ~ . ~ ~For
OcCMcd along the .tuchment line md rehnimizntion did k g a thm I x 10-6, mvasion m f mm turbulsnt to ~pninu
not ocau. 12 flow is possible, for K > 3x1W6 relrmiapiutiOn is likely,
nd for K D 5xW6 complete IClmhmidon OeMl for 2-
D flows (only very limited det.ilsd dntn1g*19 mu wulnble
to date for 3-D flows on swept wings).

Attnchment-line trnnsition. rcluninnrizntion. nnd

along the elements of the B737-100
R O S S ~ ~ O Winstability
high-LiR geometry were examinsd by vijgen et 11.3 wing
predicted pressure distributions. Results of the study
a i - thuthe ~tt.clrmmtlioedong the s l u l n d f m nq
IW likely to be laminar, while the amchnmt line dong
the fued -1 edge (main element) is likely to be -
turbulent for the 4oo-flap setting U.a = 0.. In this paper. R
nnd K a e obtained from the mensured pressurc
JJl +Wpiml W k d Tmml RN + I-Ffbhl RN-
distributions. The attachment h e dong the slnt in the
fight e * ~ rrponad by 0,Strzon the ~ i r b m u ~a
c
Reynolds number (RN) 300 nimaft was muMsd to be lnminalmda cutninhigh-
lift flight conditions.
Figure 2. Effect of attachment-line transition and
relaminarizationon maximum lift. 3. DESCRIPTION OF FLIGHT EXPERIMENT

An important parameter when examining the trmsition 3.1 Test Alrcrdt

charsctniaticsof the utachment-line bolmd.ry layer dong T k NASA Lmglcy TSRV was the p t o t y p aircraft used in
-
the lending edge of a swept lifting surface is the the development of the Bocing B737-100 md hls becn
attachment-line Reynolds number. R. as dcfmed by the significantlymodified tor fight systcmr reseawh.3 ~n its
following quntion: unmodified state, the Boeing 737-100 in a twin& shim-
haul subsonic t r a n s p ~designed
~ to cury app10ximat.dy
100 passengas with auk speed of Mach 0.78. In nda
R=-w..w
- to obtain ~ h ~ ~ t - f i Wrcoff
eld md -1 pcrfotinm~e,the
V k n f t incorporated a slnt nnd tiple-slotted tlnp high-lift
w k W,=VsinA system. Basic &edymmc ’ chacctmirtics of the Booing
737-100 configurntion from ~uind-”lel blvestigltions
and X = (v/u’)0.5
have been documented in reports by CaponeZ6 and
The charncteristic velocity, W,. is the spmwise
~nu1son27.
component of the freestreun velocity, nnd x is the
21-4

-
main-wing

15Vlap setting

-\
25O-flap setting

Fwre3. NASA TSRV (737-100 airash) in fbht with

U' Mo-tiap setting
\
4oo-llapsetting and landing par up.

Figure 3 shows the TSRV in fligbt with the tfipl&dottal

flap system deployed at a 4Oe-flnp setting (landing
configuration). Figure 4 shows a planform view of the
B737-100 d i g u r a t i o n m d illustrates the overall
geometric characteristics. The aircraft wing U Fgure 5. Wing sections at various flap settings.
cburterizedby aspnuof 93 R m.spct&of 8.82, md
a sweep mgle of 250 U the quarterordlboe md 27.580 U
tbe lcding edge outboard of the underslung, low-bypur
ruio engine nacelle.
r r f W , ~ p P r l W )0 . s
WW. uppw (3UI 0.w
wino, l a n r (iL) a18

1, @ m a o n IUD.. (uppm ,ianr)

I Pmnunml

Fylure 6. Instrumentation layout.

3.2 Instrumentatlon, Data Aequbltlon, and

Date Reductlon
Instrumentation for the full-chord, wing-section
measurements is illustr.tcd in figure 6. Surf= static
Figure 4. Planform view of the 8737-100 aimraft. pressure distributions. Rerton-tube skin-friction
medsumnents. md tlow-vi.u.liutioe mulu were obtlinsd
Inbopd luding-edge Kntcger ilap d outboard leading- on one outboard wing section of the rescar& .ira.fL The
eAge slurs pc extended in wnjlmtion with the deflection chordwise pressure distributions were measwed at a
of the triple-slottedmiling-edge flap system. As shown in nominal spmwise station of q = 0.53 on the upper lad
fi- 5, the O U h d high-lift Wing SECtbIl 6tUdiied hl this lower Mfrees of the slat, &-wing. and flap elements
papa consists of 5 elemenw the ludinguige slat, the using thin belts of plutic tubing (0.062 in. ouuide
muin wing with fixed leading edge, the fore flap. the mid dimam IO.028 in. inside dimneter) which were wrapped
flap,md the aR 0.p. At flap settings of 300 md 400,the nounduchelement. T h c b e l U W e r e U r r h e d t o t h e ~ ~
two most outboard slat I C ~ I I I ~ Uare
~ I fully extended md with thin (0.00s in.) adhesive tape. To minim& U t edge
deflected m .dditiorul increment from the 15' md 25' effects. five extra (mn-functioning) tubes were ddcd to
scaing (see Fig. 5). effectively creating a spanwise break cachside. md the sided of u c h t c l t were smoothed with a
in the wing-planfomk leading edge between the inboard silicon-rubba cnnpound. Thc technique of using extend
 21-5

pesnne belts h been wnunonly used in previous studies &-time display and stored on an optical disk for post-
and was shown to provide good results when cornpad to flight playback and data analysis. Ressurea were recorded
flush surface orifces.28 he possible e h t of ~ t onsthe at a rate of 10 samples per sewnd while aircraft fight
pressure distributions in the luding-edge regions of the pumnem wmMu 1 of 2o sampies par ssmnd.
psent high-lift system will be determined in funuc flighrs Theplnrmuic lag for the longest tube length was mcMmd
with flush orifices (we extion 6). With the instaUation of in g r d tests and determined to be approximately 0.5
the pressure belts, the minimum flap deflection was limited d and was ukcninto Uxamt in the reduction o f b u
. while the standard maximum flap deflection of 40'
to 1
5
'
was unchanged. In& to provide cometions of the suticpressm due to
probe position mor for ea& flap setting, &speed
calibration f l i h t WM conducted prim to the research
. .. QW using a uacking-radu method29. hue comctioM
a l o n g w i t h ~ ~ m u n n c s n e nWere
t l used to compute
fxeuaUn st& a n d d y n a m i c p s s ~U well M M d md
Reynolds numbers. A h . angle-of-lark conections were
obtained by calibrating the aircraft angle-of-attack vane
against pitch attitude obtained from the aircraft inertia
navigation system in steady-state flight conditions. Lift
coefficients were determined from atuay-sute. 1-g fight
RIMQIversusing aircraft weight c.lculated horn aircraft fuel
consumption measured by fuel-flow c-znsors. Thrust
concctions obtained from standud engine perfomunce
decks were .pplialto the lift data to muut forthe thrust
contribution to lift.

Flow visu;lliution was obtained by applying nylon yam

lufm to rhe upper slnfsus of the outbondflap and the main-
wing elements (lee Fig. 7) to indicate local Icgions of flow

(a) ar -
40'
Figure 7. Pressure be# and Preston-tuba installation.
quation. The tuft pattcms were Iccorded with still and
video photography to d o w for post-fight analysis and
melation. In addition, audio rewrding of pilots and
flight test enginscn on the video cassettes were used to
dirccm quality of flight data sampled.

Preston tubes were installed on the upper and lower surfaced

of the main-wing element, and one Reston tube was
installed on the slat upper surface (see Figs. 6 and 7).
These probes contain a sutic orifice in addition to the pitot
p o m e Mi to measure total and sutic pnrsuns within
the bounduy The Preston-tube probes were
installed just outboard of the pressm belt and staggered
spmwise to minLnizc interference effects of the probes on
u c h other. Reston-tubemelsurements near the treilhg
edge of the flap elements were obt.ined during the Phue I
fights st the same spmwise location and were rcpoacd in

I detail in reference 3. The local *in friction coefficient.

Cp was determined based on the measured pressure

(b) dose-up View,

Figure 7. Concluded.
-
15"
differential at the tube using the Law-of-the-Wall
d M o n for equilibrium tlnbulmt &W. pn the modired
~Klnblbeuscdherr.the W d rhan S-S WMdCtermined
08 fOhJWS:30

A total of 160 pressure tubes (144 for pressure

distributions. 12 for static and total measurements of the
R u t o n tubes. and 4 spare tubes) were connected to five (3)
elecuoniully scmming ponne @SP) modules which were
louted in the wing cove region (see Fig. 6). The ESP
modules were maintained at a constant temperature to
minimize zero shift of the measurement, and two where Ap represents the differentid pressure m d by
differential-pressure t r d u c c r ranges (2.5 and 5.0 p i ) the Rsrton tube. d reprwnts the outside dimneter of the
allowed high resolution of the preasure data. A plenum Preston tube, and A1 and A2 are calibration constants
chamber was housed m the wing m e region to provide the where A1 = 38.85, and A2 = 1073. Th outside diametaB
reference pressure for the ESP transducers; the reference of the Reston tube were d = 0.083 in. for the win#
p u ~ was e monitored with an abmlute p o m e gauge and butions and d = 0.042 in. for the slU M o n . The above
wurelntcd to the static purm meamred by the aircraft cnlikation exposion is based on the -ption that the
pitot-static pobe. A s m d data-acquisition unit using static and total poru of tho probe m louted within the
single-board computer technology was located in the re& of the turhlau boundary layer that i# governed by
outboard flq track faking of the wing (K. Fig. 5) to y ) o l s the L.w-of-thbw.1L since the bounday-laya lute 8s well
and address the ESP . -
U The digital output data U its thickness are depsnaau on the flap &g and the
were trmrmilled to a small. portable on-boardcomputer for vmiour flight pnr.mcm. it was Imcsrt.inthat d Preston
21-6

tubes would operate within the appropriate calibration 4. FLIGHT-TEST RESULTS

range at dl M conditions. B e c a w of this uncertainty. Cy
wmpuutions fmm the Preston-Nbe measurements using In the d y s b of flight data, it should be noted that test
eq. (1) uc listed in this paper u1 c; to indicate that these pinta were obtained at 1-g. level flight wnditims when
increasing mgle of a t k k wnuponb to dower .irrpodq
measurements do M t necessarily reflect acNd skin-friction and thcrcfore. both the Rcynolds numba and the Mach
values if the Reston-Nbe readings were obtained outside nutnbcr chmags with u c h twt p i n t (KC Pig. B).
the valid cllibration range. e.g.. if the flow WLS h i n u .
4.1 Trfmmed-Llft Data
3.3 Fllght-Test Condltlons Trimmed-lift cocffiicnts from rcpresPlutivt flight &U at
The fight experiments wvcred a range of Reynolds and h = 1O.ooO ft are shown in figure 9 for the TSRV
Mach n u m h as the aircraft was flown to prednne altitudes wnfiguration with flap defections of 15'. 25'. 3 0
'. and
up to 20.000 feet. The chord Reynolds number, R-,ranged No. For the angleof-attack range tested. the lift curves
fium 10 to 21 million, and the freestream Mach numbcr remained nearly linear with no significant slope changes
varied from 0.16 to 0.40. Flap deflcctims of 15". 25O. 3P. except for a slight decrease in the lift-curvs slope for the
and 40' were investigated at pressure altitudes of 5.000 R. 15'-tlap caw at high angles of attack.
10,000 ft, 15.000 ft. and 20.000 ft. Test conditions and
test points obtained in flight are show in f p e 8 well
U lines of constant Reynolds and Mach numbers for
standard atmospheric conditions. As indicated by fuure 8.
flight at increasing altitudes provides conditions of
innusing Mach n u m k and decreasing Reynolds number
for a given indicated airspeed. Vi. which corresponds
approximately to a wnstant lift coefficient for a given
aircraft weight.
2.0 -

0.0
d 0 4 8 12 16
a, d.g

-
Fgure 9. Trimmed liidata from flight. h 10,000R.
0 50 tm E4 100 m
VI, KIAS
--a-- Flight. a-naps
Fgure 8. Flight test conditions.

The flight-test points were obtained for each of the flap

settings at appximately I-g, steady-state conditions in
level flight with the aircraft initially flown at a high
nominal airspeed and then slowed to the stick-shaker 2.0
sped. Daw wuae sampled fox lppmximately 30 seconds at
each constant-airspeed test point. In addition, data were
rewrded during the deceleration of the aircraft between
selccted test points. The aircraft was decelerated at a
nominal rate of 1 knot per 5 seconds while constant
dti& wns maintained. Penincnt wt points were repeated
to mure data npeatabili. AU data were obtained with the
landing gear reIracted. The research flight deck4 on the
l.ot
TSRV allowed auto-bale and auto-pilotopnations ofthe
aircraft for nirspead- and altitude-hold modes of testing.
Use of the auto-pilot oprations increased the pmductivity
of the flight test and enhanced the quality of the fight data
OS
0.0
tU
sampled. -4 0 4 8 12 16
a, deg
Fgure 10. Comparison offlighland wind-tunnel liidata.
 0 21-7

In figure.10, the flight lift data are compared with available 4.2 Pressare Dlstrlbutlon~ and Skin-Frlctlon
wind-Nmel lift data2' for the 30'- and 4oo-fllp SettingS. Measarementr
The wind-numel investigation used a 1/8-scale model of the Resnua distributions measured for dl elamcnU on the
TSRV with flow-through nacelles. The wind-tunnel data TSRV. high-lift wing Scctioa are pruonled in figure 11 for
were obtained U test mnditions of M = 0.14 and R- = 1.4 the W-. 25"-, 30"-. and 4oo-flap settings from flight data
million, significantly less than the flight Reynolds for a range of angles of attack at a test altitude of h =
1O.OOO ft. In general. pressure distributions at other
numbers which ranged from about R- = 11 million at the dti& &owed simila ruults. Several p o d am noted
high-angle-of-attack (low-sped) conditions to about 18 here b u d on the overall flight-measured pressure
million at low-angle-of-attack (high-speed) conditions. distributions: (1) Changing angle of attack affected
Wind-tunnel fmcs and moment data were recomputed for e a r i l y slu and main-wing pressure loading. and hd
18-percent E, the nomind center-of-gravity location in little Snect on the flap pressures. The amdl vnimce of
flight, and trimmed-lift coefficients were estimated based flap loadii with angle of .nrL U explained by the fixed
on stabilizer requirements for trimmed flight. The &flection gomnetry of the U p s ~ c u nelsmant which largely
wmparison shows siflicant differences h e t w a flight determines the in-flow angle for the flap clement. (2)
and wind-numd data due to viscous (Reynolds-number) Because of sweep effects. muimum Cp V ~ U MW R C less
effects. Flightdatapoduud asteeper and mom lincar lift- th.n the 2-Dvalue for the stagnation point. A h . flow
curve slope thau the wind-tunnel data. At high angles of rcmdunent in the main-wing cove is indicated by Cp
attack, the wind-tunnel lift-curve slope decreases with values close to the l u d i y c d g e atrhmsnt-be p w e .
increasing angles of attack, indicating stronger viscous (3) Rsrsrnw U the trailing-edge of d element, except
effects. The undqmdiction in the wind-tunnel data is for the aft flap. do not recova completely due to the
likely due to premuure flow sepaation U the much lower influam of the d o u ~ ~ t r melement.
n The wceluuk
wind-tunnel Reynolds-numbu conditions. In the mid- flow region between the elemam llso wllw~the W i l i n g -
angle-of-attack range. thae was agrrnncnt in the lift and edge pressure distributions to reflect the highu flow
incremental lift values between fight and wind-tmmel data. velocities. For example, note the decreasing main-wing
However. at low angles of attack, the flight data were element upper-surf- pressures llcn the bailing edge with
d the wind-tunnel dau This ovuplediction
O V ~ e d i C t e by
hcredsii flap deflection. The higher velocities near the
may he I result of an inverse Reynolds-number effect on trailing edge llso relieve the pressure. rise on the -1
multi-element airfoils where lift decreases with increasing edge of the Qwnstrum element, thus alleviahg potential
Reynolds numher (discussed by Morgan et and separation problem that could cause loss in lift. These
woodward et ~1.24. multi-element flow phenomena are attributed to the
"circulation effect" and "dumping effect" U duaibed by
smith32.

-1 4
-13 I o 2.72" 20.2 x 10' 0.33 1
6.68" 16.4~10'
-12
0 9.39" 14.7 x108
-1 1
A 13.46" 12.7~10' 0.21
-10
b 16.42" 11.7 x10' 0.20
-9
-8
cp -7
-6
-5 b
k4
-4
-3
-2
-1
0
1

-
(a) 8 1 5 O
Figure 11. Wing-section pressure distributions.
21-8

I a RE M
0 -1.51" 17.9~10' 0.28
0 2.47" 14.2~10' 0.23
0 7.70° 11.8~10' 0.20
-9 r
-8 c A 9.57" 11.2~10' 0.1 9

cP -4 b 13.38" 10.2 x10' 0.1 7

-5
-4
-3
-2
-1
0
1
0 .1 .2 0 .1 .2 .3 .4 .5 .6 .7 .8 .9 0 .l .2 0 .1 .2 .3 0 .1 .2
( X W slat (x'c) main-wing (x~c)fore.liap (X~C)mid-llap (x'c)aft-fl#p

-
(b) h 40'
Figure 11. Concluded.

- Slat

3'01
2.5
-++-
Main element
Fore llap
Aft nap

t
-0.5
2 4 6 8 10 12 14 16 -2 0 2 4 6 8 10 12 14
a, dog a,dag

-
(a) 6f 15O
Figure 12. Wingsedion normalforce coefficients.
-
(b) k 40"
Fgure 12. Concluded.
Pressure distributions for each of the elements were Flow charncmistics of the individual hi&h-lift, wing-
integrated along the wing chord in the flap-reaacted m i s e section elements are examined in more detail in thc
position (8f = 0') to provide section normal-force following sections.
coefficients UI shown in figure 12. The results clearly
indicate that the wing and slat loadings increased with 43.1 Slat-rlcmcnr flow charactcriSrtic8
angle of attack while the flap l o a d i i remained relatively The slat pressure distributions are prcsentcd in more detail
constant. in figure 13 for the 15"- md @-flap s ew md at w a d
angles of attack. A comparison of the slit prssrurs
distributions for the 15'- and 40'-flip Betting shows
 21-9

similar res1 r a- - of attu (e.g.. note the

given angle For flap raings. the appoxim position of the
pressure diseibutions at a = 2S0. 9.Y. and 13.59. Both uuchment line, as indicated by the maximum Cp value. is
flap settings show favorable upper-surface pressure locucd on the upper surf- of the slu for angled of attack
gradients at low angles of attack, nearly flat pressure less than about 4 O . Consequently. for the 40O-flap setting
disbibutions at mid agles of attack. and an upper-surface and a = -1.51'. the slat is shown to produce neguivc lift
suction peak near the leading edge with a subsequent (sec Fig. 1Zb). At this neguive angle-of-attack condition,
& e m gradient domstream U high angles of aaack. Pur the b w a surf= flow lppurs ro be separated as evidenced
the W-flap setting, the upper-surfwe suction peak r e d d by the nearly constant Cp value in the range of 0.05 c a/c c
a minimum value of Cp = -13.18 st the highest available 0.15. On the lower surface,slat pressure porn aft of x/c =
angle of attack. a = 16.4Z0. This Cp value eonespwds to a 0.30 were r d y beusd in the slat mve (sec Fq.3). The
local M.ch number of 0.83. Even though the frecstream nearly mnstmu pressure level in the slat wve region for
Mach number was only 0.20, these high local velocities both flap settings at high an+ of attack is indicuive of
produce Mach effects which can adversely affect the separated flow region in the cove. Near the lowa-
performance of high-lift systems. surface trailing edgc. the pressure distributions indicate
increasing flow velocities as a result of the slot flow
U a=2.72' betwecn the slat and the main-wing elements. In this

-
a=6.68' region, the numw gap for the 15'-flap letting caused L
-0- a=9.39' larga Icccler.tion compared to the 40'-flap setting.
* a=13.46'
a=16.4P In figure 14. the R ~ ~ t o n - mcasuronmts
~be on the slu are
-11 plotted ss a function of aircr.ft angle of atuck for two
Solid symbol denotes altitudes. h = 10.000 R and 20.000 ft. and two flap
-10 kwsr surfaca
-9 settings, 6f = 15" and 40'. Thc data show a s i g n i f w t
increase in the measured values of C; for the mid-angle-of-
CP attack range ( 2 9 c a c 9.0') for both tlap settings except
in the lower altiNde, higher Reynolds number data of the
150-flap setting. his increase in C{ can be t r ~ tod a
sudden increase in the keston-tube total-pressure ~ ~ . a d h g
as opposed to the static-pressure r e d i which varied
smoothly throughout this angle-of-uuck r q e . The flow
behavior may be related to the attachment-line boundary-
layer state and the phenomenon of relaminarization as
discussed next.

-
0 0.2 0.4

X'CSliIt
0.6 0.8 1

I - h = 10,OW R.
h=20,WOft. I
-
(a) €f 15" 0.014-
Figure 13. Slat pressure distributions.

- a=-1.5ID
0.012 -

'R
a=2.47'
a=7.70' 0.010 -

-6
+
A a=13.38*
a=9.5P

Solid symbol denotes

c,
.
0.008-

-5
0.006
4
CP
-3 o.oM~l I I I I I , , I ,

-2
0.002
-1 -4 0 4 8 12 16
a, deg
0
-
(a) & 15O
Figure 14. Slat Preston-tube measurements.
1
0 0.2 0.4 0.6 0.8 1
X'Cs1at
-
(b) st 40"
Figure 13. Concluded.
21-10
e
sudda iaausc in Cf' w u measured (lee Fii. 14). Thuh
the increase in cf' in ~e intermediate mglwf-.tt.cl
0.014 Rnga appears to indicate a IrmhLlr bound.ry-laya state at

i
tbc~esmtube. ~ h e m o p i n ~ata-90indicueath.1
c
the boundq-l.ya nuc b.s become turlmlent again. At
0.012 this angle of attack, the change in bnmdary-layer state is
explained by the development of a suction peak on the slat
upper nufacc mu its leading edge (see pressures in Fig.
0.010 14).

Cf 0.008

0.004

0.002
-4 0 4
a. deg
8 12 16
-R 300i X

(b)&-w
Fgure 14. Concluded.

Using the measured pressure dis&tinns (Fig. 13). fi is

calculnted 8d plotted in figure 15. The ruultr show thu 0- 0
the slat utaclrmcnt line is expected to be Lrminp for most -2 0 2 4 6 8 10
test conditions at a 400-Bap setting bued on the crituion
a, deg
of c 245 for laminar attachment lines. HOWNU. it
should be noted that the Reston tube WM locued just
outbod of the pssure belts (ree Fii. 5). md thesc belts
will cause the attachment line to become turbulent.
(a) & 15O -
Fgure 15. Attachment-line Reynolds number and
rehminarization characteristics for the slat
Although the turbulence introduced into the attachment- upper surface.
line boundary layer by the belts will decay for z c 245. a
cornpaison with Gaster's e x p e h e n h l results14 d e p i i
the decay of turbulence behind lrip w i m on the uuchmmt
line .ppcns to indicate that the slat Reston tube WM
locued in the turbulent-flow region outboard of the belt
since the Reston-tube dbratim MsUmes that the probe
is located withii a mrbulent boundary layer. this
introduction of turbulence h the belt wake explains the
approximately nmnd readings by the pmbe at low-d 300-
high angles of attack in spite of the low vdues of R (c -
E
245). But attachment-be mdysis using C ~ C U ~ U ~ O M R
dw not explain the in- in cf'.f intermediate angled -
'\
'I, n
I
.I/ 1 6'
of attack. To help explain this result, the possibility of 200 K,X
relaminnrization is mdyzed for the slat uppcr-surface
region. Rel.miwiution produces a s i g d k n t reduction
in bounbry-llyer thickness. md this uusu the R e s m
probe to be only patidly s u h d in the b o a layer -
100
M compared tobeing m y slhmsedinthc turbulmtuse.
The e n d d t is m iacrusc in the total-prrrm reading of
Ihe Rsrm tube, md thuefm m aaonlpanyiag increase -4
in tbc vdue of cf'.

In G p 15, the muimum vduc for K dong the upper-

nufw inviscid srrumline, U c u fmn the menured
pressure distributions ( ~ s n m i u g l o e d y infiite swept
wing unlditilms). h dm plotted. Ths results. exmined
firstfortheW-tl.p~indiutethUicl ' . lion,
b.Md on K = 3 x would occur for m g h of attack
gmws thrn .pproximucly 250; a value thu h in IrUIkally
.good l&m.m5twithcbc o l l w ~ o utrk f at which thc
 21-11

to predict the occummx of relcluninuiutia in highly 4 3 3 Main-wing element frow characterislics

d e n t e d flows. However for the 15'-flap setting. the The nuin-wing p n u ~ e disuibuticns arc presented in more
results in figures 14 and 15 show th.1 h a x . a puanetcz deuil in figure 16 for the 15O- and 40'd.p settings at
bved on a l d kpkh number, is Mt adqIate to fully several angles of ma&. In both CMW, them is a luge
capaxe the flow physics of the rellminuiwirmpoccss fm suction peak at the x/c = 0.12 locltion of the uppa nnf.ca

firstoccu~~ed -
3-Dflows. Fm the 15'-flap setting, the incream m C:
at .ppoxirmtely a 7'ud h = 2O.ooO A md
didnot ODM at dl U h = 10,oM)A. The results Of f w e 15
A Local flow loccldon WM registeed by the paslm belt
meuurement U thir location which canerpond. to a
pxsure port loutad just bshitd a notch (i.e., a forward-
facing step) in the lding&e muface. The notch in the
.how that in ~ c u e Kmax values must be larger thm airfoil conllour in a mlt of the h t clement maacting into
approximately 6 x for rcluninuization to ocm. the main-wiry leading edge for d r n fight conditio^.
Also. in Ibis use.the demise of luninu flow at the Reston Luge flow acceleruionr dong the leading-edge uppm
tubs cm be correlated with the developInenl of the *dine- nntrc make thia region conducive to mlslunLuriution u
edge ~ c t i o npeak. Mors deuiled munncmmtc using s ~ d i e din the paper by Vijgen a 11.3 HOWWW. M
t l u s h - m d stuic pesm nificg md hot-film sensors insmemation w u available during thew initial PhUs-II
arc phnned to iiuther study the amuition phcMInem on experiments to measure the bounday-layer state lhud of
the sl.t (sec section 6). the pressure minimum poin~. For the 4oo-W case, the
suction peak oceunsd at lbout x/c = 0.036 which in .bud
of the notch in the surface and m a t e s a double-peaked

-
pressure distribution for most of the angles of attack
shown.
+a=1S.4e0 On the lower surfies. the 6 most-& pressure porn are
a=16.42O louted on the spoiler lower surface in ths wing cove
5 Solid symbol denotes region. Note th.1 the Iowa-surface flow lppusntly re-
uuchu onto rhs spoiler lower surf- M indicated by the
- 4 . high t+ value wbich is similar to the uuchmant-lime value
neu the leading edge. The influence of the slat lower-
surfax. scpanted-flow region on the main-wing element is
evidenced by the rsduccion inpeslurrr in the loWer-aurfrn
nose region ( 4 - b S 0.20). For the 15*-flap setting, the
slat and main-wing elements form. in effect, a single
elenrent at l o w a angles of attack The slot between the slat
md the main-wing element is very namw (sec Pig. S),
unuing the slu luding edge to function U the sugrution
0 zegion faboth the slat md the main-wing elements in bw-
md mid-range &ea of ltuct
1
b t o n - t u b s .kin-friction mo.1UIQnc110 for the mlin-aing
0 0.2 0.4 0.6 0.8 1 elme111 m Y t e d in 17 for the 15'- md 400-flap
"Clllain-wing usesuahctionof&eof& Threcuppu-lurfua
md two lower-surf- RM~OII t u b were attached to the
(a) 8 I1 5 O &-wing element (res P i . 6 md 7). Ths data indicated

-
Figure 18. Main-wing prassure distributions. high V l i u e a of c: for p b c m . 1 u in tk U~-flapcase md
forpobes 1u. 2u. and3u in thew-napcuy. T h c r c q l
readhp am indicative of thc high flow veloeitiu and din
-7 r

-
a=2.4P boundary layan at thorn locltions. Robe no. 1U in located
+a-7.70' just bchind the Mtch in thc h d h & g S UpPC? nnfrs (We
+a-S.5P Fig. 7b). Also. in both cues. the lower-surf- p b s near
a=13.38* c h krdiy
~ edge. p b e no. IL, experienced IOW C{ v d w
at low angles of attack. This result is indicative of the
SolidSymbol denotes
k w rs u b lower-surfw slu repuntion bubble at low of uack
engulfing the nuin-wing luding-edge region on the lower
surf-. At 40°-fhp setting, the slat is deflected an
additional mount (sec Rgs. 5 and 7). Por the 40'-tbp
case,the C: values for the midchnd location, &robe no.
2U. md fm the aft-chod location, &robe M. 3U. exhibited
higher values than those for tha ISo-- cue. This result
correlates with the widening of the slat gap, wbich allow8
higher velocity flows over this region.

4.23 Flq-elUnmr flow choracterutics

A deuiled study of the flap prusurs distributions w u
prmted m the paper by Vijgen et al.3; therefore, only a
limited discussion will be prkm~tedherein. The foreflap,
0 0.2 0.4 0.6 0.8 1 mid-flap, md aft-fhp pxsute distributions are shown in
detail in figura 18. 19. md 20, respectively. for the 15'-
XICMaln-wlng and 400-flap Scnings and m m d angled of attack.
@)4-W
Fuun 16. Concluded.
21-12

Preston tube localion

0.012
+a=Q.3Q0
+a= 13.46'
0.010

0.008
F
*I 0.006

O1 .o . U
0.000
2 4 6 8 10 12 14 16 0 0.2 0.4 0.6 0.8 1

-
(a) 6f 15"
a, d e g

Figure 17. Main-wing Preston-tube measurements.

-
(a) 6f 15'
X'CFore-flrp

Fgure 18. Fore-flap pressure distributions.

0.012
I
Preston t u b location
-
+
+a=Q.SP
a=-1.51D
a=2.4P
a=7.70"

a=13.38'
0.010 Solid symbol denotes
lower surface

0.008

Cf
0.006

0.004

0.002

1
0.000 0 0.2 0.4 0.6 0.8 1
-2 0 2 4 6 8 10 12 14
a,d e g
-
(b) 6f 40'
Figure 17. Concluded
(4sr -40"
Fgure 18. Concluded.
 21-13

-
-
::::I
-1 .o
-
+
a=2.72'
a-6.68'
a-B.3B0
a- 13.46"
a= 16.420
Solid tymbal denotes
bwer surface
-1 .o
+
a=2.72'
a=6.68"
a=8.38'
a=13.46"

0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1

X'CMld-flap X'CAR flap
(a) & - 15"
Figure 19. Mid-flappressure distributions.
-
(a) & 15'
Fgure 20. Aft-flap pressure distributions.

- a=-1.51' - a=2.47'

-
a-2.47' a=7.70"
-3.0 a-7.70' + ad.570
-2.5
+ a=8.57"
a=13.38*
Solid symbol denotes
-2.0

-1.5

cp -1 .o
-0.5
0.5
0.0

0.5

1 .o 0 0.2 0.4 0.6 0.8 1

0 0.2 0.4 0.6 0.8 1
X'CAft-flap

(b) Sr -
Fgure 19. Concluded.
X'CYld-flap
(b) Sr -40'
Fgure 20. Concluded.
21-14

For the 15*-flap . c t h g . mmeasmg the q l c of utwk For the 40'-flap letting, the upper-nnf.ce loading .Im
increlsed the upper-surface loading slightly while the i n d dightly with incrrui angle of uuck except U
l o w e r - d u x pesm remained nearly unchanged. These the highest angle of atuck shown (a = 1338O). Flow
increases in loading for the fore-flap dement are mall wp.r.tionnerr thehailiq edge of the fore flap is indieucd
compared to the increws for the slat and main-wing for dl anglea of ittack ahown except the lowest &e of
elemenu because the in-flow angle is predominantly tixed
uuck (a= -1.51O). The flow squation ia i d i d by the
by the gametry of the flap system. No separation is
indicated in the pressme distributions for the ISo-flap
setting even U the highest angle of Lttwk shown (a =
16.4'2'). although the decreasing lrding-edge p w ~ ~ ofe s
d y COnuuDt p r u ~ s
-
level near the ailing *e uppsr
surface of the fore flap starting at xlc 0.80. At the
angle of uuck (a= 13.38'). iplonolmcad e f f a of
the aft flap ue indicative of a thickening hounday layer at the sepmtion is .Is0 o h c d on the mid-flap elrmeu.
higher angles of w k .
Mururements of modified Reston-tubs loutcd at the
cr.iling-Cdge of ucb flap elemcnt m presented in figure 21
to further exunine the wmmencc of flow rapation (sec
0.005 r ref. 3 for Raton-tub locitiona on the flap). Raton-tub

o.oMl
mc.dmanents mpasantcd U afuaction of mglc of uuck
for the 15'- and 40"-flap lettings. Nesative v a l w of C{
indicate flow reppuion at the Preston-tub loution. For
the 15O-flap setting. no separation is indicated for the
angle of attack conditions m d . The h i g h rudinga
for the fore flap MKCSPOSWJ to the higher velocities
0.003 associated with the fon-flq now. For the 40'-flq setting.
the cf' values indicate s e p a r d flow on the fore f l . for
~

*I
Cf 0.002

0.001

0.000
t
angles of atrack patex than qpmximuely Oo.

-0.001
4 0 4 E 12 16
a, dag
-
(a) 8f 1 5 O
Figure 21. Flap trailing-edge Preston-tuba
-
measuremenls (from ref. 3). h 10,OW h.

0.005 r -
(a) 8 1 5 O , a - P
Figure 22. In-flight tuft fbw visualbation.
I

.
Cf 0.002

-0.001
-4 0 4 8 12 18

R0q-W (b) & -Ma.a - ,

Figure 21. Concluded. Fuure 22. Concluded.
 21-15

Typical flow-visualization results are shown in figure 22 Equation thmugh a distribution of constant-strength aource

the tuft photo (Fig. 228) is shown with the aimaft at a

9'. Even at this relatively high-angle+f-attack condition,
-
for the 15"- and 40°-flap settings. For the 15"-flap setting. and vortex singularities distributed over the surface of each
element. Compressibility effects are pecountcd for through
the Kuman-Tsien compressibility correction.
improved version of the Ooradia integral. confluent
An
the flow on the flap surfaces appeared streamwise with no
indication of separation along the span betwem the flap boundary-layer model is employed. allowing MCARF to
track fairings. For the 40°-flap setting. the tuft photo is

-
shown with the lircratt nurr the stick shaker speed (Vi
105 KIAS. a 7') For this condition, the flow p a m
- predict the presence of separation bubbles as well as the
onset of confluent boundary-layer separation. The basic
assumptions of the d e n t boundmy-layer model are that
the static pressure normal to the element surface remains
showed generally attached flow on the flap system, wnatant and that near the trailing edge of each element the
although, in the wake of the flap track fairings, localized confluent boundary layer has degenerated to an ordinary
unsteady and separated flow regions are evident. In the turbulent boundary layer. The potential- and viscous-flow
region of the pressure belt locations. the flow remained solutions are coupled through the boundary-layer
attached on the main wing and the flaps except for flow displacement thickness and solved iteratively. This
separation near the trailing edge of the fore-flap element, inviscidlviswus coupling precludes the malysis of flow
The tuft patterns of figure 22h indicate that flow separation regions with scp.ration.
occuTTed over appmximately the lest 20 percent of the fore-
flap chord. The tuft patterns correlate well with the
previously discussed observations based on the pressure The MSES multi-element airfoil code37.38 is a reccntly
distributions and Reston-tube results. Figure 22b also developed extension of the single-element ISES
indicates that the flow near the trailing edge of the aft flap wmpressible flow airfoild 9 s 4 0 . The streunline-bd
is on the verge of separation at this condition although the Euler equations and a twoquation integral boundary-layer
pressure distributions does not show incipient separation. formulation are solved simultaneously using a full Ne-
Finally. the Nft patterns near the flap/aileron edge show iteration method. The initial streamline grid is generated
three-dimemional tip effects in the trailing-edge region of through the solution of a panel method at the specified
each flap element due to the flap-edge vonical flow field. angle of attack and is modified aftcr each Newton iteration
as part of the solution Displacement bodies based on the
5. COMPUTATIONAL ANALYSIS OF PRESSURE boundary-layer and wake displacement thicknessa are used
MEASUREMENTS to displace the element surface geometries and are
A limited computational analysis of the experimental incorporated into the solution after each iteration. This
pressure distributions at the span station of q = 0.53 is strong invkid/viscous couplii procedure enables MSES
to predict the effects of laminar separation bubbles md
presented using two currently available 2-D wmputational
methods. The current lack of fully 3-D viscous multi- other regions of limited flow separation on the pressure
element alpi is methods makes the use of 2-D methods in distribution MSES is formulated to account for the effects
of both asymmetric wake shapes and confluent boundary
conjunction with simple-sweep theory necessary for the
layers. although a confluent boundary-layer model is not
analysis of 3-D high-lift system.33 The sectional included m the MSES version used hue38
geometries used in the computational analyses. shown in
figure 5. have been smoothly faired in the wve areas of the
main-wing and mid-flap elements to facilitate the flow 5.2 Comparlson wlth Fllght Data
calculations. In the present analysis. simple-sweep Results of the MCARF and MSES codes. modified as
theory34 is used to account for the (inviscid) sweep effects described previously. are wmpared to the flight-measured
on the pressure distributions. pressure distribution8 in figure 23. The wmparison shows
that the predicted pressure distributions are in reasonable
agreement with flight-measured pressures for flap settings
One of the major challenges in applying 2-D sectional of 15' and 40' at the angles of attack shown. For the 15"-
analysis methods to 3-D wing geometries is the flap setting, where no separation was predicted or
determination of the w e c t local angle of attack for input measured, the magnitudes of the flap element upper-surface
to the 2-D methods. For single-element wings at low pressures are slightly overpredicted. However, separated
angles of attack. where viscous effects are not dominant, flows in the lower slat-cove and along the main-wing
this may be accomplished in a relatively straightforward lower-surface leading edge n e not well predicted, and the
mmner by matching the predicted and experimental section suction peak near the forward-facing step on the main-wing
normal-force coefficients. In the case of multi-element upper surface is not predicted in the MCARF computations.
high-lift systems, however, the use of the sectional For the 40°-flap setting, where flow separation was
normal-force coefficient is not feasible due to the strong measured on the fore flap, the pressures over the fore flap
viscous interactions and the presence of confluent are substantially overpredicted. Although the flow
boundary layers and flow separation. In the results separation nen the trailing edge is not modeled in MCARP
presented here, the local angle of attack was determined by calculations. the location of separation onset is predicted
matching the predicted and experimental upper-surface slat and is in agreement with the experimental separation
pressure distributions.
locuion on the fore flap.3 ~nddition. pressures are
overpredicted for the leadingdge upper surfaced for dl the
5.1 Deserlption of Multi-Element Aualyslrr elements aft of the slat. This overprediction of the
Codes P s n U e S NggeStS thU d k g Of the CdUent boundary
The MCARF (Multi-Component Airfoil) computer layem in MCARF as well as the present nppximation of
cod~35.36is a widely-used aerodynamic analysis code for three-dimensional sweep effects is not adequate for the
2-Dmulti-element airfoils. It is a potential-flow panel complex high-lift gmmeuy studied.
method and employs integral boundary-layer methods to
predict the vismus flow over multiilanent airfoik. The
potential flow solution is obtained by solving Laplace's
 0
21-16

c, -4
-3
-2
-1
0
1

0 .1 .2 . 3 . 4 .5 .6 .7 .8 .9 0 .1 .2 0 . 1 .2 .3 0 .1 .2
x/c x/c x/c x/c
(a) a - - -
15". a 9.39". RE= 14.7 million. M 0.24
Figure 23. Comparison of predicted pressure distributions with flight measurements.

! !
c, -4
-3 1

-2
-1

r:
0

b7T fl
1

z / c-.:1 f = l I I 1 I

-.2
0 .1 - 2 0 . I - 2 . 3 . 4 .5 .6 .7 -8 .9 0 - 1 .2 0 . 1 .2 .3 0 .1 .2
x/c x/c x/c x/c x/c
(b) - -
40'. a 9.57". Ra = 11.2 million, M = 0.19
Figure 23. Concluded.
 21-17

The camputationd grid of the MSFS mde are presented in rhese flights. allowing insmenIaIion to bE embedded in
figure 24 for the 40O-flap setting. Figure 24. is the fd the d a m s . A number of multichannel ESP t r m u d m
solution grid used by MSES to solve the Euler-flow will be used to obtain h o s t instantaneous pressure
equllions and shows the complex streamline patterns as d i n g s at .ppoximucly 1200 pressure ports distributed
well as several dominant features of the solution. Most over several Ip.nwisc wing rutions. The F u r e d will be
notahle are the scpuated flow regions f m e d on the slat measured using flush-mounted Orificcr installed on the slat.
lower surface and near the trailing edge of the fore-flap. wing spoiler, and flap surfaces. Since the &-wing
Figure 24b is a m.gnifed view of the solution grid in the element is used as a "wet wing" for fuel sIorage. pressure
region of the flap system and reveals small regions of belts are still needed to provide main-wing pressure
sep.Rtionnurthemilingedgedofthemidand~-flapsas distributions. Hot-fh aensors will be installed along the
well as in the cove of the mid-flap. The flow over the fore- slat, &-wing. and fore-flap leading-edge surfeces to
flap appears to be separated over approximately the last detect transition, attachment-line. and separation
twenty percent of the surface at h i s condition. The chuacteriatics. Boundary-layer rakes will be installed to
agnrcmrnt of the mcaMed and predicted separation near the determine the effect of slat wake on the nuin-wing flow in
fore-flap trailing edge indicates that the direct thi
sphase of the research p g r a m .
inviscid/viscous coupling procedure used in MSES allows
reasonable modeling of the flow separation observed in
flight The nur-constant pressure nur the trailing edge is
predicted by the MSES mde (see Pi. Ub).
\\

Fgure 25. Phase llB instrumentation plans.

(a) 5-element hoh-lii system
Fgure 24. MSES computationalgrid solution.
a-9.570.R~-11.2million.M-0.19.
8 - 40'. Because of the large number of sensors. a comprehensive
data-acquisition system for flight is being developd to
handle in real-time the large volume of pressure and hot-
tilm anernometex data for analysis purposes. A multi-
channel anemometer data napisition system with digital
storage capability is planned in n d to ~ expedite the
analysis of the high-frequency boundary-layer
measurements. In addition to insmentation for flow
measurements. flow visualization is planned to provide
furher ~ ~ ~ h u n dofi nthe g flow physics on the high-lift
system. Also. anmturaldefonnation measmanem will be
required to complete the m n e l d o n of flight data with
ground-based W i n d - I l m n e l d CFD r d t s . puna in-flight
flow measurements from this research program should
provide cunently unavailable flow details at full-scale
flight Reynolds numbem for m e l a t i o n of ground-hed
wind-tunnel result8 and a chaUengiq validation t8st case
for CFD analysis. particularly in the modeling of 3-D
transitional and confluent boundary-layer d scpnuion
phenomena.

7. CONCLUDING REMARKS
Flight experiments an being conducted as part of a multi-
phased subsonic tr;msport hah-lift research program for
correlation with ground-based wind-tunnel and
(b) flap fbw characteristics computational results. The NASA Langley Trmsport
Fgure 24. Concluded. Systems Resureh Vehicle (B737-100 aircrafI) is used to
obtain in-flight flow charecteristics at full-scale Reynoldr
6. FLIGHT EXPERIMENT PLANS n u m k s to mntribuw to the understanding of 2-Dand 3-D
Future flight experiments on the TSRV high-lift high-lift flows including attadunent-line transition and
configuration are planned to provide measurements that relaminarization. confluent boundary-layer development.
addrus in detpil S W Rof ~ the critical multialement flow and flow separation charactaistics. This paper nrrmnniW
isms. Figure 25 illustrates the insmunenution for Phase- the test results to date of the high-lift fliiht resureh
IIB flight experiments planned in 1993. Slat, wing program on the TSRV as well as planned furthct flight
spoiler, and flap spare parts have bem obtained for use in experiments.
21-18

Flight tests were pertormed OVR 1range of chord Reynolds

numbers from approximately 10 to 20 million and Ilstkms
freestream Mach n u t n k s from appximately 0.16 to 0.40
for angles of attack of up to near-stall conditions. Flight IDillner. B.. May. F.W.. and McMuters. J.H..
test results showed that aircraft lift coefficients obtained "AaodynunicIssues in the Deaign of H i - L i f t Systems
from steady-state. trimmed conditions in level flight for Trm~sportAircrq" Proceedings of tk Symposium
differed from available wind-tunnel trimmed-lift on Improvement of Aerodynamic Perfarmanse Through
coefficioltr (30'- and 40'-flap settings only) due to v h u s Bowhry-Lqcr Control md High-Ljft System. AGARD
(Reynolds-nutnk) effects. The flight-test data exhibited CP 365, May 1984.
m n linear lift curves md steeper l i f t a w e slopes than t
b %reE. E.. "In-mght ~-ement of static p l e s ~ l r rmd
~
wind-tunnel results for the 30°- and 40°-flap settings. Boundary Layer State with Integrated Sensors,"Journal
Flight test results of pressure distributions, Preston-tube ofAircraj?, Vo1.28. May 1991. pp. 289-299.
skin-friction meknmmcnfs. and surfaceflow visualization 3vijgen. P.M.H.w.. ~ u d J.D.. i md Yip. LP., "plow
over a fullchord wing seetion on a uiple-slotted h p high- Prediction over a Transport Multi-Element High-Lift
lift system were presented for the 15'- and 40'- flap System and Compuison with Flight Measurements,"
settings. Measurements of the pressure distributions Fifth SynyMsium on Numerical and Physical Aspects of
showed that increasing the angle of attack primarily AerodyMnje Flaws. T. Cebeci (ed.),California State
increased slat and main-wing pressure loading. but had only University. L a g Beach. CA, January 1992.
a small effect on the flap-element loading. For the 40O-flap 4 ~ i p L.P..
. vijgen, P.M.H.w.. ~udin.J.D.. and van D U ~ .
deflection. the upper surface pressures indicated flow C.P.. "Subsonic High-Lift Flight Ruearch on the NASA
separation near the &ailing edge at all but the lowest angles Transport Systems Research Vehicle (TSRV)." AUA
of attack. Tuft-flow visualization and Preston-tube results Paper 924103CP. August 1992.
corroborated this separation on the fon-flap trailing edge. 5Bme. G. W.. and McMlstus. J. H.. "Computational
Pressure distributions for the slat and main-wing elements Aerodynamics Applied to High-Lift Systems," Progress
showed lower-surface-separation regions. re-attachment in AaoMvries and Astrmauics: Applied ConpvcntiMor
locations, and aft movements of the attachment-line Aerodynamics. P.A. Henne (03.). Vol. 125, AIAA, Ncw
location with angle of attack. Pressure distributions on the
slat upper surface reached high suction values which
-
York. 1990. pp. 389 433.
corresponded to locaUy high, but subsonic Mach numbers. 6Bengelink. R.L., "The Integration of CFD and
Experiment: An Industry Viewpoint (Invited Paper)."
AIAA Papa 88-2043. May 1988.
Preston-tube measurements on the slat outbard of the thin
pressure belt indicated a likely laminar boundary-layer state 701son. L.E.. .nd oriott, K. L.. "on the suu~nvcof
at the Preston tube for certain conditions of angle of attack Turbulent Wakes and Merging Shear Layers of Multi-
and altitude. The apparent laminar-flow readings of the Element Airfoils." AUA Paper 81-1238. June 1981.
Preston-tube p m h could be explained by relnminuization 8Brune. 0. W.. and Sikavi, D. A.. "Experimental
of the belt-tripped flow under the influence of large Investigation of the Confluent Boundary Layer of
favorable streamwise pressure gradients near the leading Multi-Element Low-Speed Airfoil." A I M Paper 83-
edge. Further flight experiments using hot-film 0566. Januuy 1982.
instrumentation are planned to study the transition and 9Braden. J.A.. whipkey. R.R.. Jones, G.S.. Lilley, D.E..
relnminarization process in more detail. '"Experimental Study of the Separating Confluent
Boundary-hyer." NASA CR-3655. June 1983.
A limited analysis of the experimental pressure loNaLayam~ A., Kreplin, H.P., and Morgan. H.L..
distributions using two 2-D. viscous, multi-element "Experimental Investigation of Flowfield About a
computational codes with integral boundary-layer methods Multielement Airfoil." AIM Journal. Vol. 28. No. 1.
showed that the predicted pressure distributions were in pp. 14-21. Janumy 1990.
reasonable agreement with flight-measured lxeswes for llKirkpauick. D.. and Woodward, D.. "Priorities for High-
m h e d flow conditions. For the 40°-flap setting, where Lift Testing in the 1990's." AIAA Paper 90-1413. June
trailing-edge flow separation was measured over the fore 1990.
flap, the pressures over the flap elements were 12Guner. P.L. Meredith, P.T.. and Stoner, R.C.. "Arus
overpredicted. This overprediction of the flap pressures for Future CFD Development U Illustrated by Trmrport
suggests that current 2-D modeling of the contluent Aircraft Applications." AUA Paper 91-1527-CP. June
boundary-layer and three-dimensional (sweep) effects are 1991.
not adequate for the present, complex high-lift geomeuy
studied. More detailed in-flight boundary-layer 13Valarezo. W.O.. Dominik, C.J., McGhee. R.H.,
measurements and the application of advanced (three- Goodman. W.L.. Pa&& KB.. "Multi-Element Airfoil
dimensional) viscous analysis methods for complex Optimization for Maximum High-Lift Reynolds
gmmeeies will be required to further address the effects of N m k . " A I M Paper 913332, Septemba 1991.
confluent boundary-layer development on three- 14Pfenninger. w.. " L . m i Flow C o m l hinariution.
dimensional high-lift systems. USAF and NAVY Sponsored Noahrop LFC Research
Betwecn 1949 and 1967." Special Course on Concepts
ACKNOWLEDGEMENTS for Drag Reduction. AGARD Report No. 654. March
The work of the second author was supported under NASA 1977, pp. 3-1 to 3-75.
h g l e y Contract NAS1-19299. The work of the thud 15Gasm, M.. "On the Flow Along Swep Lcding Edges,"
author was supposed under NASA Langley Contract NASl- The Aeronautical Quarterly. Vol. 18. May 1967. pp. 165
19ooo. The work of the fourth author was supported in part - 184.
under NASA Langley Cooperative Agreeanent NCC1-163. l6Poll. D. I. A.. "Transition in the Infinite-Swept
Attachment-Line Boundary-Layer." The Aeronautical
Quarterly. Vol. 30, Pan 4. N o v e m k 1979, pp. 607 -
629.
 21-19

17~-&. B. E.. md JW. w. P., *o~ the ~sdictionof 36Bnme, G.W.. M d e . J.W., "An Impoved Vasion of the
Relaminarization," ARC CP 1036. 1969. NASA-Lockhced Multielement Airfoil Analysis
8Hardy. B. C.. "Experimental Investigation of Computa Rogram"NASA CR-145323. 1978.
Attachment-Line Transition in Low-Speed High-Lift 37Drela. M.. "Newton Solution of Coupled
Wind-Tunnel Testing," Proceedings of the Symposium Viswus/Inviscid Multi-Element Airfoil Flows," A I M
on Fluid D y m ' c s of Three-Dimensional Turbulenr Papa 90-1470. lune 1990.
Shear Flows and TrwiIion, AGARD CP 438. 1988. pp. 38Drela, M.. "A Usa's Guide to MSES V1.2." MIT
2-1 to 2-17. Computational Fluid Dynunks Laboratory. July 1991.
Amal. D.. and Juillen. J. C.. "Leading-Edge 39Giles, M.B. and Drela, M, 'Two-Dimeslsinul T r d c
Contamination md Relminuiwion on a Swept Wine U Aaodynunic Design Method" A I M Journal, Vol. 25.
Incidence." Fourth Symposium on Numerical and No. 9. September 1987. pp.1199-1206.
Physicd Aspects of Aerdynamic Flows, T. Cebeci (ed.), 40Drela, M. md Giles. M.B. "Vismus-Inviscid Analysu of
Cal. State Univasity. Long Bedch. CA, January 1989. TrmsOnie and Low Reynolds Numba Airfoils." A I M
2oWdward, D.S.. Hardy, B.C., and Ashill. P.R.. "Some Journal. Vol. 25, No. 10. October 1987. pp. 1347-1355.
Types of Scale Effect in Low-Sped High-Lift Flows,"
ICAS Paper 4.9.3, 1988.
21Crabtree. LP..Kochemarm. D.. and Sowerby. L.. 'k-
Dimensional Boundary Layers," Laminar Boundary
Layers. L. Roaenhead (ed.),Oxford Ress. 1963. pp. 409-
491.
.
22Ha11, P.. Malik. M R., md Poll, D.I.A.. "On the
Stability of an M,te Swept Attachment Line Boundary
Layer," Prcc. R. Soc.Lond.. A395. pp. 229-245. 1984.
Z3Narasimha, R. and Sredvasan. K.R., "Relaminarization
of Fluid Flows." Advances in Applied Mechanics. Vol.
19. 1979. pp. 221-309.
24Beasley. J.A.. "Calculation of the Laminar Boundary
Laya and Rediction of Transition on a Sheared Wing,"
ARC R&M 3787. 1976.
25White. J.J.. "Advanced Transport Operating Systems
Program," SAE Paper 901969. October 1990.
26Capone. F J.. "Longitudinal Aerodynamic
Characteristics of a Twin-Turbofan Subsonic Transport
with Nacelles Mounted Unda the Wings," NASA TN D-
5971. October 1970.
27Paulson. J.W.. "Wind-Tunnel Results of the
Aerodynamic Characteristics of a 1/8-Scale Model of a
Twin-Engine Short-Haul Transport," NASA TM X-
74011. April 1977.
28Montoya, L.C.. and Lux. D.P., "Comparison of Wing
Pressure Distribution from Flight Tests of Flush and
Extemal W i e s for Mach Numbers fmm 0.50 to 0.97,"
NASA TM X-56032 April 1975.
29Gracey. W.,"Measurement of Aircraft Speed and
Altitude," NASA RP 1046. May 1980.
3 % k n e ~ d . A. 'Total HeadlStatic Measurement of Skin
Friction and Surface Pressure," AIM JOWMI. Vol. 15.
No. 3, March 1977, pp. 436438.
3 1 M o r g ~H.L.. Ferris. J.C.. and McGhee. RJ.. "A Study
of High-Lift Airfoils at High Reynolds Numbers in the
Langley Low-Turbulence Pressure Tunnel" NASA TM
89125. July 1987.
32Smith, A.M.O.. "High-Lift Aerodynamics," Journal of
Aircrafr. Vol. 12, No. 6. June 1975. pp. 501-530.
33Bme. G. W.. and McMasters, J. H.. "Computational
Aerodynamics Applied to High-Lift Systems." Progress
in Aeronautics and A s r r o ~ ~ I i c sApplied
: Computational
Aerodynamics. P.A. Heme (ed.). Vol. 125, AI& New
York. 1990. pp. 389 - 433.
34Lock, R.C., "Equivalence Law Relating Three- and Two-
Dimensional Pressure Distribution," ARC R&M 3346.
May 1962.
35Stevens. W.A.. Goradia. S. H.. and Braden. J. A.,
"Mathematical Model for Two-Dimensional Multi-
component Airfoils in Viswus Flows." NASA CR-1843.
1971.
 e e 22-1

HIGH LIFT AND THE FORWARD SWEPT WING

by
Dr. Lawrence A. Walchli
Wright Lahoratory
Wright-Patterson AFB OH 45433-6553
united States
ABSTRACT followed the envelope expansion and
was completed in December 1987. Drag
A broad overview of the X-29 Forward reduction during maneuvering exceeded
Swept Wing (FSW) Technology design goals by about 15 percent
Demonstrator Program traces the subsonically. Finally, Ship X1
aircraft's history from design testing provided evaluations of
through flight test. Brief handling qualities, military utility
descriptions of the aircraft and its and agility metrics below 20 degrees
flight control system provide insight angle of attack (AOA).
for evaluating this unique vehicle.
Wind tunnel design data substantiate Ship #2, which was modified to allow
theory and highlight potential high AOA testing, began flying in nay
solutions to a more "missionized" 1989. Its spin chute was designed to
aircraft. Flight test results assist the pilot in regaining control
validate the X-29's wind tunnel data in the event of a departure from
base and provide for piloted controlled flight. Control surface
simulation of possible improvements tutorial lights mounted in the
for the specific X-29 technologies. cockpit assist in this task. The
flight control system software was
INTRODUCTION significantly modified in order to
best utilize the various surfaces in
The X-29 integrates several different controlling the aircraft in a post-
technologies into one airframe as stall environment. One g envelope
depicted in Figure 1. The expansion was completed to 66 degrees
aeroelastically tailored composite AOA and ten degrees sideslip.
wing covers cause the forward swept Accelerated entry high AOA expansion
wing to twist as it deflects, allowed all-axis maneuvering to 45
successfully delaying wing degrees. The inherent high-lift
divergence. The thin supercritical capability of the forward swept wing
airfoil, coupled with the discrete allowed the X-29 to roll at 70
variable camber produced by the degrees per second under
double-hinged full span flaperons, approximately 2g conditions at 30
provide optimum wing performance at degrees AOA. The military utility of
a l l flight conditions. The aircraft this vehicle ie in a class by itself.
was designed for 35 percent Static
instability (time to double amplitude AIRCRAFT AND FLIGHT CONTROL SYSTEM
of about 0.15 seconds) by use of a DESCRIPTION
close-coupled, variable incidence
canard. Without it the wing-body Two essentially identical X-29s were
combination is near-neutrally stable. designed and built by Grumman
The canard, which has an area about Aerospace Corporation, Bethpage, New
20 percent of the wing area, produces York. To reduce overall program
lift and its downwash delays flow costs, the Air Force supplied several
separation at the wing root. The major components of the aircraft to
three-surface pitch control--the Grumman. These included the F-5A
canard, flaperon, and strake flap--is forebody and nosegear; F-16 main
used by the digital fly-by-wire gear, actuators, airframe-mounted
flight control system to control an accessory drive and emergency power
otherwise unflyable unstable veicle. unit: F-18 F404 engine: SR-71 HDP5301
The success of the X-29 really rests flight control computers: and F-14
with the integration of these accelerometers and rate gyros. Use
technologies into a single of these time-proven components also
synergistic configuration built for increased the reliability of the
drag reduction in turning flight. flight vehicle.
Two X-29 aircraft were designed and The X-29 flight control system (FCS)
built. The first entered flight is a triplex digital fly-by-wire
testing in December 1984 and system with triplex analog backup (as
concluded in December 1988, shown in Figure 2). The fail-
completing 242 flights and Over 200 op/fail-safe system used MIL-F-8785C
flight hours. The primary objective and MIL-F-9490D specifications as
of ship #I testing was to validate, design guides. Flying quality design
evaluate, and quantify the benefits goals were Level I for the primary
of the technologies on board, both digital mode and Level I1 for the
individually and COlleCtiVelY. analog back-up mode.
The first two years of Ship #1 Normal aircraft operation is
testing were primarily dedicated to accomplished through the normal
altitude and Mach number lg envelope digital (ND) mode with its associated
expansion. Performance testing functional options such as automatic
22-2

camber control (ACc), manual camber provides for better roll coordination
control (MCC), speed stability, at high AOA. Also assisting in roll
precision approach control (PAC), and coordination is a rate-of-sideslip
direct electrical link (DEL). ND feedback to the rudder. Since
also contains options in its gain substantial wing rock was predicted
tables for power approach (PA), up- for the X-29 above 30 degrees angle
and-away (UA), and degraded of attack, a high gain roll rate-to-
operation. aileron feedback loop has been added
to comperlsate for the unstable
The normal digital mode has a pitch rolling moment coefficient due to
rate control law with gravity vector roll rate. For a more detailed
compensation, driving a discrete ACC description of the control system,
system. This mode is gain-scheduled see Referenee 1.
as a function of Mach number and
altitude and incorporates a EVOIXlTION OF THE X-29 CONFIGURATION
sophisticated redundancy management
system allowing fail-op/fail-safe The X-29 Forward Swept wing
flight. MCC is a pilot-selected, Technology Demonstrator Program was
fixed flaperon sub-mode of ND used established to flight test an
for landing. The PAC function is a advanced air vehicle which
pilot-selected auto throttle system. incorporated forward sweep on the
The DEL function is a ground contact wing. Forward swsep was a well-known
control law set which is active when concept with fairly well-defined
any landing gear weight-on-wheel benefits, but because of the absence
relay is opsn. This function fades of certain enabling technologies, the
out the longitudinal forward loop concept had never been proven on a
integrator, allowing direct pilot high performance fighter. The
control of the canards during taxi, emergence of composites, and
take-off, or landing roll-out. Gain specifically their compatibility with
tables for degraded operation are aeroelastic tailoring, allowed a
activated by a failure of the lightweight, efficient solution for
Attitude Heading Reference System or delaying wing structural divergence.
any two of the three angle-of-attack
sensors. This function cannot be A n aircraft wing is swept, either
pilot-selected, nor can it be exited forward or aft, to postpone the drag
in flight. Degraded normal digital rise associated with transonic
operation is the last option flight. It can easily be shown
available during sensor failures (Reference 2 ) that for an equal shock
prior to automatic down-moding to wave location, shock wave sweep, wing
analog reversion. taper ratio, aspect ratio, and area,
a forward swept wing has a lower
The analog reversion (AR) mode is the leading edge sweep than an equivalent
back-up flight control system, aft swept wing. The result is a
designed to bring the aircraft safely lower profile drag and a lower root
back to base. The AR mods provides a bending moment. Conversely, if the
highly reliable, dissimilar control two wings are designed to have the
mode to protect against generic same bending moments, the forward
digital control failures. It swept wing's aspect ratio increases,
incorporates UA and PA functions producing a further reduction of the
similar to those of the ND mode. AR induced drag. These points are
contains no londtudinal trim illustrated in Reference 2 and
reproduced hers for convenience as
Figures 3 and 4.
$11 other aspects, it performs In addition to the transonic
like the ND control system. efficiency provided by an FSW, it
proves to be a good performer in the
The Ship # 2 flight control laws were near- and post-stall region of
modified to permit all-axis flight. Unlike an aft swept wing
maneuvering to 40 degrees AOA, and where flow separation starts at the
pitch-only mansuvering to as high as wing tips and may cause control
70 degrees AOA. Below 10 degrees, degradation, flow separation on an
the control laws are identical to FSW originates near the wing root and
those last flown on Ship #1. Between proceeds outboard with the wing tips
10 and 2 0 degrees, the high AOA and ailerons maintaining flow
modifications are faded in until attachment longest. The addition of
above 20 degrees they are fully canards on the X-29 enhances ths
functional. lifting performance of the FSW by
providing a downwash on the largest
The high AOA changes are fairly surface arsa region of the wing,
simple. A spin prevention logic is delaying separation in that zone.
active above 40 degress or below Upper surface pressures showed that
minus 2 5 degrees AOA with increasing separation actually originated just
yaw rate. The logic increases the outboard of the canard tip butt line
authority of both the rudder pedals and proceeded both inboard and
and lateral stick and disconnects all outboard from there. Wing separation
other lateral/directional feedbacks. is delayed in the shadow of the
Besides the spin prevention logic, an canard until the lightly-loaded
aileron-to-rudder interconnect canard itself begins to separate.
 22-3

The close-coupled, variable-incidence polar. The appropriate flaperon

canard on the X-29 provides positions were then scheduled with
additional benefits. Its large area angle of attack to achieve this
(20 percent of the wing) contributes polar.
significant pitching moment and
direct lift with small deflections. HIGH LIFT DEVICE INVESTIGATION
This contributes trim lift without
significant trim drag and enhances Early in the design process for the
the maneuverability of the X-29. The X-29, an effort was undertaken to
canards and Psw were specifically investigate various high lift schemes
optimized to produce minimum induced which might be beneficial to a
and profile drag over the X-29 low supercritical forward swept wing and
AOA flight envelope. canard configuration. Reference 4
provides the details of this wind
The addition of close-coupled canards tunnel study. A moderately cambered
on the near-neutrally stable wing- canard was tested. n o wing leading
b&y combination of the X-29 makes edge devices were examined as well as
the aircraft up to 35 percent two trailing edge devices. These
statically unstable. Figure 5 from particular devices were generally
Reference 3 shows significant relaxed accepted as producing the best high
static stability (Rss) at all lift performance on aft swept wings.
subsonic Mach numbers, with stability
returning at about Mach 1.3. By The remotely driven canards were
using the RSS to keep the canard from fitted with detachable leading and
overloading and helping to produce trailing edge flap pieces in order to
positive lift to trim, high levels of increase the camber of the baseline
trimmed lift are maintained through low camber airfoil. The canard was
the maximum lift coefficient for the movable 20 degrees trailing edge down
configuration. By employing RSS, the and 40 degrees trailing edge up.
variable incidence canard can unload Leading edge flap extensions were
at very high angles of attack, provided for 0, 5, and 10 degrees
providing the pitching moment down: trailing edge extensions were
necessary to recover to level flight. 0, 10, and 20 degrees down. The test
data indicated that canard lift
The final consideration in the effectiveness was independent of wing
evolution of the X-29 configuration leading and trailing edge devices
was the concept of optimizing the over the -4 to 22 degree angle-of-
wing profile for maximizing attack range and -4 to 16 degree
efficiency over all regions of the sideslip angles tested. Canard
flight envelope. This, of course, camber always provided a positive
requires variable camber. The lift increment, but in conjunction
simplest, most cost-effective method with a Krueger flap on the wing it
of producing variable camber on a reduced pitching moment effectiveness
thin supercritical airfoil is with at low angles of attack. The results
leading and/or trailing edge devices. clearly showed benefits from canard
various wind tunnel tests were variable camber, but when the added
conducted to ascertain the most complexity of incorporating movable
appropriate configuration. The next leading and trailing edge surfaces
section of this paper will provide was considered, the final X-29
details of this overall effort. Here configuration used a single low
I w i l l discuss only the concept that camber, all-movable canard.
was actually adopted for the
x-29. The wing extensions examined in the
wind tunnel study were a Krueger
The discrete variable camber system flap, a leading edge slat, a single
on the X-29 consists of double-hinged slotted extensible flap, and a double
full span flaperons. The flight slotted extensible flap. These
control system has a manual camber devices are shown in Figures 7 and 8.
control (Mcc) mode wherein the The Krueger flap had a 50 degree
flaperons can be set at a discrete angle of incidence, while the slat
setting by the pilot. The MCC mode was designed for 15 and 29 degrees.
was utilized to obtain aircraft The single slotted flap was
performance at a fixed flight adjustable in increments of 20, 30,
condition over a range of wing and 40 degrees down. The double
camber. The automatic camber control slotted flap had a fixed 20 degree
(ACC) mode was used to optimize the deflection on the first flap and
wing's performance over all flight incremental settings of 20, 30, and
conditions. In the ACC mode, the 40 degrees relative to the first
flight control computers used real flap.
time air data to adjust the flaperon
settings according to a prescribed Figure 9 provides the lift
schedule. Figure 6 shows coefficient results for a plain flap,
schematically how the ACC scheduling a configuration quite similar to the
was derived from MCC results. wind actual aircraft hardware. Integral
tunnel-derived drag polars were to both curves are the effects of
plotted for specific flaperon canard and wing stall. The 25 degree
settings. The dashed line, faired flap deflection significantly
tangentially across the polars, then increases lift. It also accentuates
defined an optimum variable camber the separated flow effects on the
22-4

canard-wing combination. degrees, the ACC schedule was not

correctly optimized for lift and
Figures 10 and 11 give the lift drag.
performance for the single and double
slotted flap in combination with the The manual camber control mode of the
two leading-edge devices. In all flight control system was also
cases, the improvements were utilized during performance testing.
significant. The best Combination in Figure 13 shows both MCC and ACC
this test was the K N e g e r flap plus flight data at 0.90 Mach number and
the double slotted trailing edge flap 30,000 feet. As is evident, the ACC
resulting in an improved lift scheduling did an excellent job of
coefficient of about 70 percent. optimizing the discrete camber
However, it is clearly evident from polars. The MCC uses fixed flaperon
the data that the leading edge settings to achieve discrete values
devices lead to total separation and of wing camber.
loss of lift in the 15 to 20 degree
angle-of-attack range. Only the To complete this low angle of attack
trailing edge flaps acting alone are performance analysis, let's look at
suitable for the high angle-of-attack the opposition. Figures 14 through
regime in up-and-away flight. The 16 provide aeroperformance
leading edge devices excel in the comparisons between the X-29 and a
power approach mode of flight. current high performance fighter. No
attempt has been made to normalize
AIRCRAFT PERFORMANCE the data or optimize either aircraft.
It is simply an aircraft-to-aircraft
The lift and drag characteristics of snapshot. The induced drag polars
the X-29 were determined by presented show a consistent trend.
performing push-over, pull-up ( P O W ) The polar shape of the X-29 has been
and wind-up turn (wrpr) maneuvers at greatly improved. Although the X-29
constant power settings. POPU was specifically optimized for a 0.95
maneuvers were used to obtain data lift coefficient at 0.9 Mach number
for load factors under 2 9's and the and 30,000 feet, the improvement
WUT maneuvers for 2 g's and above. exists through its entire performance
Data was obtained for angles of range.
attack up to 20 degrees and for
symmetric load factors up to 6.4 g's While these comparisons show a
as corrected to 15,000 pounds gross significant advantage to the X-29,
aircraft weight. Both the automatic the X-29's lift-to-induced drag
camber control and manual camber performance is far superior to its
control options of the normal digital lift-to-total drag performance. This
mode of the flight control system is explainable by the fact that the
were investigated. X-29 was put together from many
pieces and parts. This resulted in a
Drag polars were acquired for 0.4 to rather large profile drag for the
1.3 Mach numbers. The polars were aircraft. However, at 0.95 lift
compared to wind tunnel predictions coefficient, 0.9 Mach number and
which also used the ACC control 30,000 feet altitude, the X-29 excels
surface schedules. In general, in spite of its high profile drag.
subsonic and transonic flight test In the fighter-class arena,
results showed lower drag performance of the X-29 is
coefficients than predicted. impressive.
supersonically the test data either
matched predictions closely or showed MANNED SIMULATOR PREDICTIONS
higher drag coefficients. The
highest lift coefficient reached in One of the fundamental
flight on Ship #1 was 1.63 at 0.4 characteristics of a forward swept
Mach number. This was a result of wing is its ability to achieve
the conservative angle of attack (15 relatively high angles of incidence
degrees) and load factor (6.4 g) before the wing totally separates.
limitations imposed on the aircraft. Once the aircraft has entered post-
These restrictions were later stall flight, its lifting performance
relaxed, but a repeat of the is highly dependent on the
performance measurements was not interactions of the wing's wake
accomplished. region and disturbances from other
portions of the aircraft. In the
Shown in Figure 12 is the comparison case of an aft-swept wing with high
between a wind-tunnel-predicted drag lift devices such as those on the
polar and the actual flight-derived F-18, fuselage, LEX and leading edge
polar at 0.9 Mach and 30,000 feet flap influences are swept generally
pressure altitude. The wind tunnel outboard, adding additional
data clearly over-predicted drag. turbulence to the wing's wake (Figure
Note the break in the curve which 17, reproduced from Reference 5).
corresponds to an angle of attack of The forward swept wing of the X-29,
about seven degrees. It happened without additional high lift devices,
that the scheduled canard position is able to avoid additional
reversed direction and the flaperon interactive turbulence since its Wing
position schedule reached a limit tips are always in clear air.
simultaneously. This would suggest Without these additional influences,
that at angles of attack below seven the turbulent detached flow is able
 22-5

to reattach itself, in a reverse magnitude improve when moving the

direction, over a rather large four c.g. aft. Although an improved
foot section of the wing at 30 control surface schedule and a more
degrees AOA. This is shown in Figure aft center of gravity produce lift
18, reproduced from Reference 6. benefits, care must be taken to avoid
This suggests additional lift, making canard saturation and subsequent loss
the X-29 a natural high lift, high of pitch authority needed to recover
angle-of-attack aircraft. the aircraft to low angle of attack.
Continuing this theme of increased C O N C W I N G REMARKS
capability of a forward swept wing
configuration, what can be done to The x-29 with its forward swept wing
further improve the X-29's has been shown to be an inherently
performance? Although the X-29 Ship high lift aircraft. A wind tunnel
#2 was not specifically instrumented study of canard and wing high lift
to directly determine the lift-drag devices was reported. Flight tests
performance at high AOA (the of the unique X-29 design have been
propulsive thrust measurement system performed at high and low angles of
had been removed from the engine), attack using both test aircraft. The
the aircraft flight-validated math drag reduction design goals of the X-
model was sufficiently mature to 29 were demonstrated below 20 degrees
provide accurate manned simulation using Ship #1, while the high angle
predictions. of attack capability of the forward
swept wing design was explored using
Figures 19-22 show the results of a Ship X2. A flight validated math
canard optimization study for the model was used in the X-29 manned
very high AOA regime. With flaperon simulator to show that improvements
and strake deflections defined and in overall aircraft performance were
fixed, a unique trimmed canard possible through a simple
schedule exists at the prescribed rescheduling of the flight control
flight conditions. These schedules surfaces. This forward swept Wing
are shown in Figure 1 9 . Also shown canarded configuration has proven
are two ACC canard schedules which aerodynamically superior to all
bracket the usable center of gravity current high performance fighters.
range for the aircraft. The ACC
schedules were established early on REFERENCES
in the program and were intended to
be the most efficient at producing 1. Pellicano, et al, "X-29 High
lift and pitching moment while at the Angle-of-Attack Flight Test
same time providing a margin of Procedures, Results, and LeSSOnS
safety suitable for a highly unstable Learned, "Society of Flight Test
aircraft. Only for the plus 30 Engineers 21st Symposium, August
degree strake deflection did the 1990.
canard achieve positive deflection
(trailing edge down). This was its 2. Frei, D., and Moore, M., "The X-29
most highly loaded schedule. - A Unique and Innovative
Aerodynamic Concept, 'I SAE
with all three control surfaces Aerospace Technology Conference
lifting, the lift curve should show a and exposition, October 1985.
large improvement over all other
combinations presented. However, 3. Walchli, L., "X-29: Longitudinal
Figure 2 0 shows that this only holds Instablilty At High Angle of
true above 23 degrees AOA. Based on Attack," AGARD Guidance and
the lift curves of the various Control Panel Workshop On
combinations, a more efficient Stability In Aerospace Systems,
scheduling of the control surfaces 23-26 June 1992.
would result from using the original
ACC below about 23 degrees AOA and 4. Frei, D., "Forward sweep wing
then transitioning to a higher canard Technology Integration and
-
loading.
such a solution would, of course,
-
Evaluation Study
High Lift Systems
Task 1 8 and 1 9
Investigation," AFWAL TR-81-3024,
increase overall drag on the January 1981.
aircraft. Figure 2 1 confirms that
canard loading has increased since 5. Fisher, et al. "In-Flight Flow
additional drag begins to occur above visualization Characteristics of
20 degrees AOA. But Figure 22 shows the NASA F-18 High Alpha Research
clearly that this solution results in vehicle at High Angles of Attack,"
a net improvement in the drag polar NASA TM 4193, May 1990.
above 23 degrees AOA, allowing for
improved turning performance. C 6. Tate, B., "Surface Flow
improves almost ten percent overUme visualization -
Interim Report,"
baseline ACC. unpublished, circa Fall 1990.
The ACC curve for an aircraft center
of gravity of 453.9 inches exhibits
the benefit of relaxed static
stability. For the case presented,
both the lift curve shape and
22-6

FIGURE 1. X-29 TECHNOLOGIES

t9.w.-

FIGURE 2. X-29 FLIGHT CONTROL SYSTEM

YAINTAININO
WINO AREA * TAPER RATIO
ASPECT RATIO SHOCK WEEP

T
M O
AR = 6.81

ARC OF CONSTANT RADIUS

INTERSECTION ABOUT COMMON INTERSECTlON
OF STRUCTURAL AXIS 6
PIVOT STATION

Am

FIGURE 3. FORWARD VERSUS AFT SWEEP

REOUCED BENDINQ MOMENT
- I
FIGURE 4. FORWARD VERSUS AFT SWEEP
AR = 6.04

-
HIGHER ASPECT RATIO

c
FLAP 3
-ORIOINAL onumm WT.
-I FLAP 2
0 WINO TUNNEL DATA
40 0 FLIOHT DATA. 3OK Fl
0
k
-s
FLAP 1
UNSTABLE
20 t
W
c,
CL 8
%MAC t
2

STABLE I
0 0.4 0.8 1.2 1.8 2.0
I
MACH NO. DRAG COEFFICIENT (C D)
RGURE 5. X-29 VARIATION OF LONQINDINAL STATIC
STABILIN WITH MACH NO. FIGURE 6. X-Z$ ACC SCHEDULE DERIVATION
 22-7

a..

p-
US 1..5

111s * o I D

a.115

a- ?U. 0-

flGURE 7. FSW HIGH UFT LEADING EDGE DNICES FIGURE 8. FSW HIGH LIFT TRAIUNG EDGE DEVICES

10. KRUEQER F U P E

2.4

2lD

1J

CL 1.2

0.I

0.4

I I I I I I
O 10 20 30 40 50
I I I I I
ANGLE OF ATTACK. (DEGREES) \ 10 m x . m m
ANOLE OF ATMCK, (DEOREES)

FIGURE S. LIFT CURVE FOR PLAIN FLAP FIGURE 10. LIFT CURVE FOR SINGLE SLOTTED FLAP

3 1.2

g 1.0
aE 0.8
B
0.6
t 0 FLIGHT TEST DATA( 0.1 MACH
3
0.4 - - - W.T. PREDICTION
0.2

I0 I
10
I
20 w
I I
40
A 0
o
I ,
0.05 0.10 0.15 0.20 0 . u
I I
oso
I
0.55
I
a.a
ANOLE OF Ana=. IDEQREESJ INDUCED DRAG COEFFICIENT (Col)
FIGURE 11. LIFT CURVE FOR DOUBLE SLOTTED FLAP FIGURE 12. X-28 AIRCRAFT PERFORMANCE
 r
22-8

1'6
1.4
.. . [I 1.4 -
1.2 1.2 -
u4

s 1.0
8k 0.8
YACH 0.6
H=SOKFf
8 0 0 FLAP POS.
0.6 0 0 -50
3 0 0:
0.4 A A
- ACC

L
0 0 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40

INDUCED DRAG COEFFICIENT (CDI)

0 0.02 0.04 0.00 0.08
INDUCE
0.10
Dma WEFFICIENT
0.12

FIGURE 14. DRAG POLAR COMPARISONS MACH 0.6

0.14 0.10 0.18 0.20
(cDI)
-
FIGURE 13. X-28 ACCMCC COMPARISON

i.4 - 1.4

- -
4
i.2 1.2

0 -1

YACH 1.2
HrWKFl
wcn 0.0
0.0 H.2QKFT
0

= 0.4 CURRENT
FIOKTER
t
3
0.4

01 0.2 FIGHTER

0-
0 0.02 0.04 0.00 0.08 0.10 0.12 0.14 0.10 0.18 0.20 0 0.02 0.04 0.06 0.06 0.10 0.12 0.14 0.111 0.18 0.20
INDUCED DRAG COEFFICIENT (Col ] INDUCED ORA0 COEFFICIENT (Col )
FIGURE 15. DRAG POLAR COMPARISONS -MACH 0,s FIGURE 16. DRAG POLAR COMPARISON -MACH 1.2

FIGURE 18. X-20 FLOW VISUALIZATION

FIGURE 17. F-18 F L O W VISUALIZATION 30 DEGREES AOA
 22-9

-
-.-.-..-.-.--.
. .. ACC ca.445.1

10. - . --
Ycc
YCC
YCC
ACC
CQ445.1
CC.445.1
CQ.445.1
CQd53.0
FLAP.+ZQ
FLAP-rZO
FLAP.*ZO
8TRAKE.rlQ
STTRAKE.0
STRAKE..%O

I 1 1 1 1 1 1 1 1
.BO. 5. 10. 15. 20. 25. 50. 35. 40. 45. 50. 5s.
5. 10. 15. 20. 25. 30. 35. 40. 45. 50. 55.
ANGLE OF ATTACK (DEGREES)
ANGLE OF ATTACK (DEGREES)
I FIGURE 19. CANARD OPTIMIZATION STUDY - CANARD SCHEDULE FIGURE 20. CANARD OPTIMIZATION STUDY - LIFT CURVE

2.4
- ACC CG1445.1
2.4 . -
. ....
hcc
, YCC
c0.4.5.1
Cad4s.l FLIP-rZO STRAKE-40
. . . . . . .- ---.-
., , . , .
---
YCC CO-445.1 FLAP-.20 STRAKE10

--
- ----
- - -.
* MCC
MCC
CG1445.1
CGd45.1
FLAPnt20 STRPXEn40
FLAP=+20 STRAKEsO
FLAP;+ZO STRAKE=-30
2.2 -LZZi
YCC
KC
C0..4s.?
ca.4s3.s
FLIP-.ZO STRAKB--l?. ,,

-
2.0
- MCC
ACC
CGn445.1
CG.453.9
A Pi0 -
0"
- 1.6 0,
t- 1.8 -
I-
z
z Y
w0 1.2
0
U
1.6 -
w
U
U
w
0
0
0
-1.4
0
L 1.2 -
A

1.o 1 p" I;.I I , I I I I I I

O b. :o :5 o: 25 310 5: lo 315 io 15 io. i
.8
0 .2 .4 .6 .8
DRAG COEFFICIENT (C Dl
1.0 1.2 1.4 1.6 1.8 1 0
ANGLE OF ATTACK (DEGREES)

FIGURE 21. CANARD OPTIMIZATION STUDY - DRAG CURVE FIGURE 22. CANARD OPTIMIZATION STUDY - DRAG P O W R
 23-1

A FAST COMPUTING METHOD FOR THE FLOW OVER HIGH-LIFT WINGS

K.Jamb
German Aerospace Research Establishment (DLR)
Institute of Theoretical Fluid Mechanics
Bunsenstr. 10,3400 Goettingen, Germany

SUMMERY S starting point of rear separation re-

gion at upper surface
A quasiSdimensional method for analysing the starting point of rear separation at
viscous steady subsonic flow over wings with flaps lower' surface, close to trailing edge
for high lift is briefly presented. The total iterative point on upper boundary of rear sep-
procedure combines a 3-dimensionalinviscid lifting aration region, see fig. 4
surfaa theory with a 2-dimensional surface-singu- local flow velocity and free stream
larity method for analysing multielement airfoils in
a cuwed basic flow field. This method also includes velocity
boundary layer calculations and a model for rear vortex distribution on airfoil surface
separation. Also, small compressibilityeffects are = tangential component of flow ve-
accounted for by simple cofiections, and ground ef- locity
fects are included by means of the reflected image source distribution on airfoil surface
technique. normal component of flow veloci-
ty
First attempts to validate the method by a few theo- W velocityyinduced by a vortex system
ry-experiment comparisonsare reported.The results XYY Y Cartesian coordinate system of the
are encouraging but more experimental data are wing, fig. 2
needed for a thorough validation. The computing a angle of attack of an airfoil section
time requirements of the method are modest.
a*,a** effective basic flow angles at trailing-
LIST OF SYMBOLS
edge and quarter chord point
ag wing incidence
influence coefficients in eq. 2 'YYc1 dimensionless quantities related to
chord of airfoil local lift and pitching moment ac-
lift-, drag-, and pitching moment coeffi- cording to eqs. 2a
cients of the total wing 6f flap deflection angle
lift-,drag-, and pitching moment cocfi- 4% potential function of basic flow field
cients of an airfoil section rl nondimensional spanwise distance Y/s
cp pressure coefficient @-p,)/(~~oop/2) A aspect ratio
f damping factor in eq. 6
hD ground distance in fig. 12 Subscripts and superscripts:
L reference length for a wing; usually
chord of center section with flaps re- B basic flow
tracted 2-d two-dimensional
M number of wing sections in the Trucken- ef effective
brodt method v, n at section No. v resp. n
M, or Mawfree-streamMach number * trailing edge
P static pressure
Re, Reynolds number, based on V, and L
** quarter chord point
r straight distance between two points
S wing semi-span
co-ordinate along S,
Sj surface of airfoil element No. j
23-2

1. INTRODUCTION crease when the wing approaches the ground.

Eventhough the most accurate solution for such
Safety and economy at take-off and landing of an complicated 3-d flow might be obtained with Navi-
aircraft are dependant on the quality of the high lift er-Stokes methods, for reasons of economy and fea-
system of the wing. This consists of various extend- sibility these methods are not considered here.
ed flaps, by which the lift coefficient CL can be in- Instead, we use a combination of available, ap-
creased considerably, thus enabling the .wingto proved, economical 2 d and 3-d methods, try to in-
maintain sufficient lift L , - q VW2atlow speed V
,. clude the essential flow features, but also accept
some simplifications and restrictions.
For the design of good high lift systems engineers
need very extensive testing or a good and fairly fast At a wing of high aspect ratio and low sweep the
computing method for the flow around multi-com- flow around a wing section is approximately 2di-
ponent wings. The most important features of such mensional. But, contrary to the rectangular infinite
a flow are the complexity of the geometry and the wing, the effectivebasic flow field at each section of
importance of viscosity effects, including flow sep- a general finite wing is not identical with the on-
aration. coming parallel flow. It is somewhat changed, be-
cause the different vortex systems of the finite wing
A good review of existing methods has been given and the infinite wing induce different velocities.
recently in ref. 1. The development of “coupled sep- The differenceof these velocities has to be added to
arated flow methods” for multi-element airfoils the oncoming flow to give the effective basic flow
started in the 70th(refs. 2,3,4). Some progress with for a 2-dimensional calculation at a section. Gener-
those methods has been achieved since (e.g. refs. ally, this basic flow is curved and the average angle
5,6,7,8) and recently several authors dealt with ap- of attack is smaller than the wing incidence.
plying Navier-Stokes methods to multi-element air-
foils (e.g. refs. 9.10). But none of these methods is If the effective basic flow field is known, the span-
perfect and all are limited to 2-dimensional flow. wise lift, pitching moment and drag distributions
can be computed by applying a 2-d viscous flow
Can 2-dimensional methods, even if further im- method for multi-element airfoils to each section.
proved, be a real help for the design of a 3-dimen- On the other hand, with a known lift and moment
sional wing? With this important question in mind, distribution, the induced velocities and thus the ef-
a quasi-3-dimensional method has been designed, fective basic flow at each section can be computed
combining 2-d viscous flow calculations for the approximately by a reverse application of a 3-d lift-
wing sections with a 3-d lifting surface method for ing surface method. Therefore, both methods are
taking account of the essential effects of the trailing applied by turn. starting with a first approximation
vortices. First the method was worked out and test- for the spanwise lift and moment distribution and
ed for clean wings only (ref. 11) and recently it has computing the converged lift distribution by itera-
been extended to wings with flaps by H.N.V. Dutt tion. The total procedure is roughly sketched in
and the author (ref. 12). In the present paper this figure 1 and the main components are treated briefly
method will be explained briefly and some results in the next chapters.
will be shown and discussed.
2.2 3-dimensional lifting surface method
2. QUASI-3-DIMENSIONAL COMPUTING
METHOD The method used at present for the 3 d calculations
is the well known lifting surface method of Mul-
2.1 Over-all concept thopp and Truckenbrodt (ref. 13) with Hummel’s
extension for including the ground effect by reflect-
In order to develop an economical computing meth- ed image technique (ref. 14).
od to support the aerodynamic design of high-lift
systems for civil aircrafts, one has to consider the This method uses horseshoe vortices in the wing
flow around multi-component wings of moderate to projection plane (figure 2) with a continuous 2-pa-
high aspect ratio, operating at low speed and high rameter vortex distribution in the chordwise direc-
lift coefficient. For such a flow it is crucial to take tion and a spanwise discretisationwith M wing
account of viscosity effects, including partial sepa- sections at
ration. At take-off and landing also ground effects
may be important causing sometimes lift de- 11, = Y,/S = c o s ( m / ( M + 1 ) ) , v = 1,2,...J4. (1)
 23-3

The law of Biot-Savart leads to linear equations

for the induced velocities w * at~ the trailing
edge and w * * ~at the quarter chord points of
input of wing geometry and flow data
and preliminary calculations these sections:
M
w : K = Ya+c:a Pa)
first approximation of spanwise a= 1
lift and moment distribution *
and
M

n= 1
3-0 LIFTING SURFACE THEORY
(inviscid. incompressible)
Here the y and p are closely related to the local
lift and moment coefficients c1 and c,, namely
+
effective basic flow
I
read next
r, = Cln LV,f I (4s V,)
at each section incidence
and
I 4
P,, = c,, LVef J (4s V J

and the coefficients B*, thru C**, are determined

(incl. boundary layer, separation. by the geometry of the wing projection and the num-
compressibility), applied to each section
ber of sectionsM. In case of ground effect they also
depend on ground distance and incidence.
spanwise lift and moment distribution The normal application of the method consists of

-no
I
converged ? yes -
final calculations +
output for present
wing incidence
determining the velocities to be induced, from the
“equivalent thin airfoil” geometry at each section.
and then computing the values of y and j~by solving
the system of linear equations (2). For more details
seeref. 11.
In our total procedure (fig. 1) the lifting surface
Fig. 1 Sketch of total procedure method is used in the normal way only once in the
beginning, to get a first approximation to y and p.
Within the iteration loop it is used only in a reverse
way, i.e. finding the induced velocities w*, and
~ known approximationsto y and p by
w * * from
simple matrix-vector multiplication.

2 Y Then, the effective basic flow velocities at quarter

chord and trailing edge of wing section No. v are ob-
tained by vector addition:

v,
.. ..
= v _ + w , -94 v
*.

and (3)
root wing projection plane X-Y
SeclOn
I .
v: = Vm+wv--&d v

Fig. 2 A multi-component wing geometry, with w2-d meaning the veloctiy vector induced by
vortices and some notations the vortex lines of the infinite rectangularwing (2-d)
with the section No. v and constant y = yv and p =
p,,. The formulas for evaluating the basic flow ve-
locities are given in detail in ref. l l.
23-4

L
A

Generally, the resulting basic flow velocities will

have somewhat different directions at quarter chord
and trailing edge. So the basic flow field for the sub-
sequent 2d calculation is no longer parallel but
curved. At present this flow is approximated by a
circular flow, produced by a remote vortex of proper
strength and position (see firmre 3).

2.3 2-dimenslonal multlelement alrfoll

method

For analysingthe approximately2-dimensional, vis-

cous flow mund the multielement airfoils at the
wing sections, the author's method (ref. 2) is used,
with the extensions for multiple separation (ref. 6),
ground effect (ref. 15) and cuwed basic flow (ref. Fig. 3 Airfoil section in a curved basic flow
16). Given below is a very brief presentation of the field, induced by a distant single vortex
essentials of the method only. arlcr kviscid llow

The flow field around a multi-element airfoil is di-

vided into different regions (see fimre 4). The outer
inviscid flow is treated by a surface-singularitytech-
nique, using a vortex distribution vt(s) on the sur-
face S of each element. In addition, source distribu-
tionsVJS) are used to simulate displacement effects
SLAT W I N G FLAP
of boundary layers, wakes and separated flow re-
gions, via the "transpirationmethod". For the
boundary layers Rotta's integral method (ref. 17) is
applied, and small compressibility effects are ac-
counted for by a simple correction applied to the
surface pressure distribution of the incompressible
flow ( see ref. 11, page 23).

For the inviscid incompressible flow with a basic

flow potential $B the kinematic flow conditionleads
to the following system of coupled linear integral
equations
3-element airfoil system
Re = 1.8 x106 , M, = 0.123

with (4)

,
with k=1,2,...,N for an airfoil system with N ele- -a 0 cd- '. 0
em-
ments. Here, t and n indicate the directions tangen-
tial resp. normal to the airfoil surface.
Fig. 4 2-dimensional 3-element landing con-
Accounting for ground effect via reflected image figuration with different flow regions.
technique leads to extending eqs. 4 by additional Pressure distribution and lift-, drag- and
terms ( see ref. 12, page 9). moment coefficients from theory and
experiment.
 23-5

From eqs. (4). after some transformation and discre- ities in the wake see ref. 6 and 21.
tisation, vt can be computed numerically, if the basic
flow (b)and the source disvibutions v,, aR given Apart from the relative simple wake model, our
and if one condition is added for each airfoil element present method contains some more simplifications:
to fix its circulation (Kutta condition or specified closed separation bubbles at the lower surfaces of
circulation r ). the slat and the main element are simply replaced by
Solving eqs. (4) for different right hand sides and estimated contour fairings. Small compressibility
different additional conditions, a set of “fundamen- effects are postulated, and possible confluence of
tal flow solutions” can be obtained. Then, by super- boundary layers and wakes is not yet accounted for.
position, a flow can be constructed, which contains Nevertheless, as can be seen from fig. 4, the method
a simulated separated flow region. This is demon- gives good results with respect to lift and pitching
strated in figure 5 for the simplest case of parallel moment, as long as rear separation is limited to the
basic flow and sources on the rearmost element only. flap, i.e. below maximum lift. However, drag pre-
(For curved basic flow a solution with the $B of that diction is not yet satisfactory.
flow has to replace solutions (al, ad).
2.4 Some details of the total procedure
(a.~Parallel
- 1 ~- basic flow in (b1Circulatorv flow around one
17 x -direction of the airfoils
Input and preliminary calculations (see fig. 1):
The wing geometry shall be given by streamwise
wing sections at a sufficient number of spanwise
:@vn=o stations, e.g. 8 “input stations” on the half wing in
figure 6. Their number and distributionshall be such
(a,)ParalleI basic f l o w in (d1Outflow from o n e of the that the data of intermediate sections can be found
z-direction by simple interpolations.For the M “Truckenbrodt
stations”, defined by eq. (l), the geometric data are
then to be prepared, includingthe influencematrices
B* thru C**. Also the aerodynamic characteristics
cl(u) etc. of all sections are now computed for given
Reynolds and Mach number, assuming parallel ba-
sic flow. Finally, normal application of the lifting
surface method gives a first approximation to y and
P*

Fig. 5 Four fundamental flow solutions and

flow with model for rear separation.

For the separated flow region an isobaric model is

adopted, obtained approximately by demanding
equal pressure at the points S,T, and U. This can be
achieved by finding proper weight factors for the Semi span = 4 L , Aspect ratio = 7.878
fundamental solutions (b) and (d). Airfoil: inboard - GA(W)-1 + 30% Fowler flap
outboard - GA(W)-1 clean
So far a potential flow with a rear separation model
can be found for any position of the starting point S. Fig. 6 Geometry of part span Fowler flap con-
But to be physically realistic S must just be the figuration (for flap deflection 6f= 10’ ).
boundary layer separation point of this flow. To find
the right position of S the potential flow calculations Iterations :
and subsequent boundary layer calculations are ex- For reasons of economy the inner iteration loop in
ecuted for various positions of S,starting near the fig. 1 is used in two different versions. First the 2-d
trailing edge and moving S upstream until the result- calculationsare replaced by simply interpolating the
ing separation point coincides with S . lift coefficients etc. from the pre-calculated aerody-
For extending the method to multiple separation namic characteristics, neglecting the curved basic
and for refining the wake model by adding singular-
23-6

Port span Fowler l h p config~ration.6~

=20*
flow effect. Here the effective parallel basic flow is ~ , = 0 . 2.Re=22.106 .a, = 8.
taken to be defined by

vef = V” and a=,= (a” + 2a*)/3

An improved approximationfor the lift distribution

(5)
-
Symbol

-6-
Damping
factor
OJO
0.20
0.25
Na of
it ero tions
7
6
5
can thus be obtained quickly with about 20 fast iter- + 0.30 L
--*--
0.35 diverged
ations. Then, starting from this approximation and _,.- 0.LO diverged
applying now the full 2d calculations with curved
basic flow, the final lift distribution etc. can be ob-
tained with only a few of these comparatively ex-
pensive iterations.
In both phases of the iteration process it turned out
to be important to apply smoothing and damping to
the intermediate results and to take care for a smooth
transition to zero at the wing tip. Especially the
damping factor for the “curved basic flow iteration”
can have a considerable effect on the convergence
history, as demonstrated in firrure 7. Here the damp
ing factor f is defined by 2 3
Naof iterations -
y i+ y ‘-‘+f. (7 L yi - 1 ) ( 6)
Fig. 7 Effect of damping factor on con-
withy being the result of the i-th iteration. Figure 7 vergence history
shows best convergence with f = 0.30, but diver-
gence with too high f. Port span Fowler flap configuration. 61 = L O O
No. of input stotions:8. damping factor 0.25
M, -0.2 , Re =2.2x106 ag = B O +
Final calculations and output: ag = L o +
Finally the coefficients CL, CD, CM for the total lift,
drag and moment of the wing are found by integra-
tion of the spanwise distributions, taking account of
the different local directions of lift and drag, for ex-
1 -0.62

ample r. I
+1 -0.58

cr. = J[ y(q) cos (ag- ac,)-

-1
(7)
- 6 (W sin (ag- a=/>
I (VCf / V,) drl
with 6 = cd L Vef/ (4s V
,
) being related to the drag.

Now, all results of the last iteration are available for

output, including the chordwise pressure distribu-
tions at the wing-sections.
At the end it may be mentioned, that the inviscid lift-
ing surface method gives good first approximations
to y and p only for low wing incidence a,.The com-
puting time per case can be reduced considerably if
the computation is done for a whole series of a,-Val-
ues, starting with a moderate ag,increasing agstep- I I 1

20 30 LO 50 60
wise and using always the converged y and p of the M-
previous a, as first approximation for the next one.
Thus all the preliminary calculations have to be car- Fig. 8 Effect of number of “Truckenbrodt-
ried out only once and there are always reasonable stati0ns”h.l on converged values of
first approximations; CL, CD, CM.
 23-7

3. SOME RESULTS Next, figure 9 shows that the final results can be
somewhat affected by the choice of the input sta-
Before trying to validate the method by comparing tions. With stations No. 4 and 5 being positioned
computed and experimental results, some computa- nearer to the geometricdiscontinuity(flap tip atq =
tional studies were carried out to see how conver- 0.50). the y-distribution drops more sharply and the
gence and final results are affected by the choice of total lift coefficient CL is slightly reduced. Recom-
various parameters. mendations for favourable positions of the input sta-
tions can be given only after comparing with
For the wing configurationof fig. 6 the effect of experimental distributions, carefully measured es-
damping factor f on convergence history has been pecially in the vicinity of geometric discontinuities-
shown in fig. 7. The effect of the chosen number of The preliminary tests with the wing of fig. 6 are
“Truckenbrodt-stations”M on the final force coeffi- concluded by comparing our results for a low flap
cients of the wing is demonstrated in figure 8. From deflection and a low incidence with the results of an
these and other examples we found that f = 0.25 and inviscid vortex lattice method (ref. 18). Fipure 10
M > 40 can be recommended. contains the y-curves and the values of the total lift
0.151 I 1 coefficient CL. As expected, our viscous method
gives somewhat reduced lift.

Port span Fowlerflap configuration, 6, =20°

M..=0.2, R e = 2 . 2 x 1 0 6 , C ( g = 8 ’ , M : L 7

Fig. 9 Variation of spanwise ydistribution with altered positions of the input stations.

Part span Fowler flap configuration, S, = 10’

k=
0.2, = 4 O
vortex lattice method, inviscid;
t CL = 1.025
--.e --
present method with Re = 2.2 x 10’; CL = 0.950

Fig. 10 Comparison of spanwise y-distributions, computed by inviscid vortex lattice method

and present viscous method.
 -
23-8

The first comparisons of theoretical results with ex- is most important at wing regions with a strong
periments were camed out for the simplest case, a spanwise lift gradient.
clean rectangular wing of aspect ratio 3.1, shown in In finure 12 the lift characteristics are shown for the
finure 11. For this wing detailed measurements are airfoil section (A = =) and for the wing (A = 3.1)
available in ref. 19, including high angles of attack with and without ground effect (h& = 0.33 resp.
with partial flow separation. The measured pressure -). In all cases theory and experiment compare
distributions compare fairly well with the results of well, including maximum lift. Finally, in fimre 13
the “advanced method”, which contains the calcula- the ground effect was investigated in more detail.
tions with curved basic flow. The curved flow effect Besides fair prediction of all aerodynamic charac-

/-
Rectongu tor Y. rl=y/s teristics, it may be noteworthy that the maximum lift
wing is lowest at the medium ground distance hD / L =
0.99. This “hight instability of lift” is indicated by
A = 2 S / C = 3.1 theory and experiment as well.
Rectangular wing, section NACA LL15
Re = 2.1 x IO6 , Ma, = 0.17

section NACA 1,415

-3

IO

-2

CP

I -I
ground

0 ,,,
experiment
0

Chordwise p r e s s u r e distributions
ot 3 s e c t i o n s
0
0”

-
ioo
present theory
ail - io0

D NACA Rep 826 (extropol.1

R e r 2 . 1 = lo6 , M0,:0.17
1 o -
experiments trom DFVLR IB 222 -82 A10
0 OFVLR- FE B I - l 2
Fig. 1 1 Pressure distributions at different Fig. 12 Lift characteristic from theory and
stations of a rectangular wing with experiment for airfoil section (1) and
aspect ratio A = 3.1. for the rectangular wing without and
with ground effect (2), (3).

Rectongulor wing, hs3.1, section NACALLlS -----

-
- - present theory
Re :2.1 = 10’ . Ma, x 0.17 o I o experiments i t r ~ r nDFVLI(-IB 212-c.2~101

Fig. 13 Total lift, drag and pitching moment coefficients of the rectangular wing for three
different ground distances hD / L.
 23-9

Next, a rectangular wing of aspect ratio 6 equipped and the main element nose at ag= 6'. There is prob
with full span leading and trailing edge flaps is ably a large separation bubble which was not mod-
shown in figure 14. For this configurationmeasured elled Properly. At the high ag= 18" there is very
pressure didbutions at section A-A are availablein good agreement in this =@on, Probably due to dis-
ref. 20 and compared with our theoretical results for appearance of the bubble. Also the separation on the
flap (level and extension of constant pressure) is
a low and a high wing incidence ag At ag= 6'
well predicted. The pressure differences on the rear
the results for 2-d flow are also shown, and the effect
part of the main element upper side might partly be
of the finiteness of the wing can well be seen.
due to geometrical differences.
The xsults of experiment and present method com-
pare fairly well, except for the region below the slat

Vm
r-3L 1
! .A

I
,5101

I
l y
3-component rectongulor wing
1' ;
f !

F wkrflo
I

p[

A
I
-6.01
Mm=O.I.Re=2.36xlO6,

0 upper aide
-do : 6 *

Experimcnl(NASA TM 4 0 4 0 , 1 9 8 8 1
ix 0 lower r i d e

- Theory (presenl method)

---Theory (2-dlmenrionol)

/ ' -4.01

L.E. s:ot 1 Moin e l e m e n t TE. f l o p

- 3.0 - 3.0
CP

- 2.0

- 1.0 -I .o

0.0 0.0

1.0
0 x/L 0.1
..-
0 0.5 X/L
'
1.0
1.~
.0
0 x/L 0.25

:;Dl
~0.1,Re = 2 . 3 6 ~IO6, d g ? 18'

- 6.0 o upper side

Experiment(NASA TM 4 0 4 0 . 1 9 8 8 )
0 lower r i d e

- Theory (presenl method.improvedl

- 5.0
-5.0[
-4.0 L.E. -4 0
) Moin e l e m e n t T. E. f l o p
CP -3.0
CP

-2.0

-1.0 -1.0

0.0 0.0

0 .
f # ? & J , O . O W.
0 ,

1.0 1.0 1.0

0 x/L 0.1 0 0.5 X/L 1.0 0 r/L 0.25

Fig. 14 Geometry of a 3-component rectangular wing and pressure distributions at section A-A for
a low and a high wing incidence ag.
23-10

Next, a swept wing with 28" sweep and a constant For more general and more realistic wings with
3-element section, is shown in figure 15, and the taper and part span flaps no proper measurements
computed wing lift coefficients CL are compared were available to the author. So only computed re-
with measurements. Here we found that the agree- sults are shown in fimre 16 for an almost realistic
ment can be improved, if the trailing vortices in the wing landing configuration with full span slat and
lifting surface theory are turned upward by %D. 75% span Fowler flap. The converged results were
(This has been done approximately by just rotating obtained with 7 iterations and they look reasonable.
the induced velocity vectors w* and w** counter- The spanwise lift distribution drops near the flap tip
clockwise by ag/2.) and the pressure distributions show some separation
The computations do not yet include an infinite on the flap and on the outer wing near the flap tip
swept wing correction for the pressure distributions. (station No. 3). Given on the figure are also the total
Nevertheless, for wing incidences up to about 25" wing force coefficients and the computing time re-
the lift of this 3-component swept wing is fairly well quirement.
predicted. For comparison the computed lift for the On an IBM 3090 computer about 4 minutes CPU-
unswept wing is also shown. time were needed for this case, and in the pwious
example (fig. 15) an average of about 2.5 minutes
was needed per wing incidence.

lv- S= 2.531. -4 3 - componen t

F
+
L .

I
Y+-
10 = 3 R O
s w e p t wing-

slat, b s = 16.3'

X
I main element
constant wing section:
3-elem. airfoil SCCH Fowler flap, tif = 1 2 . 5 O

Present method
Pres. method, improved
(trailing vortices turned

0.0
0 100 % 20° 3 Oo

Fig. 15 Geometry of a 3-component swept wing and total lift coefficients from theory and experiment.
 23-1 1

3-component general wing

Landing configuration of a wing with
full-span slat and 7 5 % -span Fowler flap. Chordwise
span: 2slL =9. pressure
distributions:
Onset flow data: ag=lZo
7 stat ion
Re=lO
M,=0.2 -L

CP
-2
Wing geometry : Sponwise
'I -distribution:
station 0
+1
1.00-

0.75-

S
Yls

0.50-
,'II -L

CP
-2

0
0.25- *1

0
I

Total lift-,drag-,and moment-coefficient:

CLz2.25 , CDzQ2L7, C,0=-0.626
CPU-time on CRAY(X-MP216)180 sec.

Fig. 16 Geometry of a 3-component general wing and some results obtained with the present method.

4. CONCLUSIONS
were available to the author for theory-experiment
A relatively fast and simple method for approxi- comparison. The few presented results are encour-
mately predicting the subsonic steady viscous flow aging but not sufficient for thorough validation of
around wings of moderate to high aspect ratio has the method. Further validation is required.
been presented. The total iterative procedure com-
Certainly the method is restricted to wings of mode-
bines a 3-d inviscid lifting surface method with 2-d
rate to high aspect ratio and low to moderate sweep.
viscous flow calculations for the wing sections (air-
For such wings the proposed way of coupling fast
foils in cuwed basic flow fields). Essential high-lift 2-d and 3-d methods seems to by very useful.
flow features, e.g. complex geometry and viscosity Present limitations with respect to wing incidence
effects including separation and also ground effects and Mach number originate mainly in the 2-d meth-
are taken into account. od, presently applied. To extend the scope of the to-
tal method and increase the accuracy, it may be
After successful tests for clean wings at high angles helpful to use a more advanced 2-d multi-element
of attack, the method has been extended and applied airfoil method, containing e.g.
to various multi-component wings. The computa- - refined models for the open wakes (e.g. ref. 21) and
tions converged, the computing time requirements for the closed bubbles,
are modest and the results look reasonable. Very few - allowance for possible wake-boundary layer con-
proper experiments for multi-component wings fluence,
23-12

- more accurate treatment of compressibility near nat. Concil of Aeronautical Sciences, Paper 88-
suction peaks. 4,6.4,1988
Moreover, the total method might .be improved, for 11. Jacob, K.: Advanced Method for Computing
example, by using, the Flow around Wings with Rear Separation
- infinite swept wing corrections for the local pres- and Ground Effect. DFVLR-FB 86-17.1986
sure distributions, 12. Dutt, H.N.V., Jacob, K.: Viscous Subsonic
- a more sophisticated model for the curved basic Flow Computation for Wings with Flaps for
flow, High-Lift. DFVLR-IE3 221-89 A 16,1989
- more careful treatment of the wing regions in the 13. Truckenbmdt, E.: TragflBchentheorie bei in-
vicinity of geometric discontinuities. kompressibler Strtimung. Jahrbuch 1953 der
Wissenschaftl. Gesellschaft fiir Luftfahrt.
Despite of the mentioned restrictions the author is 14. Hummel, D.: Nichtliieare Tragfliigeltheorie in
hoping, that the presented method -or it’s next im- BodennSUle. Zeitschrift fur Flugwissenschaften
proved version-will become a helpfil tool to engi- 21, Heft 12,1973
neers for the design of high-lift wing configurations. 15. Steinbach, D.: Berechnung der StrOmung mit
AblCisung fiir Profile und Profilsysteme in
REFERENCES Bodenniihe oder in geschlossenen Kanaen.
Zeitschrift fiir Flugwissenschaften und
1. Brune, G.W., McMasters, J.H.: Computational Weltraumforschung 2, Heft 5,1978
Aerodynamics Applied to High-Lift Systems. 16. Chen, Y.Q., Jacob, K.: Computation of Flow
Applied Computational Aerodynamics, Vol. around Wings with Rear Separation, Including
125 of Progress in Astronautics and Aeronau- the Curved Basic Flow Concept. DFVLR-IB
tics, AIAA, ISBN 0-930403-69-X, 1990 221-84 A 08,1984
2. Jacob, K., Steinbach, D.: A Method for Predic- 17. Rotta, J.C.: FORTRAN-Rechenprogramm fiir
tion of Lift for Multi-Element Airfoil Systems Grenzschichtenbei kompressiblen ebenen und
with Separation. AGARD CP- 143, April 1974 achsensymmetrischenSVtlmungen. DLR-FB
3. Henderson. M.L.: A Solution to the 2-d Sepa- 71-51, 1971
rated Wake Modeling Problem and its Use to 18. Rajeswari, B., Dutt, H.N.V.: Non-PlanarVortex
Predict c h a x of Arbitrary Airfoil Sections. Lattice Method for Analysis of Complex Mul-
AIAA Paper 78-156, Jan. 1978 tiple Lifting Surfaces. NAL India, TM AE
4. Olsen, L.E.: Optimization of Multi-Element 8606, Aug. 1986
Airfoils for Maximum Lift. NASA CP-2045, 19. Bippes, H., Turk, M., Jacob, K.: Experimen-
Vol. 1, March 1978 telle Untersuchungenzur abgeltisten Strtimung
5. LeBalleur, J.C., Neron M.: Calcul d’coule-
an einem Rechteckflugel. DFVLR-FB 81-12,
ments visqueux decolles sur profil d’ailes par
1981
une approche de couplage. AGARD Cp-29 1,
Aug. 1980 20. Applin, Z.T., Gentry, G.L.: Pressure Distribu-
6. Jacob, K.: Berechnung von Profilsystemen bei tions from Subsonic Tests of an Advanced
UnterschallsVtlmung mit mehrfacher Ablti- Laminar Flow Control Wing with Leading and
Sung. DFVLR-FB 81-24, 1981 Trailing Edge Flaps. NASA TM 4040, parts I
7. Dargel, G., Jakob, H.: Berechnung von Klap- and 11, July 1988
penprofilstrtimungen mit Abltisung auf der Ba- 21. Jacob, K.: Improved Wake Modeling for
sis gekoppelter Potential und Grenzschicht- Flow Around Multi-Element Airfoils.
ltisungen. Strtimungen mit Abltisung. DGLR- DLR-IB 22 1-92 A 06 ,1992
Ber. 88-05, S. 267-278,1988
8. King. D.A., Williams, B.R.: Developments in ACKNOWLEDGEMENTS
Computational Methods for High-Lift Aerody-
namics. Aeronautical Journal, Vol. 92, No. 917, I want to express my thanks to:
Aug./Sept. 1988 my institute for supporting this work, my colleagues
9. Shima, E.: Numerical Analysis of Multiple El- at DLR Dr. Steinbach and Dr. Bippes and our former
ement High Lift Devices by Navier-Stokes guest scientists Mrs. Chen from BUAA Beijing and
Equation Using Implicit TVD Finite Volume Dr. Dutt from NAL Bangalore for much help and
Method. AIAA Paper 88-2574-CP, June 1988 good cooperation, and Mrs. Alberti for carefilly
10. Fritz, W.: Numerical Simulation of 2-d Turbu- preparing the manuscript.
lent Flow Fields with Strong Separation. Inter-
 24-1

CALCULATION OF MULTIELEMENT AIRFOILS AND WINGS AT HIGH LIFT

Tuncer Cebeci*
Aerospace Engineering Department
California State University, Long Beach
Long Beach, California 9 0 8 4 0 , USA

SUMMARY limitation is a function of computers and

A calculation method based on an inter- programming methods and these are likely
active boundary-layer approach to multi- to improve with time so that solutions of
element airfoils and wings is described. the Navier-Stokes equations, with proper
For two-dimensional flows, the method is consideration of momentum conservation in
applied to three types of airfoil configu- two directions together with longitudinal
rations with and without flap wells in diffusion, are likely to be a major compo-
order to demonstrate its applicability and nent of design methods of the future. The
accuracy to general high-lift configura- combination of the largest main-frame com-
tions. This method, extensively tested for puters and structured and unstructured and
single airfoils as a function of shape, multigrid techniques has already been shown
angle of attack, and Reynolds number, is to be ver powerful as, for example, by
shown here to apply equally well to multi- Mavriplisy2, Rogers et al.13, Barth.14 and
element airfoils. The calculation method as we shall see, in the papers to be pre-
is also applied to a wing and wing/flap sented in this meeting.
configuration in order to demonstrate its
promise for addressing three-dimensional Regardless of which approach is used to
flows. Preliminary results indicate that solve the conservation equations, one must
with further development,the method, as for calculate the onset of transition in the
multielement airfoils, will also become a development of the design algorithm so
practical, accurate and efficient tool for that the effects of wind tunnel and flight
multielement wings. Reynolds numbers can be properly identi-
fied. As we shall see later, the compo-
nents of the multielement airfoils at wind
1. INTRODUCTION tunnel Reynolds numbers can have relatively
In recent years, there has been a renewed lower Reynolds numbers than that of the
interest to design more efficient and eas- main airfoil. At chord Reynolds numbers
ier to build high-lift systems. Primary less than 500,000. the components can have
objectives are lower cost and lower drag, large separation bubbles, with the onset
which lead to lower noise. Extensive of transition occurring inside the separa-
measurements have been reported for this tion bubble. As a result, their behavior
goal by Nakayamal, Alemdaroglu, summarized can be significantly different than the
by Nakayamal, and Valarezo et a1.3-4 The main airfoil at higher Reynolds numbers.
data of Nakayama are for a three-element For this reason, an accurate calculation
airfoil with a leading-edge slat and for a procedure not only requires the solution
single-segment flap; these data were of the conservation equations by either
obtained at NASA Langley's Low Turbulence approach, but also requires the prediction
Pressure Tunnel (LTPT). Those of of transition, modelling the transitional
Alemdaroglu are essentially for the same turbulent flow, which is different than the
but smaller model and were obtained at the flow at high Reynolds numbers.
low-speed wind tunnel of California State
University, Long Beach. The data of The approach to be described in this paper
Valarezo et al. were also obtained at NASA is based on the solutions of inviscid and
Langley's LTPT and correspond to measure- boundary-layer equations with the onset of
ments at high Reynolds numbers. These transition computed with the en-method
data add to the previously obtained data based on linear stability theory; this
on multielement airfoils by van den Berg5, method has been tested extensively for
van den Ber and 0skam6, Oskam et a1 7 , single-element airfoils, as described for
Omar et al.8*g and Olson and Orloffli)and example by Cebeci et and by
allow the development and validation of Cebeci.17 These studies have shown that
computer programs to analyze high-lift the flows around a number of airfoil geom-
systems. etries, with angles of attack up to and
beyond that of stall, and including regions
Several airfoil-analysis and design algo- of extensive separated flow, can be repre-
rithms have been developed in the past sented accurately and with low cost for
decade and have been based on one of two both high and low Reynolds numbers.
approaches, either as numerical solutions
of the Reynolds-averaged Navier-Stokes The present paper is concerned with the
equations or as interaction between invis- extension of this method to multielement
cid and boundary-layer equations. l1 The airfoils and wings with and without flap
former approach involves the numerical wells. The following section, Section 2 ,
solution of elliptic equations so that describes the calculation method for two-
information travels in all directions dimensional flows to obtain the results of
through pressure, velocity and viscous and Section 4. The calculation method for
turbulent stress gradients. As a result, three-dimensional flows i s described in
the solution method requires simultaneous Section 3 with results presented in Section
calculation of the pressure, velocity 5. The paper ends with a summary of the
components, and stress tensor throughout main conclusions and a statement of the
the flowfield and this, in turn, implies a further steps t o be taken toward the pro-
trade-off between accuracy and cost. This vision of a general method for calculating
*Professor and Chairman
24-2
e
multielement airfaile ana wingt, at high
lift. (6*)'+l = [l + o (-'ev
(6*)' - 1)l ('6)
'ei
2. CALCULATION METHOD: TWO-DIMENSIONAL
FLOWS is used in t h e inverse method to replace
The calculation method for two-dimensional the Hilbert integral formulation of the
flows uses the panel method of Hess and external boundary condition. The new edge
Smithla and a solution of the boundary- boundary conditions are given by Eq. (3b)
layer equations in which the turbulence and Eq. (6). where Uey and uei correspond
model is given by the algebraic eddy vis- to the external velocities computed by the
cosity formulation of Cebeci and boundary layer and inviscid methods,
Smith.16Em&th b denoting 1 + Em/V. the respectively. and o is a relaxation
continuity and momentum equations can be parameter. At the end of the flap-well
written as region, the solution procedure reverts to
the Hilbert-integral approach.

2.2 Turbulence Model

The turbulence model used to represent the
flow on the airfoil is expressed in terms
of the Cebeci and Smith eddy-viscosity
formulation,
In the absence of mass transfer, the
boundary conditions on the airfoil are:

u = v = o , y = o (3a)

U + Ue(X), Y + - (3b)
and in the wake, where a dividing line at
y = 0 is required to separate the upper and
lower parts of the inviscid flow, and in
the absence of the normal pressure gradi- 1
ent, they are:

y + U + Ue(X); y = 0, v = 0 (4) where a is a constant assumed to be 0.0168

and
2.1 Interaction Law
To perfoLm the calculations for flows with
separation, an inverse procedure is used
and the external velocity is computed as (Ea)
part of the solution. According to the 1
formulation discussed in Ref. 15, the edge Y =
boundary condition is written as 1 + 5.5(y/6) 6

The condition used to define yc is the

continuity of the eddy viscosity so that
where 6Ue(x) is computed from the Hilbert Eq. (7a) is applied from the wall outward
integral (inner region) until its value is equal to
that given for the outer region by Eq.
da (7b). The expression Ytr represents the
e x - a (5b) transition region and is given by

This inverse boundary-layer formulation is

appropriate to airfoils and to those parts
of airfoils without surface discontinu-
ities such as flap wells. Where flap Here xtr is the location of the beginning
wells occur, a different formulation of of transition and G is defined by
the inverse procedure is required, and
that used here is described below. ..3
The calculation of the flow in the flap-
well region is similar to that over a
backward-facing step. A large portion of where C is 6 0 for attached flows and the
the flow separates immediately after the transition Reynolds number Rxtr= ( U ~ X / V ) ~ ~ .
sudden change of the geometry, and the size
of the reversed-flow region depends mainly In the flap-well region, the above formulas
o n the step height, o n the gap, and the are modified so that
overhang. The flow reattaches and gradu-
ally recovers downstream in the flap-well
region or in the wake. The calculation of
flows of this kind is difficult, and poten-
tial theory is not adequate because of the
singularity that occurs at the geometry
discontinuity and the strong viscous
Here ELI)denotes the eddy viscosity cor-
responding to the velocity rofile above
effects in the separated flowfield. Thus, the separated region, and t i F ) includes
a n initial distribution of displacement the total re ion from the wall. The expres-
thickness is assumed, and the relaxation
formula
sions for EA 4
and cbF) are given by:
 24-3

are obtained for given velocity profiles U

and f i l l and for a set of specified dimen-
sional frequencies U * . Amplification rates
Y > Yo ai are then computed as a function of x,
and for each frequency the value of n is
obtained from the integral
I
a X
I (1 - ")ay n = - I aidx
Y O
ue xO

Transition is obtained from the resulting

curves corresponding to the maximum ampli-
fication factor by assuming a value of n,
commonly taken to be between 8 and 10.

-y/A 2 2.4 Solution Procedure

(0.4Y [l - e 11 A brief description of the solution pro-
cedure used in this method is ds follows.
The panel method provides an external vel-
for y > 0 ocity distribution based on a body shape
in which the flap-well region is assumed
to be absent and identifies the stagnation
point. The interactive boundary-layer
approach leads to solutions on the upper
where yo is the location of U = 0. h is a Surface from the stagnation point through
relaxation parameter (usual y around lo), L the regions of laminar, transitional and
is a characteristic length, and xo is the turbulent flow to the trailing edge.
beginning of the flap-well. Similarly, it provides results for the
lower surface up to the beginning of the
The corresponding expressions in the wake flap-well. A displacement thickness dis-
are: tribution, 6*(x), is assumed in the flap
well and, with the continuation method
cm = (Em)W + described in Ref. 2 4 and with the initial
velocity profile similar to that of a
(x - x o ) backward-facing step, calculations proceed
- ('m)w' exp 206 1 (11) to the trailing edge. With the upper and
lower surface velocity profiles computed at
where (cm)t.e. is the eddy viscosity at the trailing edge, the calculations are
the trailing edge computed from Eqs. ( 4 . 7 ) extended into the wake. As a consequence
and (4.10) and ( E ~ )is~ the eddy viscosity of the above, a blowing velocity is avail-
in the far wake given by the larger of able on the airfoil and in the wake. In
the flap-well region, the blowing velocity
Vn is defined by

and
W where 6; = 6* - bt. Here 6t corresponds to
( c U~ =) 0.064
~ I (Ue - U)dy (12b) the body shape assumed to exist over the
flap-well. Elsewhere, the blowing velocity
Ymin is given by
with ymin denoting the location where the vn = (ue6*) (17)
velocity is a minimum.
2.3 Transition Method With the blowing velocity distribution
For two-dimensional high Reynolds number known, a new distribution of external vel-
flows, the onset of transition can con- ocity Uei(x) is obtained from the panel
veniently be calculated from the formula method. As before, the boundary-layer
given by MichelZ0 solutions on the upper and lower surfaces
of the airfoil are obtained with the
Re = 1.174 (1 + (13) Hilbert integral. In the flap-well, with
U known from the previous flap-well calc-
ufation, a new &*-distribution is available
where Re = Uee/V and RX = UeX/V. For low from Eq. (6). which is used to obtain solu-
Reynolds number flows, Eq. (13) is not tions up to the trailing edge. This
appropriate and it is necessary to use the sequence of calculations is repeated for
en-method. This method first suggested the whole flowfield until convergence is
by Smith and GamberoniZi and Van IngenZ2, achieved. Additional details are provided
is based on the linear stability theory and in Ref. 23.
can also be used for high Reynolds number
flows as well as three-dimensional flows, 3. CALCULATION METHOD: THREE-DIMENSIONAL
as we shall discuss later in the paper. WINGS
According to this method. for two- The extension of the method of the previous
dimensional flows the solutions of the section to three-dimensional flows has the
Orr-Sommerfeld equation, ingredients shown in Fig. 1. The purpose
of the interface program placed between the
inviscid and three-dimensional inverse
boundary-layer methods is to process the
geometry and inviscid velocity data for
24-4

for a prescribed flow condition is, there-

fore, obtained by simultaneously satisfying
a condition of zero normal velocity at a
control point on each panel of the body
together with a Kutta condition at each
I trailing-edge panel.
The nature of the Kutta condition adopted
by the many panel methods that are cur-
rently available varies greatly. While the
Fig. 1. The interactive boundary-layer condition adopted by Hess assumed equal
method. upper and lower surface pressures at the
trailing edge, other methods make use of
input to the boundary-layer program. The other derived conditions which do not
basic input to this program is the defini- guarantee a pressure match at the trailing
tion of the wing configuration which is edge. Since we are also interested in the
used by a geometry subroutine to construct computation of inviscid flows with viscous
a nonorthogonal coordinate system and com- effects, and since the behavior of the
pute the associated geometrical parameters, boundary-layer at the trailing edge can
which consist of geodesic curvatures and have a significant effect o n the overall
metric coefficients needed in the boundary- solution, the approach adopted here is
layer calculations. Some of the generated more realistic and blends in well with the
data are used later in a velocity sub- previous studies on two-dimensional flows
routine to determine the inviscid velocity employing interaction procedures.
components at the boundary-layer grid
points and to transform the inviscid vel- 3.2 Inverse Boundary-Layer Method
ocity components on the surface, calculated Viscous flow calculations are performed by
in a Cartesian coordinate system. into the solving the three-dimensional boundary-
boundary-layer coordinate system. This layer equations in the inverse form. These
operation consists of dot products of vel- equations for a nonorthogonal coordinate
ocity vectors as well as chordwise and system with the eddy-viscosity concept
spanwise interpolation. Further velocity are25
and geometry data processing is done in a -
a (uh2 sine) + a (whl sine)
subroutine which separates the generated ax
information into upper and lower surfaces a (vh h
of the wing for boundary-layer calculations + -
ay 1 2
sine) = 0 (18)
3.1 Inviscid Method
The inviscid flow is computed by the Hess
surface-source panel method, which is
applicable to a complete airplane configu-
ration. In this method a general body is - K~U’ cote + K ~ W ’ cosece + K ~ ~ U W
represented by means of a set of quadri-
lateral panels as shown in Fig. 2 . A
three-dimensional configuration in general I -
cosec 2 0 aJ cot0 cosec0 aJ
consists of lifting sections, such as a phl ax + ph2 az
wing o r pylon, for which there is a well-
defined trailing edge and nonlifting sec-
tions such as fuselage. Under the Hess
formulation, all panels are assigned a n
independent source distribution, while
those on a lifting section are assumed to
carry a bound vorticity distribution. The
variation of this bound vorticity in the - KZw 2 cot0 + K1u 2 cosec0 + K uw
streamwise direction is assumed, while its 21
variation in the spanwise direction is
adjusted to satisfy the Kutta condition at
the trailing edge. The complete solution
- cot0 cosec0 aJ cosec 2 0 a~
phl ax - Ph2 az

+ U
a
- aw
( b -)
ay ay
where x denotes the axial direction. z the
COMTAOC spanwise direction, and y is normal to the
surface. Here h denotes the metric coef-
ficients. K the curvature parameters, and
0 the angle between the coordinate lines
x t’const. and z = const.
The above equations, together with those
for the Plane of symmetry and along the
chordwise direction, are solved by Keller’s
two-point finite-difference method (box
scheme) subject to the following boundary
conditions

Fig. 2. Wing/body configuration.

 24-5

To account for flow separation, as in two-

dimensional flows, a form of the interac-
tion law of Veldmana6 is used by modifying where y is the angle the wave makes with
the two-dimensional interaction formula the x-axis. With a and B connected through
given by Eq. ( 5 ) to account for the inter- Eq. (26). and with the disturbance propa-
action in the x- and z-directions in three gating along the way given by the two terms
steps. In step one, an initial displace- on the right-hand side of Eq. (26). the
ment surface is generated by solving a disturbances are damped if the amplifica-
quasi-three-dimensional form of Eqs. (18) tion rate defined by
to (20) with all derivatives with respect
to z neglected, that is, r = ai - si (-1aa
as U.R
a (uh2sine) + -
- a (vhlh2 sine) = 0 (22) is > 0 , neutral if r = 0 and amplified when
ax ay r < 0 . Once a and f3 are computed with the
au
U ax + v
hl & - K1u 2cote + K2w2cosece constraints of Eq. (26). the amplification
rate r is obtained from Eq. (27); addi-
tional calculations are then made for dif-
+ K u W = -
cosec2e a a au ferent values of aa/as s o that new values
1 phl ax + (b of a and t? are calculated to determine the
(23) maximum value of r.
2 aw
ay
2
+ v - - K ~ cote
W
2
+ K ~ Ucosece This method has been evaluated in terms of
measurements reported for the flow around
a swept wing equipped with a cambered
leading edge and attached to a half fuse-
lage and for the flow around a prolate
with the external velocity distribution spheroid at 10 degree incidence.30 It
has been shown that it is convenient to
ug(x) obtained from the panel method. The use, particularly because of the neutral
second step involves interaction between stability curves (zarfs) which facilitate
the inviscid flow equations and the quasi- the calculation and avoid uncertainties
three-dimensional flow equations. As in associated with the choice of magnitude
two-dimensional flows, the solutions of the and location of the critical frequencies.
boundary-layer equations are used to com- In general, the calculated values of the
pute distributions of blowing velocity on onset of transition are in very good agree-
the surface and these allow the inviscid ment with measured values. 30
flow solutions to be updated. In step
three, after the calculation of the initial 3.4 Solution Procedure
conditions in the (y,z) and (x,y) planes, While the calculation of the onset of tran-
the fully three-dimensional boundary-layer sition is important for airfoils, it is of
equations are solved with the external utmost importance for three-dimensional
velocity components resulting from step flows. Unlike the two-dimensional flows
two. As before, the spanwise velocity
component is assumed to correspond to its where transition occurs in the region of
inviscid value. The viscous flow solutions flow deceleration, in three-dimensional
are obtained by marching in the spanwise flows it often occurs in the region where
direction at each advancing chordwise loca- the flow accelerates. The sweep angle and
tion. This represents the first phase in Reynolds number strongly influence the
an interactive loop which involves the location of transition and requires that
fully three-dimensional boundary-layer its calculation becomes a part of the comp-
equations. In the subsequent phases, as utational strategy.
before, the blowing velocity distribution In the present method, the stability/
is used to obtain improved inviscid flow transition calculations are first performed
solutions so that the fully three- for three-dimensional laminar boundary
dimensional boundary-layer equations can layers obtained for a prescribed pressure
be solved in accordance with the iteration distribution, so that the inviscid and
scheme shown in Fig. 1.
viscous flow equations can then be solved
3.3 Transition Method according to the interaction scheme shown
We again use the en-method and solve the in Fig. 1. The quasi-three-dimensional
Orr-Sommerfeld equation, which for three- forms of the equations are solved chordwise
dimensional flows is given by in the inverse mode to obtain the inter-
action coefficients needed in the solution
aiV - 2(a2 + e2)+ + (a2 + s212+ of the full three-dimensional boundary-
layer equations. Several sweeps on the
- iR(aii + t3ij - a)[+" - (a2 + s2)+1 wing and in the wake are performed and new
+ iR(aG" + B i j l ' ) + = 0 (25) inviscid flow solutions with viscous
effects are obtained. This procedure is
Here a and $ denote the dimensionless wave repeated until the convergence of the solu-
numbers in the x- and z-directions and o tions; the transition calculations are then
the radian frequency. Our eigenvalue pro- repeated to obtain new transition locations
cedure differs from those used by Malik27 on the wing for the next phase of the
and Mack2*. It is based on the saddle- interactive boundary-layer calculations.
point method of Cebeci and S t e w a r t s ~ n ~ ~ , The whole process is repeated until the
which, unlike the approaches of Malik and flowfield and transition locations
Mack, does not assume a relationship converge.
between the two wave numbers a and f3 but 4. RESULTS AND DISCUSSION:
computes it from the requirement that aa/aB TWO-DIMENSIONAL FLOWS
is real. According to this requirement, We now use the interactive boundary-layer
the wave orientation and growth direction procedure of Section 2 to calculate the
of the disturbance are given by
24-6

performance characteristics of single represented by

airfoils without (Section.4.1) and with 6
(Section 4.2) flap wells, two-element 1 + 2RT(2 - RT)
airfoils (Section 4.3), and three-element
airfoils corresponding to slat, airfoil,
and flap combinations (Section 4.4). For for RT < 1.0. For RT 2 1.0, B is taken as
single airfoils, we also consider flows at
low Reynolds numbers since these flows
become important on the components of
high-lift systems at wind tunnel Reynolds
numbers. Introducing the above relationships into
the definition of F and using Eq. (27a).
4.1 Single Airfoil Without Flap Well the following expression for a is
In calculating airfoil flows at high obtained
Reynolds numbers with the interactive
method of Section 2, the onset of tran-
sition is obtained from Eq. (13) with the
transition region represented by the
ytr-term in Eq. (ab), except where the where B is given by Eqs. (27d) and (27e).
boundary layer separates upstream of this
location, in which case transition is Figure 3 shows a comparison between calcu-
assumed at the separation location. Stud- lated and measured lift coefficients for
ies indicate that. while the turbulence the NACA 0012 airfoil at two chord Reynolds i
model of Section 2.2 is adequate for almost numbers. As can be seen, in accord with
all airfoil flows at low and moderate the measurement^^^, the calculation
angles of attack, it is necessary to make method is able to compute ca for all angles
improvements to this model at high angles of attack and satisfactorily account for
of attack corresponding to near stall and the effects of Reynolds number. Figure 4
post-stall conditions. One approach dis- shows similar comparisons for the NACA
cussed and developed by Johnson and KingJ1 23012 airfoil at two chord Reynolds num-
and Johnson and C ~ a k l e yis~ ~to adopt a bers, indicating again good agreement with
nonequilibrium eddy-viscosity formulation measurements. Additional comparisons and
E,, in which the C S model serves as a n equi- discussion are provided in Ref. 33.
librium eddy viscosity ( E ~ distribution.
) ~ ~
An ordinary differential equation (ODE), The calculation of airfoils at low Reynolds
derived from the turbulence kinetic energy numbers requires changes to the turbulence
equation, is used to describe the stream- model of Section 2.2 and to the procedure
wise developmentofthe maximum Reynolds used to obtain the onset of transition
shear stress -(pU Vm), O r -U Vm for short, location. According to the ytr model used
in conjunction with a n assumed edd%
viscosity distribution which has -U vm as
its velocity scale. In the outer part of
the boundary layer, the eddy viscosity is
treated as a free parameter that is
adjusted to satisfy the ODE for the maxi-
mum Reynolds shear stress. In essence,
this model treats the parameter a in Eq.
(7b) as a variable and determines it as
described above.
Another approach to improve the predictions
of the C S model in flows with adverse pres-
sure gradient and ~ e p a r a t i o n ~ is
~ . to
relate the parameter a to a parameter F by
a = -0.0168
F2. 5

Here F denotes the ratio of the product of

the turbulence energy by normal stresses
to that by shear stress evaluated at the
location where shear stress is maximum,
that is

Before Eq. (27a) can be used in Eq.

9):
a n a d d i t i o n a l e l a t i o n s h i p between (U
v) and (-ulvl) at (-U'V')max is needed.
This is done by assuming that the ratio in
Eq. (27b)
- -
12 - V' 2
B = f U - 1 - (27c) 0 5 10 15 20
alpha
1s 30 I5

-U'V' (-U'V')max
Fig. 3. Effect of Reynolds number on the
is a function of RT = -tW/(-U1V') which, lift coefficient of the NACA 0012 airfoil;
according to the data of Nakayam:Qg is (a) Rc = 6 x IO6, and (b) Rc = 8.8 x lo6.
 24-1

-
Lm
/W 65-213. RC Zb 1 1@
% LNVlO9A. R. 3.75 d
10' 20 cl13

"
.. .
'.
.. :
a Pig. 5. Variation of C2/3 with transition
0 '
0 5 10 15
.lDh.
20 25 30 15
Reynolds number R, ,...
Care should be taken in the use of this
equation outside the range of experiments,
particularly at high Reynolds numbers where
the separation bubble is likely to be small
and a limiting value of C = 60 applies.

The application of Eq. (30) requires the

location of the onset of transition and
this is obtained with the en-method
discussed in Section 2.3. The manner in
which it is obtained is described in Ref.
17.
In the studies reported in Ref. 17, the
accuracy of the interactive method was
evaluated for a number of airfoils and
0 I LO 15 20 25 10 15
overall good agreement with experiment was
alpha observed. W e now consider the Eppler
airfoil and present a sample. of results
Ffg. 4.. Effect of Reynolds number on the for a chord Reynolds number of 300,000 for
lift coefficient of the NACA 23012 airfoil; a wide range of angles of attack and
(a) Rc = 3 x lo6, and (b) Rc = 8.8 x lo6. compare the predictions with the experi-
mental data of McGhee et al.35 obtained
in the C S model and given by Eqs. (8b) and in the Langley Low-Turbulence Pressure
(8c). the extent of the transition region Tunnel (LTPT). The tests were conducted
Rfix is related to the transition Reynolds over a Mach number range from 0.03 to 0.13
number Rxtr by and a chord Reynolds number range from 60
x lo3 to 460 x lo3. Lift and pitching-
2/3 moment data were obtained from airfoil
RAx = CRXtr surface pressure measurements and drag data
from wake surveys. Oil flow visualization
was used to determine laminar-separation
which shows that R A increases
~ with and turbulent-reattachment locations.
decreasing Rxt . This expression was
obtained from 6ata based on attached flows Figure 6a shows a comparison between the
and is not applicable to flows with sepa- calculated and measured distributions of
ration. Experiments show that the extent pressure coefficients for angles of attack
of a separation bubble and the location of of 0 and 8 deg. The calculated lift and
transition depend upon the Reynolds number. drag coefficients shown in Fig. 6 b indi-
At high Reynolds numbers, transition cate remarkably good agreement with experi-
usually corresponds to the location of ment for all angles of attack up to stall.
separation and the length of the bubble is There the computed lift coefficients begin
relatively short. At low Reynolds numbers, to deviate from data, indicating higher
transition can occur inside the bubble and values than those measured; the discrepancy
can strongly influence the nature of flow. increases with a n increase in Reynolds
To take account of the corresponding number, and the solutions do not follow
effects, a correlation formula was devised the poststall behavior. W e believe the
by Cebeci17 to represent C in Eq. (8c) reason for this discrepancy lies in the
in terms of Rx , and based on experimental turbulence model. Whereas the results
data which shobrseparation-induced transi- presented in Figures 3 and 4 made use of
tion at low Reynolds numbers. This corre- the C S model with the parameter a com-
lation formula is shown in Fig. 5 together puted from Eq. (28). those in Fig. 6 were
with the experimental data obtained for made with a = 0.0168.
several airfoils. The data, which encom-
passes a typical low Reynolds number range The calculated values of the chordwise
from R~ = 2.4 x 105 to R, = 2 x 106, falls location of laminar separation (LS), turbu-
conveniently on a straight line on a semi- lent reattachment (TR), and the onset of
log scale and can be represented by the transition are given in Table 1 for several
equation angles of attack. The experimental results
C2 = 213 [log R - 4.73231 (30) of this table are subject to some uncer-
'tr tainty because of difficulties associated
24-8

1 .o{ I
0 0.2 0.4 0.6 0.8 1.0
(a)

Io o 0, oa 0, 01 01 01 ar am
I
om
x/e

Fig. 7. LB 572 single airfoil with flap-

well cut.
0 0.i G.4 0.6 0.8 1.0

c,::;J

0.6

0.2 0.2 L
0.01 0.02 a.03 0.00 0
0
0. 4- 8' 12'
cd

Fig. 6. Comparison of calculated (solid

lines) and measured (symbols) pressure-
coefficient distributions at: (a) a = 0
deg, (b) a = 8 deg,.and (c) lift and drag
-U c
coefficients for R, = 3 x 105.

Table 1. Experimental and calculated

laminar separation (LS), and turbulent
reattachment (TR), and transition locations
on the upper surface of the Eppler airfoil.

Experiment
a
(;Ix Calculated
-
TR -tr -
LS -
TR
0 0.48 0.69 0.63 0.51 0.72
2 0.45 0.62 0.58 0.46 0.67
4 0.40 0.58 0.52 0.43 0.60
5 0.39 0.55 0.49 0.415 0.57
6 0.38 0.50 0.43 0.42 0.50
6.5 0.38 0.44 0.40 0.41 0.46

with the surface visualization technique.

With this proviso, comparison between
measured and calculated values must be
considered outstanding. It should be
noted that when there is a separation
bubble, the transition location obtained
from the en method occurs within the
bubble in all cases and in accord with
experimental observation, leads t o reat-
tachment some distance downstream. Fig. 8. LB 572 airfoil pressure distribu-
4.3 Single Airfoil with Flap Well tion for R, = . 0 . 5 x io6 at (a) a = 5.0.
Figure 7 shows the geometry of the LB 572 (b) a = 8.0, (c) a = 12.0.
single airfoil with flap well tested in the
low-speed wind tunnel of the California comparison between calculation and the
State University, Long Beach (CSULB). The experimental data for the Reynolds number
pressure distributions of Fig. 8 allow of 0.5 x lo6 and transition locations at
 24-9

25% of the chord on both surfaces. In downstream, close to the trailing edge. the
general, the results are in good agreement computed profiles have larger negative
for angles of attack up to 12O. The upper velocities near the wall than measured
surface peak, registered by the calcula- profiles, which may mean that a too weak
tions close to the leading edge, is in part mixing model was applied to the reattach-
due to the better spatial resolution of the ing flow. Figure 10 shows that the skin-
calculation method. Close to the trailing friction distribution vanishes in the
edge. there is some disagreement between plane of the step and becomes negative,
the upper surface pressure distributions corresponding to the reverse flow up to
that may stem from approximations made x/c of around 0.07. The location of
about the flap well region and the way the reattachment in the flap well region was
Kutta condition was specified in the panel observed to move upstream from the
method. However, the pressure coefficient trailing edge with increasing angle of
curves well represent the flow behavior in attack.
the flap-well region: the constant pressure
values indicate a recirculation area fol-
lowed by a rapid increase of pressure close
to the trailing edge which signifies reat-
tachment of the separated flow.
Velocity profiles corresponding to angle
of attack of 5 O and the location of the
displacement thickness are presented in
Fig. 9 together with the recirculation
streamlines in the flap well region. Fig-
ure 9a compares the velocity profiles of
the present calculation with experimental
data of Alemdaroglu2 at six locations in
the flap-well cut. The initial profiles
at the step are in good agreement, but
comparison of the downstream profiles shows
that computation predicts a less retarded Fig. 10. Calculated local skin-friction
flow than is indicated by measurement. The coefficient in the flap-well region for
possible explanation is that the eddy- a = 5O.
viscosity formulation used models for a
stronger mixing than is appropriate for 4.3 Two-Element Airfoils
the separated flow after a step. Further In a previous study, a similar interactive
approach was used to compute the perform-
ance characteristics of three two-element
airfoils.36 The inviscid flow solutions
were obtained by the conformal-mapping
method of H a l ~ e y ~rather
~, than the panel
method used here, and the viscous-flow
calculations were performed without
accounting for the wake effects, either
behind the main airfoil or the flap. The
calculation method provided results which
agreed with experimental information within
the accuracy of the measurements up to an
angle of attack that was sufficiently small
so that there was either no separation or
very small separation on the airfoil. and
the gaps between the elements were compara-
tively large. In this way, the difficul-
ties in computing the wake of each airfoil
and accounting for the merging of the
shear layers between the airfoil and the
flap, and extending the range of the
computational method to higher angles of
attack, were postponed to a later time.
In the studies reported here, we first
performed calculations with the present
method, which did not include the wake
effect and compared the results with those
obtained with the earlier code36 with its
different inviscid flow method. After
ensuring that the results of both codes
were essentially the same, the wake effects
were introduced into the present method and
calculations were repeated for the three
two-element airfoils to investigate the
role of the wake effect on the solutions.
Fig. 9 . LB 572 airfoil for Rc = 0.5 x lo6 The first two-element airfoil corres onded
at a = 5.0. (a) Velocity profiles in the to that investigated by Van den BergE and
flap-well region, (b) Recirculation stream- discussed in subsequent papers by Van den
lines and the location of the displacement Berg and Oskam6 and Oskam, et al.7 It
thickness. comprised a supercritical main airfoil
(NLR 730) with a flap of 32% of the main
24-10

chord at a deflection angle of 20 degrees,

close to the highest value that could be
used without onset of flow separation.
This configuration was studied at a
Reynolds number of 2.51 x lo6 for a
range of angles of attack up to 13.1O with
consideration of the wake of the main
element only.
Figure 11 allows a comparison between the
calculated and experimental results of the
NLR 7301 airfoil. Figures lla and b show
the surface pressure distributions at a
= 6 O and 13.1O. respectively, and Fig. 1lc
shows the effect of wake on the computed
lift coefficients. The calculated and
experimental values of local skin-friction
coefficient cf and dimensionless momentum
thickness B/c for the two angles of
attack, Figures lld and lle, show that the
boundary layer on the upper surface is
approaching separation but has not reached
it. Consistent with the studies on single
airfoils, the wake effect is relatively
small due to the absence of flow separa-
tion but has improved the agreement with
experiment.
The second case involves a NASA super-
critical airfoil, 24 in. in length, with a
7 in. flap at a deflection angle of 20
,/
degrees. The experiments were carried out
in the 36 x 96 in. wind tunnel of the
Boeing Research Laboratories at a Mach
number of 0.2 and have been documented by
Omar et a1.8e9 Figures 12a and 12b show
the surface-pressure distributions for
angles of attack of O o and 8.93O. respect-
ively. As can be seen, the inclusion of
the wake in the calculations makes a slight
improvement on the main airfoil but not on
the flap where the results obtained without
the wake effect are better. The results
in Fig. 12c show that the calculated lift
coefficients with the wake are in better
agreement with data than those without the
wake effect.
4.4 Three-Element Airfoils
With the positive results of Figs. 8 to 12
for an airfoil with a flap-well and for
two two-element airfoils, it is appropriate
to consider the application of the inter-
active boundary-layer method to a three-
element airfoil with a flap well. The
chosen configuration is shown on Fig. 13
and corresponds to the high-lift model
tested in the NASA Langley wind tunnel at
a Reynolds number of 5 x lo6. The slat
deflection angle was -30 degrees and the
flap deflection angles 15 and 30 degrees
with angles of attack of 4 to 20 degrees.
The measurements were made by a combination
of hot-wire and laser-velocimetry tech-
niques, the latter was primarily used in
regions of separated flow.
Calculations were initially made on smooth
bodies without explicitly considering the (e) I ,..,'.
U .I ,., .
U ". .
U
.
.I
.
.I
.
.e
.
et L.

flow in the flap-well region. Also, U

because the potential flow theory predicts Fig. 11. NLR 7301 wing with flap. Calcu-
flow singularities at the discontinuity of lated and measured: (a) Pressure distribu-
the airfoil geometry, the sharp corner of tion at a = 6O. (b) Pressure distribution
the slat and the flap-well cut out of the at a = 13.1O. (c) Lift coefficients, (d)
main airfoil were smoothed to prevent solu- Local skin-friction coefficients, cf, and
tions from breaking. Figure 13 shows the momentum thicknesses, B/c. on the upper
modified geometry of this airfoil with the wing surfaces at a = 6O. (e) Local skin-
flap-well fairing and the rounded slat friction coefficients, cf. and momentum
used in calculations. The so-called thicknesses. B/c. on the upper wing sur-
"experimental fairing" refers to the faces at a = 13.1O.
 24-11

Finally, the method of Section 4.2 was

applied to include the calculation inside
the flap-well. The results, shown in Fig.
14, agree well with measurements for all
the cases indicated above and including
angles of attack u p to 20°. This confirms
that it is unnecessary to make a priori
assumptions about the fairing shape, and
allows for further detailed investigations
of the recirculation flow in the flap-well,
such as the gap and overhang effects. Fig-
ure 15 shows the variation of lift coef-
ficient with angle of attack, confirming
that the present calculation method leads
to values which are in close agreement with
experiment.
Comparison of pressure coefficients for a
flap deflection angle of 30°, Fig. 16,
allows similar conclusions to be drawn to
those of the previous paragraph. Also,
the calculated lift coefficients shown in
Fig. 17 are very close to measurements.
5. RESULTS AND DISCUSSION: THREE-
DIMENSIONAL FLOWS
The calculation method described in Section
3 for three-dimensional high-lift configu-
rations is in the evaluation stage. So far
the studies are limited to wing and wing-
flap geometries. The experimental data
being used to evaluate the method is due
to L 0 v e 1 1 ~who
~ has reported lift, drag
and pitching moment coefficients over an
extensive range of configurations of the
high-lift system on a wing of basic aspect
ratio 8.35 and with a trailing-edge plan-
form extension a,nd a body added. In
Section 5.1 w e first present the results
for the wing alone and in the subsequent
section we present results for a wing/flap
configuration.
5.1 wing Alone
The wing tested in Ref. 38 has a n airfoil
section having a considerable rear loading
with the maximum thickness of 10.7% occur-
ring at 37.5% chord and the maximum camber
Fig. 12. NASA supercritical two-element occurring at 75% chord. It has no twist
airfoil. Calculated and measures, (a) nor dihedral, has a quarter-chord sweep
pressure distributions at a = Oo, (b) at angle of 28O and taper ratio of 1/4. The
a = 8.930, (c) lift coefficients. experiments were conducted at a test
Reynolds number of 1.35 x lo6.
Figure 18 shows a comparison between the
calculated and'measured lift coefficients.
As can be seen, there is some ViSCOUS
effect even at very small angles of attack
which becomes much more pronounced at
higher angles of attack. The interactive
U boundary-layer method solutions, obtained
either by using the strip theory approxi-
mation or by the full three-dimensional
Fig. 13. The three-element airfoil with approach, improve the solutions consider-
analytical and experimental fairing. 5 , = ably. In the former case, the calculated
-3OO. 5f = 15O. results agree reasonably well with data at
low and moderate angles of attack, but with
dividing streamline, which was determined increasing incidence angle they begin to
from measurements, while the "analytical deviate from the measured values of lift
fairing" was drawn arbitrarily. Reference coefficient. In the latter case, the
23 presents a comparison between the mea- three-dimensional flow solutions w i t h , t h e
sured and calculated surface pressure dis- blowing velocity vn obtained from
tributions on the slat, main airfoil, and
flap at three angles of attack ( 4 O . 12O and
16O) for the configuration with the experi-
mental fairing, and the corresponding dis-
tributions at the same angles of attack for
the configuration with analytical fairing.
Overall, the calculated results agree well
with experimental data. and the displacement surface A
24- 12

Lf -
Fig. 14. Calculated and measured pressure distributions on the three-element airfoil for
15O and R, = 5 x lo6 at (a) a = 4O. (b) a = 12O. (c) a = 16O. (d) a = 2 0 ° .

Fig. 16. Calculated and measured pressure

Fig. 15. Variation of the lift coefficient distributions on the three-element airfoil
with angle of attack for the three-element with Lf = 3 0 ° and R, = 5 x lo6, (a) a =
airfoil with 6f = 15O and R, = 5 x lo6. 5 O . (b) a = 15O. (c) a = leo, (d) a = 2 0 ° .
 24-13

A47 --- I N V.
. I S.C
..ID
.
- 2D
_. INTERACTIVE
. -
0 DATA
-.-3D INTERACTIVE
I
0 5 10 15
a

Fig. 18. Comparison of calculated and

measured lift coefficients for the
experimental wing data reported in Ref. 38.
a [uehZ sin0(A - 6;)J

0
+ a [wehlsine(A - 6*)] = 0
2 (32)

used to satisfy the Kutta condition in the

inviscid method of section 3.1, provide
very good agreement with experimental data
up to the stall angle. As in airfoil
flows without improvements to the turbu-
lence model, the calculated results begin
to deviate from the measured values and
continue to increase past the stall angle.
Work is in progress towards generalizing
the improved turbulence model for three-
dimensional flows.
5.2 Wing/Flap Configuration
Figure 19 shows the results for a wing/flap
configuration corresponding to a flap
deflection angle of 2 5 O . The viscous flow
8 calculations were first performed with
transition specified at 5% from the lead-
ing edge of the wing and flap. After the
convergence of the solutions, the stability
properties of the velocity profiles were
analyzed by using the 3D version of the
en-method for the wing and the 2D
version of the en-method for the flap.
The reason for the use of the 2D version
of the stability/transition method is due
XR to the low Reynolds number flow on the
Fig. 16. Continued.

"I

SI
CL
0 (cQ)vls
U .

d 1.5 0 1.31
40 1.77 1.63
/
. ...

:I,/.
5 O1.85 1.11
0 6' 1.93 1.78
lo 2.01 1.85
1.93

U . . , . . 1 .o
0 5 a 10
U
U U I) U a, .a U a. )U nb
wwmmam
Fig. 17. Variation of the lift coefficient Fig. 19. Comparison of calculated and
with angle of attack for the three-element measured lift coefficients for the wing/
airfoil with 6f = 3 0 ° and R, = 5 x lo6. flap configuration.
24-14

flap. While the Reynolds number varied 5. Van den Berg, B., "Boundary-Layer Mea-
from 0.66 x lo6 to 1.75 x LO6 on the wing, surements on a Two-Dimensional Wing
it varied from 0.26 x 106 to 0.68 x 106 on with Flap", NLR TR 79009U. 1979.
the flap. A previous s t ~ d yhad ~ ~shown
that crossflow effects at relatively low 6. Van den Berg, B. and Oskam, B.,
Reynolds numbers (Rc < 0.5 x LO6) are small "Boundary-Layer Measurements on a Two-
and transition calculations performed with Dimensional Wing with Flap and a Com-
either the 3 D or 2 D versions of the parison with Calculationsii,AGARD
en-method essentially yield the same CP-271, 1979, Paper 18.
results.
7. Oskam, B., Hahn, D.J. and Volkers,
The stability/transition calculations for D.F., "Recent Advances in Computational
the flap indicated the strong low Reynolds Methods to Solve the High-Lift Multi-
number effect with separation bubbles component Airfoil Problem", NLR Rpt
around 10 to 15 percent in extent and with MP84042U. 1984.
transition occurring inside the separation
bubble. These studies are still in prog- 8. Omar, E., Zierten, T. and Mahal, A.,
ress, and the results shown in Fig. 19 "Two-Dimensional Wind-Tunnel Tests of
correspond to results with transition a NASA Supercritical Airfoil with Var-
specified at 5% chord from the leading edge ious High-Lift Systems. 1 - Data
and thus do not include those obtained from Analysis", NASA CR-2214, 1973.
the stability/transition analysis. The
calculated results in Fig. 19 show discrep- 9. Omar, E., Zierten. T., Hahn, M . .
ancies even at lower angles of attack, and Szpiro, E. and Mahal, A., IITwo-
it is believed that with improved stabil- Dimensional Wind-Tunnel Tests of a
ity/transition calculations, the results NASA Supercritical Airfoil with Vari-
will agree much better with data. ous High-Lift Systems. 2 - Test Data",
NASA CR-2215, 1977.
6. CONCLUDING REMARKS
The results and discussion of the previous 10. Olson, L.E. and Orloff, K.L.,"On the
sections show that the present interactive Structure of Turbulent Wakes and Merg-
method, with its consideration of the flap- ing Shear Layers of Multielement Air-
well region and the wakes, leads to pres- foilst1,A I Mpaper 81-1238, June 1981.
sure coefficient distributions and values
of lift that are in good agreement with 11. Garner, P.L., Meredith, P.T. and
experiment for single and multielement con- Stoner, R.C., iiAreasfor Future C F D
figurations with angles of attack up to Development as Illustrated by Transport
and beyond stall. Aircraft Applications", AIAA Paper No.
91-1527, 1991.
For multielement wing flows, the pcelim-
inary results are encouraging. Calcula- 12. Mavriplis, D.J., "Research on Unstruc-
tions show the importance of low Reynolds tured Grid Techniques for C F D at
number effect on the components of the ICASE", Paper presented at the CFD
high-lift configuration at wind tunnel Conference, NASA Ames, 12-14 March
Reynolds numbers, and indicate that a 1991.
calculation method, either based on the
solutions of the Navier-Stokes equations 13. Rogers. E.S., Wiltberger, N.L. and
O K o n a combination of inviscid and Kwak, D., "Efficient Simulation of
viscous flow solutions, must include the Incompressible Viscous Flow Over
prediction of the onset of transition as Single- and Multielement AirfoilsBo,
part of the calculation procedure. AIAA Paper No. 92-0405, 1992.

7. REFERENCES 14. Barth, T.J., "CFD Algorithms on

Unstructured Meshes", Paper presented
1. Nakayama, A., "Flowfield Survey Around at the C F D Conference, NASA Ames, 12-14
High-Lift Airfoil Model LB 546". March 1991.
McDonnell Douglas Co. Rpt MDC 54827.
1987. 15. Cebeci, T., Clark, R.W., Chang,, K.C.,
Halsey, N.D. and Lee, K., "Airfoils
2. Nakayama, A., "An Experimental Investi- with Separation and the Resulting
gation of a Flow Around the Flap Well Wakests, Journal Fluid Mechanics, Vol.
of a Multielement AirfoilI8, Douglas 153, pp. 323-3470 1986.
Aircraft CO. Rpt M D C K4310, 1990.
16. Cebeci, T., Jau, J., Vitiello, D. and
3. Valarezo, W.O., Dominik, C.J. and Chang, K.C., "Prediction of Post-Stall
wilcox, P.A., "High Reynolds Number Flows on Airfoils", in ''Numerical and
Test Results for a Supercritical Multi- Physical Aspects of Aerodynamic Flows,
element AirfoiltB,McDonnell Douglas Rpt IV", (ea. T. Cebeci) Springer-Verlag,
M D C K5545, Dec. 1990. N.Y.. 1989.

4. Valarezo, W.O., Dominik, C.J. and 17. Cebeci, T., iuEssentialIngredients of

McGhee, R.J., "Reynolds and Mach Num- a Method for Low Reynolds-Number Air-
ber Effects on Multielement Airfoils", foils", AIAA J., 2 7 . 1983, pp. 1.680-
in "Proceedings of the Fifth Symposium 1688.
on Numerical and Physical Aspects of
Aerodynamic Flows18,13-15 January 1992, 18. Hess, J.L. and Smith, A.M.O., i°Calcula-
California State University, Long tion of Potential Flow About Atbitrary
Beach, CA. Bodies." Prog. in Aerospace Sci., 8 ,
Pergamon Press, N.Y., 1966.
 24-15

19. Cebeci, T. and Smith, A.M.O., "Analysis 30. Cebeci, T., Chen, H.H., Arnal, D. and
of Turbulent Boundary Layers", Academic Huang, T.T., "A Three-Dimensional Lin-
Press, N.Y., 1974. ear Stability Approach to Transition
on Wings and Bodies of Revolution at
20. Michel, R., "Etude de la Transition Incidence. AIAA J., December 1991,
sur le6 Profiles d'Aile," Establisement p p . 2077-2085.
dlun Critere de Determination de Point
de Transition et Calculde la Trainee 31. Johnson, D.A. and King, L.S., "A Mathe-
de Profile Incompressible, ONERA Rpt. matically Simple Turbulence Closure
1/1578A, 1951. Model for Attached and Separated Turb-
ulent Boundary Layersos,AIAA J., 23,-
21. Smith, A.M.O. and Gamberoni, N., 1985, p p . 1684-1692.
"Transition Pressure Gradient and
Stability Theory", in "Proceedings of 32. Johnson, D.A. and Coakley, T.J.,
the International Congress of Applied llImprovementsto a Nonequilibrium
Mechanics", Brussels, 4. 1956, p. 234. Algebraic Turbulence Model", AIAA J..
-
28, 1990, p p . 2000-2003.
22. Van Ingen, J.L.. "A Suggested Semi-
Empirical Method for the Calculation 33. Cebeci, T., "An Approach to Computa-
of the Boundary-Layer Regionlo, VTH71, tional Fluid Dynamics', book in
VTH74, Delft, Holland, 1956. preparation.

23. Cebeci, T., Jau, J., Vitiello, D., "An 34. Nakayama, A., "Measurements in the
Interactive Boundary-Layer Approach to Boundary Layer and Wake of Two Airfoil
Multielement Airfoils a t High Lift", Modelsoo,Douglas Aircraft Co. Rpt. MDC
AIAA Paper No. 92-0404. January 1992. 52403. 1988.

24. Hess, J.L., llCalculationof Potential 35. McGhee, R.J., Jones, G.S. and Jouty,
Flow About Arbitrary Three-Dimensional R., "Performance Characteristics from
Lifting Bodiesoo,Douglas Aircraft Co. Wind-Tunnel Tests of a Low Reynolds-
Ret. 55679-01, 1972. Number AIAA paper 88-0607,
January 1988.
25. Bradshaw, P., Cebeci, T. and Whitelaw,
J.H., "Engineering Calculation Methods 36. Cebeci, T., Chang, K.C., Clark, R.W.
for Turbulent Flows88,Academic Press, and Halsey, N.D., 'ICalculation of Flow
London, 1981. Over Multielement Airfoils at High
Lift", Journal of Aircraft, Vol. 14,
26. Veldman, A.E.P.,"New. Quasi-simultane- pp. 546-551, 1987.
ous Method to Calculate Interacting
Boundary Layers", AIAA J., 154, 1981. 37. Halsey, N.D.. "Potential Flow Analysis
pp; 79-85. of Multielement Airfoils Using Con-
formal Mapping", AIAA Journal, Vol.
27. Malik, M.R., "COSAL - A Black-Box Com- 17. p. 1281, 1979.
pressible Stability Analysis code for
Transition Prediction in Three- 38. Lovell, D.A., "A Wind-Tunnel Investi-
Dimensional Boundary Layers", NASA CR gation of the Effects of Flap Span and
165952, 1982. Deflection Angle, Wing Planform and a
Body on the High-Lift Performance of a
28. Mack, L.M., ulBoundary-LayerStability 28O Swept Wing", CP No. 1372, 1977.
Theorya8,AGARD Report No. 709. 1984.
39. Cebeci, T., Chen, H. H. and Lee, B.P..
29. Cebeci, T. and Stewartson, K., "Stabil- nCalculation of Three-Dimensional Low
ity and Transition in Three-Dimensional Reynolds Number Flows", to be published
Flow". AIAA J., a, 1980. p p . 398-405. in J. Aircraft. 1993.
 25-1

WAKE STRUCTURE AND AERODYNAMIC BEHAVIOR OF

HIGH LIFT AIRCRAFT CONFIGURATIONS
DURING UNSTEADY MANEUVERS
IN GROUND EFFECT

A.Baron, M.Boff adossi

Dlpartimento dl Ingegnerla Aerospaziale del Polltecnico dl Milano
Vla C. Golgl, 40 - 20133 Mliano - Italy

SUMMARY
A non linear unsteady vortex lattice scheme is used and flight dynamics equations a r e
solved in order t o predict the structure of the wakes and the instantaneous distribution
of the aerodynamic loads on high-lift aircraft configurations, during general unsteady
take-off maneuvers in ground effect.
The numerical scheme here presented can t r e a t an arbitrary number of mutually interfering
lifting and moving control surfaces having arbitrary plan form and camber. Wakes can be
released in the flowfield from any of the sharp edges of the lifting surfaces, depending
on their plan form, aspect ratio and angle of attack, while the effects of the fuselage
are ignored in the present formulation.
Turbulent diffusion of the cores of the Rankine vortex filaments is regarded as a
pre-eminent factor in a correct simulation of the development of unsteady interfering
wakes. A vortex core diffusion model is used capable t o deal even with the severe
roll-up of the mutually interfering wakes developing close and impinging on the ground.
Typical applications of the unsteady vortex lattice scheme a r e presented, aimed at
illustrating the capabilities of the code.

LIST OF SYMBOLS

C wing root chord v relative velocity ( a i r c r a f t fixed

CL lift coefficient frame of reference)
Fa total aerodynamic force V absolute velocity
I a i r c r a f t moment of inertia x,y,z coordinates in the aircraft
K constant in the turbulent diffusion fixed barycentric frame of ref.
model Xk,yk,zk coordinates in the surface
e spanwise dimension of the elementary fixed frame of reference
portion of shear layer X,Y,Z coordinates in the absolute
L load on undercarriage leg frame of reference
m mass 5 angle of attack
Ma moment of the aerodynamic forces ;r linear vortex density vector
about barycentric axes r circulation
rc vortex core radius wy angular velocity component along y
r relative position vector
suffixes
R absolute position vector
f? rotation tensor G center of gravity
S surface of the cross section of the h hinge axis
vortex core k generic K-th lifting surface
t time m main undercarriage
T dimensionless time T=tU/C n nose undercarriage
T engine thrust o initial value; undeflected position
Q a i r c r a f t weight p generic point P
U characteristic velocity y y component
25-2

1. INTRODUCTION
The antithetical requirements f o r high speed cruise and short take-off and landing
capabilities of modern aircrafts has promoted the development of more and more efficient
high-lift systems (Dilliner et al., 1984).
While the ability in numerically predicting the performances of multi-component airfoils
at high angle of attack, even in viscous and compressible flows, has been enormously
improved in recent years, limited advances have been obtained in the numerical simulation
of the complex vortical flows developing past and interfering with the lifting surfaces
of complete a i r c r a f t configurations .
However, a detailed knowledge of the geometry of the three-dimensional wakes is essential
f o r both an effective optimization procedure of the high-lift devices and the preliminary
choice of the a i r c r a f t configuration.
As a matter of fact, high-lift devices are not merely required t o produce maximum lift:
other constraints have t o be accounted for, such as their drag. Whilst there are
substantial opportunities in reducing the profile drag of leading and trailing edge
devices, by improving their sectional design and optimization, major performance
improvements, particularly at high angles of attack, will only come from an improved
capability of predicting the complex vortical flow past three-dimensional wing
geometries.
High-lift devices indeed necessarily influence the induced drag which strongly depends on
wing and flap planforms, camber and twist, and their trim drag penalty is a function of
the relative position of wing, flaps, and control surfaces. Both induced and trim drag
can only be evaluated provided the complex interaction between three-dimensional wakes
and lifting surfaces is correctly predicted (Butter, 1984).
Furthermore, high-lift devices a r e used on most aircrafts during take-off and landing
maneuvers, namely during flight phases which a r e critical because of unsteadiness and
ground proximity. Both unsteadiness and ground effect can produce strongly non linear
aerodynamic effects and interferences, which can significantly affect the controllability
and the operative safety of the aircraft.
Finally, knowledge of the time evolution of the wakes produced by high-lift
configurations is also essential f o r attempting to reduce the hazardous persistence of
the vortices in the proximity of high a i r traffic density airports.

2. GROUND EFFECT
When an airplane flies close t o the ground, at a height comparable with i t s wing chord,
i t experiences an increase in lift and remarkable changes in drag and pitching moment.
This phenomenon is called ground effect.
Ground effect is particularly significant f o r V/STOL and military low aspect-ratio
aircrafts, mainly because of the strength and the complex shape of the unsteady wakes
they produce.
Starting from the f i r s t basic theory developed by Wieselsberger in 1922, extensive
research, both theoretical and experimental, has been devoted t o understanding and
predicting ground effect.
Wieselsberger’s steady formulation ignored the effect of the bound vortex in the image
however, introducing an image of the real wing below the ground plane and using the basic
concepts of the lifting-line theory of Prandtl, he was able t o calculate a correction to
the induced drag and angle of attack of wings out of ground effect.
Since then, almost all investigators adopted the virtual image concept t o obtain a
theoretical or numerical simulation of the ground effect.
Unsteadiness was f i r s t introduced in a two-dimensional ground effect model by Chen and
Schweikhard in 1985, who considered a straight imposed wake and found that unsteady
effects can be such t o completely modify the steady state predictions.
The findings of Chen and Schweikhard started extensive experimental work on finite wings
of various plan forms in both steady and unsteady ground effects (Chang and Muirhead,
 25-3

1985 and 1987).

Finally, in 1985, Katz proposed a comprehensive numerical model f o r the prediction of the
unsteady ground effect, which also included a freely deforming wake. He adopted a
vortex-lattice scheme and investigated the behavior of finite span lifting surfaces close
t o ground, so evidencing again that the increases in aerodynamic loads due t o
unsteadiness can be approximately twice those produced in steady s t a t e conditions.
The same free-wake approach has been adopted in the present work and, using the virtual
image technique, a design tool has been derived capable to t o predict the instantaneous
load distribution and geometry of the wakes on multiple lifting surfaces, during general
unsteady maneuvers in ground effect.

3. THE COMPUTATIONAL METHOD

The numerical simulation of unsteady airplane maneuvers obviously requires time dependent
schemes, which allow t o reproduce the actual time dependent boundary conditions
representing the instantaneous velocity, flight attitude and location of the a i r c r a f t
with respect t o the ground. The motion of the aircraft can be either assigned or
computed through flight dynamics models and appropriate control laws (see next
paragraph 1.
The Navier-Stokes equations constitute a complete mathematical model f o r general
three-dimensional unsteady flows of viscous fluids and can theref ore model all of the
complex phenomena mentioned above.
However, their solution requires, even f o r an incompressible fluid, storage capabilities
and computing times which are still unacceptable. Although computer performances a r e
dramatically increasing, recourse t o simpler mathematical models, such as those based on
the irrotational flow hypotheses, is still justified, if not unavoidable.
In the present work, results concerning the take-off maneuver of high-lift aircraft
configurations are presented, obtained using an unsteady vortex lattice scheme. Details
of the mathematical and numerical approach are given in Baron et al. 1989, 1990, 1991 and
1992, where various applications of the vortex-lattice scheme t o fixed and rotary wings
a r e also reported. Only the main features and limitations of the method a r e here
summarized.
Wakes and flowfield a r e computed simultaneously, starting from an initial s t a t e of rest.
Wings are impulsively started and wakes a r e generated with a Lagrangian process during
which, at each time step, the vorticity present on their edges is convected in the field
(Belotzerkowskii, 1977; Kandil e t al., 1977; Hoeijmakers, 1983; Konstandinopoulos e t al.,
1985; Mook, 1988; Katz and Maskew, 1988; Baron et al., 1990).
A t each time step, a new row of vortex panels is added t o the wakes, which a r e convected
in the field in such a way that the f r e e vortex sheets a r e force-free.
A Lagrangian generation of the wakes does not require a f i r s t guess of their
configuration, which can be critical when complex high-lift configurations have t o be
considered.
Wings, flaps and tail planes of arbitrary plan form and camber a r e simulated as surfaces
of negligible thickness and separation can be imposed on any of their sharp edges.
Lifting surfaces and wakes are discretized into a finite number of surface panels. Each
panel is made up of straight vortex segments lying on the edges of the panel, forming a
closed loop with constant circulation (Mook, 1988).
The unknown values of the circulations on the lifting surface panels a r e determined, at
each time step, by imposing the zero normal velocity condition on the solid surfaces,
including the surfaces of the virtual image of the aircraft below the ground plane.
Pressure distributions are computed by means of the unsteady Bernoulli equation. Lift
and moment coefficients a r e obtained by integration of pressure.
In order t o compute the velocity induced by vortex segments, the Biot-Savart law is used,
together with a viscous core diffusion model, physically consistent with the turbulent
diffusion mechanism of continuous shear layers (Baron et al., 1990).
Rankine vortices a r e assumed t o be "equivalent" to the elementary portions of the
25-4

physically continuous shear layers they replace in the numerical scheme. Therefore,
their core radii spread in such a way that their cross sectional area S and circulation
r=y! a r e equal, at each time, t o spreading and circulation of an elementary portion of
continuous shear layer containing the same vorticity. This model also implies t h a t the
enstrophy of the vortex lattice is equal to the enstrophy of the continuous turbulent
shear layer and introduces a correct amount of energy dissipation in the wakes.
This implies a rate of change of the radius rc of the Rankine vortex cores given by:

-drc
-- - K r (3.1)
dt 2nrc

where r is the circulation of the vortex filament considered and the experimental
diffusion constant K, f o r both forced and unforced turbulent shear layers, assumes a
universal value equal t o 0.095 (Liepmann and Laufer 1947; Brown and Roshko, 1974; Oster
and Wygnanski, 1982; Lesieur, 1987).
Forms analogous t o (3.1) a r e also proposed by Squire (19651, Govindaraju and Saffman
(1971) and Leonard (1980) f o r the turbulent diffusion of single line vortices, however
they involve empirical constants the value of which cannot be defined in a general form.
The present approach brings t o a diffusion model consistent with the behavior of
continuous turbulent shear layers and, in addition, information of experimental nature is
introduced in a fully general way.
Moreover, ,most commonly used vortex core diffusion models turn out t o be strongly
dependent on the number of vortex filaments used t o discretize the continuous
distribution of vorticity in the flowfield (Rusak et al., 1985). The present approach,
on the contrary, is virtually independent on discretization, the turbulent diffusion
being explicitly related to the circulation of each vortex filament and, therefore, t o
the number of vortices.
This brings t o a kind of "self adaptation" of the model and explains why, in a variety of
applications, the vortex core turbulent diffusion model proved t o be capable t o cope,
without any form of tuning, with the crucial aspects related t o the roll-up process of
unsteady interfering wakes as well as t o the interaction of vortex filaments with both
the solid lifting surfaces and ground.
A final comment can be done on the computing efficiency of unsteady vortex lattice
schemes.
The simulation of unsteady vortical flows is intrinsically expensive from a numerical
point of view, as the circulations of the transversal vortical filaments, which a r e
virtually absent at steady state, must also be accounted for. In addition, a Lagrangian
generation of the wakes requires longer computing times before a sufficiently
representative portion of the shed wakes is produced.
However, a significant computing time saving can be obtained if only the mutual
induction of the wake elements and the lifting surfaces is considered, while
self-induction of the vortex filaments is neglected, f o r the f a r t h e r portion of the shed
wakes.
Self -induction effect simulation is a peculiar property of vortex lattice schemes and
proves t o be essential f o r a correct determination of the configuration of the wakes.
However, as a matter of fact, in the ''far wake" it only acts in increasing the local
roll-up and stretching processes of the vortical sheets, without affecting at all their
induction (and the aerodynamic load distribution) on the lifting surfaces f a r upstream.
I t turns out that, f o r practical purposes, in most cases, vorticity in the wake can be
simply convected, starting from a downstream distance of the order of some wing spans.
Being most of the computing time devoted t o the evaluation of the self-inducing effects,
this procedure enormously speeds up the computations, without affecting the accuracy of
the numerical predictions.

4. FRAMES OF REFERENCE AND FLIGHT DYNAMICS MODEL

In order t o determine the instantaneous unknown values of the circulations on the lifting
surface panels, the zero normal velocity condition on the solid surfaces is imposed, a t
 25-5

each time step.

This implies t h a t the coefficients of the aerodynamic influence matrix, which rely on the
instantaneous location and attitude of each surface element in the absolute frame of
reference, must be computed.
Moreover the instantaneous velocity of the control point of each surface panel must be
prescribed. This is the result of both the linear and angular velocities of the a i r c r a f t
and obviously depends on the instantaneous linear and angular accelerations. These, in
turn, rely on the mass and moment of inertia of the airplane and on the distribution of
the aerodynamic loads on i t s surfaces.
Therefore appropriate frames of reference and flight dynamics equations must be
considered.
An absolute ground fixed frame of reference O(X,Y,Z) and an a i r c r a f t fixed frame of
reference G(x,y,z), having i t s origin in the center of gravity G of the airplane, a r e
defined.
An additional k-th local frame of reference is introduced Hk(xk,yk,Zk), fixed with each
k-th lifting surface. The origin of the k-th frame of reference is coincident either
with the leading edge of the fixed k-th surface, or with the hinge axis of the k-th
control surface.
A t time t , the relative position vector r p ( t ) of a generic point P laying on the k-th
lifting surface, in the aircraft fixed frame of reference, can be expressed as a function
of i t s position r p o in the local frame Hk(xk,yk,zk), as:

where Rk(t) is the tensor associated with the rotation of the k-th surface, with respect
to i t s undeflected angular position. Rk(t) therefore expresses the instantaneous
contribution t o the displacement of point P associated t o the deflection imposed t o the
k-th control surface. r h k is the location of the hinge axis of the k-th control surface
in the a i r c r a f t fixed frame of reference.
The absolute position vector Rp(t) of point P, can be is given by:

where % ( t )is the tensor associated with the rotation of the aircraft, with respect t o
i t s initial angular position at time t=O, and R G ( ~ is
) the absolute position vector of
the center of gravity of the airplane.
Consequently, the absolute velocity Vp(t) of a generic point P laying on the k-th surface
can be determined according to:

where W ( t ) is the instantaneous angular velocity of the aircraft, V d t ) is the linear

velocity of i t s center of gravity, and vp(t) is the velocity of point P in the airplane
fixed frame of reference:

where vhdt) is the relative instantaneous velocity of the k-th hinge axis, around which
the angular velocity of the k-th control surface is wk(t). vhk(t) is obviously equal t o
zero f o r fixed hinge flaps and control surfaces.
If the motion of both the aircraft and the control surfaces is assigned "a priori", the
instantaneous values of W t ) , V G ( ~ )vhk(t)
, and w d t ) a r e known.
Otherwise, only the deflection laws f o r the control surfaces are imposed, while W ( t ) and
V d t ) are determined through the dynamic balance of the forces and moments produced by
the integration of the instantaneous pressure coefficients. In the present formulation a
simple flight dynamics model based on the following rigid body equations is adopted.
25-6

The linear momentum conservation is imposed:

m VG = Fa +T + Q + Ln + Lrn (4.5)

where m is t h e mass of the a i r c r a f t , VG i s the linear acceleration of i t s center of

gravity, F a is t h e sum of t h e aerodynamic l i f t and drag, T i s t h e engine t h r u s t (assumed
t o be constant), Q is t h e airplane weight, Lm and Ln are the forces exerted by t h e nose
and main undercarriage legs, during the take-off run.
The conservation law f o r the angular momentum about t h e center of
airplane, in its plane of symmetry, brings to:
gravity of the 1
Iyy w y = May + Xn Ln + Xm Lm (4.6)

where Iyy i s t h e moment of inertia of the a i r c r a f t about the y axis, ;y is t h e component ~

of t h e angular acceleration around the y axis, M a is the moment of t h e aerodynamic f o r c e s

with respect t o t h e center of gravity and the last two t e r m s represent t h e moments of t h e
f o r c e s exerted by t h e undercarriage legs during the take-off run. 1
Equations (4.5) and (4.6) are used t o determine the dynamic behavior of t h e a i r c r a f t 1
during i t s flight. Obviously, during the take-off run Ln, Lm and t h e i r moments are 1
unknown, an! both t h e vertical component of the linear acceleration and t h e angular 1
acceleration w y are equal t o zero, while, when Ln goes t o zero, t h e rotation phase takes 1
place, and t h e airplane angular acceleration w y must be computed. I

5. THE SELECTED AIRCRAFT CONFIGURATIONS

The unsteady take-off maneuver in ground e f f e c t i s numerically simulated and analyzed in 1

the next paragraphs f o r various high-lift configurations, which reasonably approximate

STOL military o r civil t r a n s p o r t a i r c r a f t s (see Fig.1).
The basic a i r c r a f t configuration has the following geometrical characteristics. I

The wing h a s a t a p e r r a t i o of 0.5, a root chord C equal t o 5.4 meters, a span equal t o 4 0
2
meters and a s u r f a c e of 176 m. I t i s located 0.8 root wing chords above ground.
The t a i l plane h a s t h e same t a p e r ratio, a span equal t o 3 C and i s located in t h e plane
of t h e wing with its apex 2 root wing chords downstream the trailing edge of t h e wing.
,
The hinge a x i s of t h e elevator i s located at 2/3 of the local tail plane chord.
In i t s basic configuration (A), the airplane is f i t t e d with double slotted trailing edge
f l a p s having a span and a central cutout equal t o 5.36 and 0.87 root wing chords
respectively. Each f l a p element has a constant chord equal t o 0.2 C. The f l a p elements
are set t o a n angle of a t t a c k of 11.25 and 22.5 degrees respectively. I
Additional double slotted f l a p configurations (B and C), are obtained by only modifying
the f l a p planforms, while maintaining their surface unchanged.
Configuration B h a s f l a p s with constant chord increased 50% with respect t o t h e basic
configuration, and a proportionally reduced span, the central cutout being unchanged.
Configuration C i s obtained from the basic one by introducing cutouts simulating t h e
presence of wing-mounted engines, and consequently increasing the f l a p span t o 6.1 C.
A f u r t h e r configuration D i s defined, having a single slotted f l a p with constant chord
spanning t h e whole wing, set at 22.5 degrees.
For all of these configurations, the airplane is assumed t o have a mass of 34.000 Kg, a
2
moment of inertia around t h e y axis equal t o 400.000 K g m , a center of gravity located
in t h e plane of symmetry at 50% of the wing root chord and at 0.4 root wing chords above
I
ground.
I
6. COMMENTS ON THE NUMERICAL SIMULATION OF UNSTEADY TAKE-OFF MANEUVERS ICE 1
The attention is focused in t h i s paragraph on some general aspects related t o t h e use of I
vortex l a t t i c e schemes in t h e numerical simulation of unsteady a i r c r a f t maneuvers in
ground e f f e c t .
 25-7

Fig.1 - The selected A, B, C and D high-lift aircraft configurations

The time dependent evolution of the wakes past the basic aircraft configuration, reported
in Fig.2, impressively evidences the capability of unsteady vortex lattice schemes in
treating evolutive vortical flows.
However, as mentioned in paragraph 3, this is only possible provided the turbulent
diffusion process of the cores of the discrete vortices is adequately modeled in a
general and physically consistent way.
Also the accuracy of the geometrical discretization, particularly in the spanwise
direction, brings t o an increasingly better definition of the configuration of the wakes
and t o a more accurate prediction of the load distribution on the lifting surfaces, but
obviously requires rapidly growing computing times.
A preliminary analysis of the sensitivity of the scheme t o panel density has shown that,
f o r relatively low taper and high aspect-ratio plan forms, an acceptable prediction of
the aerodynamic load distributions can be obtained even with the limited number of vortex
panels used in these simulations. Solutions are only negligibly improved by further
increasing the number of panels.
Flowfield and aerodynamic loads rely upon the instantaneous location of the shed wakes.
Therefore, an appropriate integration time step must be used t o accurately evaluate the
displacement of the nodes of the vortex lattice. Moreover, the time discretization also
affects the longitudinal dimension of the vortex panels in the wake, so influencing the
uniformity of the vortex elements. Though not strictly compulsory, time discretization
should be such t o produce nearly uniform vortex elements on wakes and lifting surfaces.
These are known t o increase accuracy of the solutions and stability of the numerical
scheme. The numerical results here presented have been obtained using a dimensionless
integration time step T=0.25, which meets rather well the antithetical requirements of
accuracy and computing time.
In order t o describe how the take-off maneuver is accomplished, reference is made in the
following t o Fig.3, where the time history of the lift coefficients of wing, flap and
tail plane is reported f o r the airplane configuration A, in ground effect (IGE).
25-8
e
A numerical analysis of the whole take-off maneuver (including the take-off run, the
rotation phase and the final climb) can be performed by using the vortex lattice scheme,
imposing appropriate deflections to the control surfaces, and integrating the flight
dynamics equations 4.5 and 4.6. However, attention is focused in the present work t o the
simulation of the rotation phase of the aircraft, during which unsteadiness of the flight
condition and time varying ground effects a r e believed to play a significant role.
Therefore, in order t o speed-up the numerical simulation, the initial part of the actual
take-off run has been replaced by an impulsive start.
Then, once the effects of the impulsive start a r e sufficiently damped and an
approximately steady s t a t e configuration of the near wakes is attained (T=7 approx., in
Fig.31, the elevator is progressively deflected from 0 t o -20 degrees, at a r a t e of 30
degrees per second.
A t the instant deflection begins,
the tail surface lift decreases and
starting vortices a r e shed in the
field at the edges of both the
stabilizer and the elevator.
Note that the relative location and
attitude of the elevator panels
vary progressively with respect t o
both the panels of the fixed
portion of the tail plane and wing.
This implies that the coefficients
of the aerodynamic influence matrix
must be computed at each time step,
during the deflection phase of the
elevator.
 25-9

1.50

0.50
-
W++Q
Wing
Flap

-0.50

-1.50 1
1

0.00
1 1 1 1 1 ~ 1

5.00
1 1 1 1 1 1 1 1 1 1

IO.0 0
1 1 1 1 1 1 1 1 1

15.00
1 1 1 1 1 1 1 1 1

20.00
1 1 1 ~

dimensionless time
Fig.3 - Time history of the lift coefficients of wing, flap and tail plane for
configuration A, during the impulsively started take-off maneuver IGE.

A s a consequence of the elevator deflection (time units from 7 t o 12 approx., in Fig.31,

the a i r c r a f t is subject t o a pitching moment which progressively grows until the nose
wheel lifts up and i t starts rotating in i t s plane of symmetry, according t o the dynamic
equilibrium laws. Initial rotation takes place around the axis of the main undercarriage
wheels, which also provide the vertical force still necessary t o balance the a i r c r a f t
weight.
This rotation produces a progressively increasing angle of attack and causes the lift of
the wing and flaps t o increase while the negative lift of the tail plane reduces.
During this phase, the distance from the ground of both the lifting and the control
surfaces of the a i r c r a f t varies significantly with time and the unsteady ground effect
begins t o play an important role.
Also the mutual induction of the lifting surfaces and their time varying interference
with the shed wakes cause non linear aerodynamic effects.
During the rotation phase, which has a relatively short duration, the horizontal
component of the velocity of the aircraft can be reasonably assumed t o vary negligibly
with time. This justifies the adoption of a simplified flight dynamics model in which
the equation f o r the horizontal component of the momentum is eliminated and a constant
horizontal acceleration is imposed t o the aircraft. Although the practical consequences
of this simplification a r e completely negligible, both in terms of computing time and
complexity of the numerical scheme, i t allows t o perform numerical simulations which can
more easily be analyzed in a comparative way. In fact, provided the same law is imposed
t o the deflection of the elevator, all of the rotation phases will start at the same
dimensionless time, with the same value of the horizontal component of the velocity,
independently on the particular examined configuration.
When the angle of attack of the airplane reaches a value of approximately 10 degrees,
lift becomes larger than weight, the airplane starts climbing, and ground effect
progressively reduces.

Also during the rotation, the take-off and the climbing phases the coefficients of the
aerodynamic influence matrix must be computed at each time step, not only because of the
control law imposed t o the elevator, but also due t o the f a c t that the lifting surfaces
vary their distances and attitudes relative t o their virtual images below the ground.
25-10

7. THE INFLUENCE OF GROUND ON THE TAKE-OFF MANEUVER

Ground proximity affects the evolutive structure of the wakes released by the lifting
surfaces. Ground confines the vertical development of the wakes and promotes a
consequent spanwise spreading.
This is clearly evident in Fig.4, where the plan and lateral views of the wakes of
configuration A a r e shown, at T=18,f o r both the OGE and the ICE maneuvers.
Although the outboard displacement of the tip vortices of wing and flaps does not
completely alter the global structure of the wakes, i t significantly enhances their
interaction with the tail plane.

Fig.4 - Plan and lateral views of the wakes of configuration A, OGE (top) and ICE.
 25-1 1

Fig.5 - Cross section of the wakes released by the wing and flaps of configuration A, ICE
and OGE.

This is also confirmed by Fig.5, where the cross sections of the IGE and OGE wakes of
configuration A, are shown f o r comparison, in a plane normal t o the ground, 1 root wing
chord downstream the wing trailing edge. Solid circles a r e used t o represent the cross
section of the longitudinal vortical filaments, which have radii proportional t o their
circulation.
N o t only stronger vortices are produced in ground effect (IGE), but also their location
relative t o the tail plane is modified in both spanwise and vertical directions (also see
Fig.4 1.
As a consequence, unsteady ground effects can be responsible f o r significant
modifications in the efficiency of the tail plane and can remarkably affect the take-off
maneuver.
In Fig.6 the time evolution of the DCL developed by the tail plane, with respect t o the
IGE maneuver of Fig.3, is reported (DCL is defined as CL(OGE)-CL(IGE)).
In addition t o a stronger sensitivity t o the starting wake released by the wing upstream
(T=2), the tail plane in ground effect is always shown t o produce a reduced negative
lift.

-0.00 L
-
-
-
-
-
-
-0.40 I I I I I I I I I J I I I I I I I I I J I I I I I I I I I I ~1 I I f I

Fig.6 - Time history of the l i f t coefficient produced by the tail plane OGE, with respect ,
to the ICE condition.
25-12

Fig.7 -- Plan view of the wakes of the aircraft configurations B (top), C and D (bottom),
in the final stage (T=20)of the take-off maneuver IGE.
 25-13

8. THE INFLUENCE OF VARIOUS HIGH-LIFT CONFIGURATIONS

The influence of various flap configurations on the take-off maneuver in ground effect
are examined.
In Fig.7 plan views are shown of the wakes produced, during the take-off maneuver, by the
a i r c r a f t configurations B, C and D defined in paragraph 5.
Although all of t h e considered configurations approximately produce the same amount of
lift, streamwise vorticity in their wakes is distributed in completely different
patterns, as evidenced in Fig.8, where the solid circles used t o represent the cross
section of t h e longitudinal vortices, have radii proportional t o the strength of each
vortex filament.
Various are t h e consequences of the different vorticity distribution in t h e wakes.
Structural loads, f o r instance, and induced drag depend on the spanwise distribution of
lift which, in turn, relies on vorticity distribution. Highly concentrated unsteady
vortices can induce vibratory loads on the tail surfaces, can be ingested by tail-mounted
engines o r induce fatigue loads on the propellers of canard configurations.

I
Fig.8 - Cross section of the wakes past various high-lift aircraft configurations
ICE, in a plane 1 root wing chord downstream the wing trailing edge (T=18).

A s a n example, t h e effect of the structure of the wakes is shown in Fig.9 on t h e spanwise

lift distribution on the tail plane of the examined configurations.

9. CONCLUSIONS
High-lift devices are mainly used during unsteady maneuvers in ground effect. Their
design and optimization, together with the controllability and operative safety of
aircrafts, relies on the capability of predicting the complex structure of the
three-dimensional unsteady vortical flows which develop close and interfere with the
control surfaces and ground.
25-14

A numerical approach t o t h e simulation of unsteady maneuvers in ground e f f e c t , based on a

non linear vortex lattice scheme and a simple flight dynamics model, h a s been presented
and applied t o various reasonably realistic high-lift a i r c r a f t configuration. Further
developments of t h e scheme will allow t o account also f o r the fuselage e f f e c t s , which
have been neglected in the present formulation. I

For design purposes, the developed vortex lattice scheme seems t o constitute a reasonable
compromise between t h e antithetical needs of complete mathematical models of complex
flows and required computing times.
All of t h e main physical f e a t u r e s of t h e examined flows seem in f a c t t o be retained in
t h e numerical simulations, despite of the quite crude vortex lattice model used t o
discretize t h e continuous distribution of vorticity in the flowfield. t h i s i s partly 1
ascribable t o t h e use of a fully general model f o r the diffusion of t h e Rankine vortex
filaments. ~

I
In addition t o t h e significant role played by ground proximity, the numerical simulations
also evidence how t h e choice of appropriate high-lift configurations can completely
modify t h e s t r u c t u r e and t h e inducing effect of the vortical wakes shed in t h e flowfield.

0
0

'(+
3 -0.50
0
a,
m

Fig.9 - Spanwise lift distribution on the tail plane of various high-lift aircraft
configurations (A, B, C and D), during a take-off maneuver IGE (T= 18).
I

REFERENCES
BARON . A. , BOFFADOSSI M. (19891, "Simulazione numerica di scie vorticose non stazionarie
tridimensionali", Proceedings of the X" Convegno Nazionale di Aeronautica e
I
1
Astronautica, Pisa. I
BARON A., BOFFADOSSI M., DE PONTE S . (19901, "Numerical simulation of vortex flows past
impulsively s t a r t e d wings", AGARD-CP-494. I

BARON A., BOFFADOSSI M. (1991) ,"Numerical Simulation of Unsteady Rotor Wakes", I

Proceedings of t h e XVII-th European Rotorcraft Forum, Berlin, Paper N"91-69.
BARON A., BOFFADOSSI M. (19921, "Unsteady Free Wake Analysis of Closely Interfering
 25-15

BARON A., BOFFADOSSI M. (19921, "Unsteady Free Wake Analysis of Closely Interfering
Helicopter Rotors", XVIII-th European Rotorcraft Forum, Avignon, Paper No 108
BELOTZERKOWSKII S.M. (19771, "A study of unsteady aerodynamics of lifting surfaces using
t h e computer", Annual Review of Fluid Mechanics, Vo1.9, pp. 469-494.
BROWN G.L., ROSHKO A. (19741, "On density effects and large s t r u c t u r e in turbulent mixing
layers", Journal of Fluid Mechanics, Vol. 64, pp. 775-816.
BUTTER D.J. (19841, "Recent progress on development and understanding of high lift
system", AGARD-CP-365
CHANG R.C., MUIRHEAD V.U. (19851, "Investigation of Dynamic Ground Effect", NASA-CP-2462
CHANG R.C., MUIRHEAD V.U. (19871, "Effect of Sink Rate on Ground Effect of
Low-Aspect-Ratio Wings", Journal of Aircraft, Vo1.24, pp. 176-180
CHEN Y.S., SCHWEIKHARD W.G. (19851, "Dynamic Ground Effects on a Two-Dimensional Flat
Plate", Journal of Aircraft, Vo1.22, pp.638-640.
DILLINER B., MAY F.W., MCMASTER J,K. (19841, "Aerodynamic issue in t h e design of high
l i f t systems f o r transport aircraft", AGARD-CP-365
GOVINDARAJU S.P., SAFFMAN P.G. (19711, "Flow in a turbulent trailing vortex", The Physics
of Fluids, Vo1.14, No.10, pp. 2074-2080.
HOEIJMAKERS H. W.M. (19831, "Computational vortex flow aerodynamics", AGARD-CP-342
KANDIL O.A., ATTA E.H., NAYFEH A.H. (19771, "A three dimensional steady and unsteady
asymmetric flow past wings of arbitrary plan forms", AGARD-CP-227.
KATZ J. (19851, "Calculation of the Aerodynamic Forces on Automotive Lifting Surfaces",
Journal of Fluid Engineering, Vol. 107, pp. 438-443
KATZ J., MASKEW B. (19881, "Unsteady low speed aerodynamics model f o r complete a i r c r a f t
configurations", Journal of Aircraft, Vo1.25, No.4, pp 302-310.
KONSTANDINOPOULOS P., MOOK D.T., NAYFEH A.H., WATSON L. (19851, "A vortex lattice method
f o r general, unsteady aerodynamics", J. of Aircraft, V01.22, No. 1, pp 43-49.
LEONARD A. (19801, "Vortex method f o r flow simulation", Journal of Computational Physics,
V01.37, pp. 289-335.
LESIEUR M. (19871, "Turbulence in fluids, stochastic and numerical modelling", Martinus
N i jhoff Publishers, Dordrecht.
LIEPMANN H. W., LAUFER J. (19471, "Investigation of f r e e turbulent mixing", NACA-TN-1257.
MOOK D.T. (19881, "Unsteady aerodynamics", VKI L. S . 1988-07 on "Unsteady Aerodynamics"
OSTER D., WYGNANSKI I. (19821, "The forced mixing layer between parallel streams",
Journal of Fluid Mechanics, Vo1.23, pp. 91-130.
RUSAK Z., SEGINER A., WASSERSTROM E. (19851, "Convergence characteristic of a
vortex-lattice method f o r nonlinear configuration aerodynamics", Journal of
Aircraft, V01.22, No.9, pp. 743-749.
SQUIRE H.B. (19651, "The growth of a vortex in turbulent flow", The Aeronautical
Quarterly, Vo1.16, pp. 302-306.
WIESELSBERGER C. (19221, "Wing Resistance Near the Ground", NACA-TM-77
 26-1

CALCUL PAR INTERACTION VISQUEUX NON-VISQUEUX DES

ECOULEMENTS COMPRESSIBLES FORTEMENT DECOLLES
AUX GRANDES PORTANCES
SUR PROFILS D'AILES ET VOILURES
-----
( VISCOUS-INVISCID CALCULATION
OF HIGH-LIFT
SEPARATED COMPRESSIBLE FLOWS
OVER AIRFOILS AND WINGS )
-----
J.C. Le Balleur

ONERA, BP 72,92322 Chatillon Cedex (France)

Abstract. pleinement tridimensionnelle en mode inverse. Une extension

nouvelle "Massive-separation 2.750" de l'algorithme de
The viscous-inviscid interaction transonic numerical method couplage "Semi-inverse" de Le Balleur est donnte et dttaillte,
previously defined by the author for computing attached or en mcme temps que sa thtorie de stabilitt.
separated flows over airfoils, including the deeply stalled flows,
Des rtsultats sont montrts pour le dtcrochage 2 0 , et pour
is extended into a new three-dimensional method for strongly des tcoulements 3D-dtcollts sur aile rectangulaire ou aile en
separated flows over wings at high-lifr and compressible freche, avec un accord calcul-exptrience satisfaisant. Une tech-
speeds. nique d'auto-adaptation du maillage aux effets visqueux est
The numerical non-linearly implicit boundary layer technique prtsentte. Les rtsultats dtmontrent que les mtthodes
(directlinverse), the turbulent models, the grid generation and d'interaction visqueux-non visqueux donnent un plein acc2.s au
grid-adaption, the coupling and wake-equilibration algorithms, calcul du dtcollement tridimensionnel.
the inviscid full-potential schemes, are extended in three-
dimension, with approximation on the viscous equations
(2.75D-local). New theoretical results are given on the singu-
larities and characteristic cones of the fully three-dimensional
boundary layer in inverse mode. A new "Massive-separation 1. INTRODUCTION.
2.750" extension of the "Semi-inverse" algorithm of Le Balleur
for coupling is given and detailed, together with its stability- La simulation numkrique par "interaction visqueux-non
theory. visqueux" offre la possibilitC de dkvelopper des methodes B
Results are shown for 2D-stall, and for 30 separated flows plus faible dissipation numCrique, et 21 plus faible coct, que les
over rectangular or swept wings, with satisfactory agreement techniques de "rCsolution directe" d'tquations de Navier-
between theory and experiment. A self adaptation technique of Stokes. L'investissement dans cette mCthodologie, mal& ce
the grid to the viscous effects is displayed. The results demon- double avantage, semble toutefois avoir CtC desservi par 1'idCe
strate that the viscous-inviscid interaction methodr give a full ancienne et maintenant inexacte que son domaine de
access to the calculation of three-dimensional separation. dCveloppement serait toujours plus limitt, notamment vis 21 vis
du dCcollement Ctendu, et surtout du dCcollement tridimension-
nel.
R6sum6. Par les progrks des algorithmes de couplage, l'accbs au
calcul des dtcollements avait pu etre ouvert, en 2D ou 2.5D.
La mkthode numtrique d' interaction visqueux-non visqueux Ces progrks sur le couplage avaient CtC d'abord obtenus par in-
transsonique dtfinie prtctdemment par l'auteur pour le calcul troduction de l'algorithme "Semi-inverse" de Le Balleur [6][1]
des tcoulements attachts ou dtcollts sur les profils dailes, in- en 1978, algorithme gCnCral subsoniquelsupersoniquedot6 d'un
cluant le calcul du dtcrochage profond, est ici ttendue en une controle de stabilitb autonome, puis par l'algorithme "Semi-
nouvelle mtthode tridimensionnelle pour les tcoulements forte- inverse" simplifi6 de Carter [lo] de 1979, 6quivalent B une
ment dtcollts sur des ailes, aux grandes portances et en fluide intkgrale premikre de la partie subsonique du prCcUent, puis
compressible. par I'algorithme "Quasi-simultaneous" de Veldman [l 13 de
La technique numtrique non-lintairement implicite de couche 1981, algorithme subsonique de type Gauss-Seidel, puis enfin
limite (directelinverse), les modkles turbulents, le gtntrateur par l'algorithme instationnaire-consistant "Semi-implicite", Le
de maillage et son auto-adaptation, les algarithmes de Balleur, Girodroux [12][13], de 1984.
couplage et de mise en tquilibre du sillage, le schtma non- Plus rkcemment, I'algorithme "Semi-inverse" originel de
visqueux potentiel-complet, sont ttendus au tridimensionnel, 1978 a conduit 21 2 nouveaux algorithmes qui le gtnCralisent,
avec une approximation sur les tquations. visqueuses (2.750- Le Balleur [l]. Le premier est I'algorithme "Semi-inverse
local). Des rtsultats thtoriques nouveaux sont donnts sur les Massive-separation", algorithme stationnaire ayant ouvert
singularitts et canes caracttristiques de la couche limite l'accks aux calculs de dCcollement massif et de dkcrochage.
26-2

Cet algorithme est ici repris et Ctendu aux calculs 3D- "pseudo-fluide-parfait".
approchts ("2.75D"). Le second est l'algorithme "Semi-inverse
Pseudo-unsteady", obtenu en modifiant le prikonditionnement
non-visqueux de l'algorithme originel par introduction du
caracttre pseudo-instationnaire des solveurs Euler, ce qui a 2.1.1. Theorie complete : Navier-Stokes
donne accts ti des mtthodes robustes et rapides pour
l'interaction choc - couche limite [l]. La thtorie de "Formulation-DCficitaire"de Le Balleur pro-
pose une resolution indirecte des kquations de Navier en
L'accts au calcul du dtcollement tridimensionnel, plus traitant d'une part la difftrence entre Quations d'Euler et de
complexe ti obtenir, a souvent ttC dtcrit ti tort comme hors de Navier-Stokes, et d'autre part les Quations d'Euler, par des
portCe de la mCthodologie d'interaction visqueux-non visqueux. schtmas numeriques diffkrents. Le systtme numkrique obtenu,
Bien que, de par sa complexitC, la thtorie 3D du couplage n'ait de rang double, vise l'obtention d'une technique ti trts faible
pu Ctre directement dCgagte, les Ctudes exploratoires, par ex- dissipation numCrique, aussi bien pour des Quations de Navier
emple de Wigton, Yoshihara, Wai [29][30][31], Carter, Ed- complttes, que pour des kquations de couche-mince. Notons
wards, Davis, Hafez [26][271, Cebeci et al. [28], Steger, van que la thtorie de "Formulation-DCficitaire"ne dCcompose pas
Dalsem [32], ont d'abord mis en Cvidence des possibilitts de le vecteur vitesse en introduisant des vitesses-dtficitaires
calcul de petits dkollements 3D. (U' - i i i ) , mais dtkompose au contraire les Quations de Navier
PrCsentement, la mtthode transsonique 2D (ou 2.5D) sous forme additive par rapport aux flux. La thCorie introduit
suggCrCe antCrieurement pour les Ccoulements de profils d'ailes un double jeu de variables primitives, pour le champ-reel et
dCcollCs ou en dkrochage, Le Balleur [7][1], est ici Ctendue au pour le champ-fictif Euler, avec couplage fort (couplage exact)
calcul tridimensionnel sur des voilures dCcollCes, en intro- des champs Navier et Euler :
duisant des Cquations visqueuses approchtes ("2.75D"-local)
mais avec un calcul non-visqueux complet d'tquation du po-
tentiel. Les mtthodes originales mises au point dans la
mtthode bidimensionnelle, Le Balleur [l], notamment la tech-
nique numdrique de couche limite directehnverse, ainsi que sa
modClisation, l'algorithme de couplage "Semi-inverse" et sa
theone de stabilitk, le gCnCrateur algbbrique de maillage et son
auto-adaptation aux effets visqueux, l'algorithme d'kquilibrage
du sillage, le solveur potentiel-complet, sont ici gCntralisCs en Ruide Reel : ,Ti, B,p. LT
i . j , k = 1,2,3
3D. Les premiers rtsultats pdsentts dtmontrent l'acchs de la Pseudo-fluide-parfait : U ' , p , p. hT

[s][d.t
mCthode au calcul du dkcollement tridimensionnel, avec large
dkollement. Notons qu'une extension instationnaire menCe,
paralltlement sur la mCme mkthodologie visqueuse, Le Balleur, gij = ax' ax' G U ,
-- J = det = (gij)]',
Girodroux [XI, avec un choix oppod d'Quations visqueuses axi axj

3D-complttes et de calcul non-visqueux approcht (petites per-

turbations), dCmontre aussi l'accks au calcul du dCcollement
tridimensionnel.

2. EQUATIONS. -
axi -
axk = G,!,
. G,!= GM = G m = Kronecker.
axk ax]
Les Quations de Navier-Stokes sont dissocikes en 2 La thCorie Ccrit le couplage exact par raccordement continu des
systtmes couplCs Quivalents en utilisant la thtorie de 2 champs ("matching") lorsque l'on s'Cloigne vers I'extCrieur
"Formulation-DCficitaire"de Le Balleur [4][7][141111, proposke des couches visqueuses, ce qui Cvite toute ddpendance de la
vers 1980, et complttte vers 1989 par l'introduction du solution ti un choix arbitraire d'epaisseur visqueuse 6,et ce qui
"RtfCrentiel de DCplacement" [l] qui permet de gCnCraliser s'est avtrC aussi, Le Balleur [5][8], tliminer toute anomalie de
l'emploi d'6quations de couche-mince au decollement massif. couche visqueuse "supercritique" au sens de Crocco-Lees :
La thCorie est valable aussi bien pour les kquations complttes
de Navier-Stokes que pour des Quations de "couche-mince".

2.1. Thhrie de "Formulation-D6ficitaire".

Ce couplage apporte les conditions aux limites qui assurent la
On note avec un signe "barre" l'koulement visqueux tur- dktermination du champ fictif "pseudo-fluide-parfait", avec uni-
bulent del, assujetti aux equations de Navier-Stokes cit6 dans le cas non-visqueux potentiel, et avec la nCcessitC de
moyenntes. On note sans ce signe un Ccoulement fictif super- conditions supplCmentaires dans le cas non-visqueux Euler,
posC "pseudo-fluide-parfait", assujetti aux Cquations d'Euler, voir Le Balleur [3][231[19][20].
dCfini partout dans l'espace, m&me A I'intCneur des dgions
visqueuses. On considtre des coordonntes cartesiennes X i et
curvilignes X ' ( X ' ) , les composantes carttsiennes T
i et contra-
variantes i i J de la vitesse, le jacobien J = det @Xilaxi), les 2.1.2. Theorie "NS-couche-mince".
symtmles de Christoffel r!k,.?e tenseur mktrique giJ ou g i j , les
composantes cartksiennes T'J et contravariantes t'J du tenseur De mCme que dans une rksolution "directe Navier-Stokes
des contraintes, la densit6 p, l'enthalpie totale LT. la pression couche-mince" habituelle, la direction x 3 peut ici Ctre
totale pT, ainsi que les variables correspondantes dans le particulariste, et les termes de contraintes tronquts :
 26-3

+rikJ.[ 0
pu'uk - pii'ak
] + J giJ-%[
ax) p -p
] = d[
ax3 - J t i 3
](2a)

i , j, k = 1.2. 3
Les Cquations "Navier-Stokes couche-mince" sont dCpendantes
du choix du maillage, cette incertitude (majeure) &ant acceptCe
dans les mCthodes de rCsolution directes "Navier-Stokes". i , j, k = 1, 2 (3a)

2.1.3. "NS-couche-mince" en "Referentiel de Deplacement".

La thCorie de "Formulation-DCficitaire" Climine cette

dCpendance abusive en dCterminant aCrodynamiquement la
direction x3 par la sClection d'un "RCfCrentiel de Dtplacement",
Le Balleur [14][1], voir Fig. 1.

a
- 4 3 )
1
E = - ax3
P4 a ( U i -E')
ax - (X1. x2. 6)

Fig. I . "Rifirentiel de Diplacement" (dicollement massif).

En thCorie de "Formulation-DCficitaire", l'obtention des

Nuations de couche-mince par dCveloppement asymptotique
nCcessite en effet que non seulement la vitesse normale
visqueuse ii3 soit petite, mais aussi que la vitesse normale
non-visqueuse fictive superposke u3 soit elle-mEme petite, ce
qui est r6alisC par choix du "RCfCrentiel de Deplacement".
Dans ce rCfCrentie1, qui &end la validit6 des Quations de
couche-mince au calcul des dCcollements-massifs, la direction
1'
x3 devient sensiblement normale B la surface de dtplacement
(inconnue avant calcul), x 3 = 0 Ctant alors la surface de
dkplacement et non-plus la paroi. De plus il est possible sans
restriction majeure, dans ce rCfCrentie1, de tronquer encore les
Cquations de couche-mince au niveau de 1'Quation de mouve-
ment en x3. Par dCveloppement en 6 (jauge d'kpaisseur de _a --cosy'. 1a , _a --cos$.- a
couche visqueuse), il vient au premier or& : as1 ax asz axz

d . [pu'
axi
' - pii'
puiuJ - pii'nJ
' ] En plus des Cquations de masse et de mouvement, le
systbme (3) Ccrit ici, choisi de rang-4, inclut l'huation
d'entrainement, qui est une collocation le long de la surface
x3 = 6,,,x3 avec les huations locales de mouvement du
+rL.J.[
pu'uk
0
-p'pk
] =d[]
aX3 - Jti3
(26) s y s t h e (2) selon X I ou x2. Contrairement B de nombreuses
affirmations, les Cquations intkgrales wnt w i n s resrricrives que
celles de Prandtl, parce que p, &, p , p. u J ,hT, pT sont ici
supposCs Ctre Cvolutifs selon x3 au sein de la couche visqueuse,
Le Balleur [4][9][22], cette extension ne rksultant que du choix
optimal de dCfinition des Cpaisseurs intkgrales ("Formulation-
DCficitaire" par rapport aux flux). Notons que seules les
Quations intkgrales de mouvement (et donc d'entrainement)
2.1.4. Forme integrale "Referentiel 6
' " (NS-couche-mince). sont ici approchCes en couche-mince. L'Quation de continuit6
de (3a)(3b) est une indgrale exacre, valable aussi pour les
Une simple integration en x3 de (2b) entre la paroi Quatiom de Navier-Stokes complbtes, Le Balleur [4][9][22].
x3 = Z, ( X I , x2) et I'infini fournit les huations inttgrales, Elle dCtermine l'exact effer de diplacement gtntralist, traduit
pleinement huivalentes. Notant q le module de la vitesse par la vitesse normale non-visqueuse B la paroi w,,'ou bien par
non-visqueuse, et f les cotes supCrieur/infCrieur en cas de un saut de vitesse normale non-visqueuse w,,>sur les nappes
nappe de sillage : de sillage. Pour simplifier, on se limitera dans la suite au cas :
 e
26-4

cos$=1, c o s v ~ = ~ s W 1+]
=[ 1’
2 1 On choisit ici de conserver les pleines Cquations 3D pour
la partie non-visqueuse, et d’introduire une troncature
visqueuse 3D-approchCe, qualifike de troncature visqueuse
“2.75D“. En chaque noeud de calcul visqueux, on introduit une
ce qui correspondrait en 3D aux cas oil 1’Ccart entre surface de direction tangentielle A la couche visqueuse h’ dans laquelle
dkplacement et paroi dCpendrait peu de xz ou s2. les dCrivCes sont tronquCes. Cette troncature n’a qu’un faible
Enfin, I’intCgration en x3 de la relation (2c) foumit poids sur les Cquations discrCtisCes, etant donnC I’aliasing qui
(p - ji)’. en ( X I , x z , 2,) ainsi que des sauts de pression serait de toute manitre introduit par I’allongement des mailles
non-visqueux correctifs sur les nappes de sillages (effet de et par une discrCtisation en s 2 insuffisante ii 1’Cchelle visqueuse
courbure gtntralist). Celui-ci, en cas d’approximation de
couche-mince, peut Ctre, soit n6gligC au premier ordre, soit
modClisC comme un terme d’ordre suptrieur. Dans cette
cos X* . a
as1
+ sin X* a -- 0
.-
asz
(3c )
seconde hypothtse, on peut en particulier adopter la
modClisation de “Courbure-induite” suggCrCe vers 1980, Le Cette troncature visqueuse “2.758” Ctant appliquCe de manikre
“locale“, ce qui la rend similaire dans ses effets ii une tronca-
Balleur [4][7][14][1], moins restrictive que la thtorie de couche
limite, et en meilleur accord avec les thdories asymptotiques de ture numtrique par maillage grossier, tous les paramttres
bord de fuite donnCes par Melnik [2][17][18]. calcults, visqueux et non-visqueux, restent totalement Cvolutifs
dans la direction h’ et donc parfaitement tridimensionnels.
Suivant le choix de h*, la direction locale de troncature de
d6rivte peut aller de la simple direction s z dtfinie par le mail-
lage (sin h’ = I), ii toute autre direction locale
2.2. Choix d’une “rndtrique de surface locale“. akrodynamiquement remarquable, par exemple perpendiculaire
aux lignes de courant non-visqueuses.
Les Quations curvilignes ci-dessus sont gCndrales et
s’appliquent dans tout systtme de coordonnkes curviligne ou
orthogonal. L’expCrience a cependant montd qu’il est avan-
3. MODELES TURBULENTS 3D.
tageux, contrairement aux usages habituels en couche limite, de
sClectionner une ”mCtrique de surface locale” simple sur la
paroi (SI, s’) (ou le long de la surface de dtplacement XI,x2), Le nombre d’6quations intCgrales ci-dessus donne un
pour les calculs de couche visqueuse. systtme fermC pour le calcul du champ local visqueux lorsque
les profils de vitesse moyenne ainsi que les profils de tur-
Utilisant la trace ii la paroi du maillage non-visqueux bulence sont supposCs relever d’une modClisation 3
i , j , k , les coordonnCes de surface locales sont ici dCfinies en paramkms (6, a ’ , a’). Avec un modtle de turbulence ii N-
choisissant pour coordonnCe s I I’abscisse curviligne le long des Quations de transport, qui augmente de N le rang du systtme
Ligces (j = CSL, k = l), et pour coordonnCe s’ l’abscisse curvi- (3) ainsi que le nombre de variables-indkpendantes, ce rang-3
ligne dans la direction orthogonale ii S I tracCe sur la surface. concerne seulement 1’Ccoulement moyen. Les paramttres
Ce choix de mCtrique, avantageux pour la simplification ma- (6, a’, a’) sont fonctions de ( X I , x’).
jeure des &pations (rjk= 0 ), apporte de surcroit des avantages
de prkcision numtrique aprts discrCtisation des Quations. La
contre-partie de ces avantages est Cvidemment que les lignes 3.1. Profils de vitesse paramktriques.
(i = csr, k = 1) ne sont plus des lignes S I = cst.
On utilise la description analytique paramktrique originale
2.3. Troncature visqueuse “3D-approchke”(2.75D). de rang-3 des profils de vitesse moyenne turbulents sugg6rte
pdc&lemment, Le Balleur [7][14][9][1], d’abord en 2D puis en
3D. La modtlisation a C t t construite pour inclure tant les
Dans de nombreux calculs visqueux tridimensionnels, il Ccoulements attaches que massivement dBcollCs, voir [ 141. Elle
reste strictement hors de portCe d’utiliser des maillages de est ici appliquCe sous sa forme la plus compltte [l], avec
discr6tisation offrant la mCme densit6 de noeuds dans les 2 sous-couche laminaire et terme D d’amortissement, Fig. 2 :
directions tangentielles (s I , s’) aux couches visqueuses.
U‘ loi de
Pour le probltme discrCtisC, le grand allongement
inCvitable des mailles (en envergure par exemple sur un calcul sillage
de voilure), mCme s’il est compatible avec la tridimensionnalitk
du fluide non-visqueux, conduit en gCnCral, pour le problbme
numkrique visqueux, ii rendre illusoire I’accts ii la tridimen-
sionnalitt compltte. Les Cquations visqueuses 3D-compBtes,
une fois discr6tisCes sur des mailles ii fort allongement,
dintroduisent numCriquement en effet des troncatures non-
conmlCes, voire des distorsions d’aliasing et de dissipation
numCrique.
Pour ces configurations de calcul 3D-visqueux h fort al-
longements de mailles, c’est ii dire pour tous les cas oil la tridi-
mensionnalitC du maillage et de la discretisation ne peut suivre
la tridimensionnalitk des Cquations analytiques, il parait plus
avantageux de controler analytiquement les mncatures de tridi-
mensionnalitC visqueuse intvitables. La m6thodologie
Z* = Fonction de [E] , k = 0.41 .
incompressible
d’interaction visqueux-non visqueux offre ii nouveau un avan-
Les paramttres de forme independants a ’ choisis peuvent par
tage, en pouvant dCcoupler les troncatures numkriques
exemple Etre ceux du reptre orthogonal tangent aux lignes de
visqueuses et non-visqueuses.
courant non-visqueuses.
 26-5

.I - 0.05 , 0.00 modClisC B partir du modble de profils de vitesse turbulents ci-

1.8

rddh - IO 000. dessus et d’un modtle de longueur de mClange, Le Balleur

. [7][9][ 14][1]. L’expression finale de ce modble original

d’entrainement d’Cquilibre E,, est donnCe dans [l]. I1 a permis
d’Ctendre la modClisation de I’entrainement aux Ccoulements
2D ou 3D dCcollCs, mCme massivement, sans faire appel 21
I
I’empirisme.
On utilise ici I’ajustement le plus rCcent suggCrC pour les
constantes ainsi que pour I’effet de nombre de Reynolds [l].

1.
e. J .a

La direction XI est ici celle des lignes de courant non-

visqueuses. a i . Cfo correspondent aux valeurs de a ’ , Cf
calculCes pour la plaque plane en incompressible, au mCme R s
et avec un parambre de Clauser Go. Les deux seules donnCes
du modtle sont la constante C.,, qui est proportionnelle au
c a d du niveau de longueur de mClange dans les couches lim-
ites d’kquilibre, et la pseudo-constante Go, paramttre de
Clauser de plaque plane (qui peut Cue fonction de Rg).

Fig. 2 . Profils de vitesse pararnetriques 3D-turbulents,

discrerises h chaque station (a’ = .05-.90, a 2 = -.15, R b = lo4) 3.3. ModiYe de turbulence 24auations k-tau.

Les expressions de T i ,W’, Cf’,D , z’ sont dCtaillCes dans Les effets de turbulence hors-Cquilibre sont calcules en
[1][14]. Le modtle reprend I’idCe de forme composite additive ajoutant le modble B 2 Cquations de transport de Le Balleur
(loi de paroi - loi de sillage) proposCe par Coles [16] pour les [7][1], prksentement utilis6 en thCorie 2D (par Ranches) le long
Ccoulements attach& mais la forme a C t t ici gtnCralisCe aux de XI. Le modtle peut Ctre utilis6 aussi bien dans les
Ccoulements dCcollCs, et les fonctions de sillage F ’ , F sont mCthodes-locales (calcul de t g ) , et dans les mCthodes-inttgrales
diffkrentes. La simple loi de sillage de Coles serait totalement (entrainement et dissipation).
irrCaliste en Ccoulement massivement dCcollC. La modClisation Ce modtle suggtr6 en 1981 a pour premitre originalit6 de
z’(S’/S) est dCduite de l’analyse des singularitCs de couche “forcer“ le modble de Launder-Rodi par la modClisation ci-
limite au dCcollement, Le Balleur [5], de I’analyse des c8nes dessus des profils de vitesse, ce qui notamment le rend mieux
caractCristiques des systbmes d’Cquations et de la rCcupCration adapt6 aux Ccoulements dCcollCs. Sa seconde originalit6 est
de “l’influence amont” en Ccoulement dCcoll6 [14][5][8], ainsi d’effectuer un calcul en termes d’Ccart aux valeurs donnCes par
qu’enfin de la capacitC B dCcrire le cas-limite des couches de le modtle d’Cquilibre (contraintes ti: et entrainement Eeq). Cet
mClanges 2D isobares, Le Balleur [ 141. Ccart B I’Cquilibre est supposC Ctre invariant en direction x3,
Le vecteur unitaire T’ est sensiblement parallble au frotte- variable en directions XI,x2, et Ctre calculC 21 partir de 2
ment Cf’,et Cf est dCduit d’une modClisation de loi de paroi Cquations (intkgrales) de transport. Celles-ci calculent des effets
“universelle” projetCe sur T’ (le profil n’est pas une courbe d’histoire pour 1’Cnergie cinttique turbulente moyennCe en x 3
plane dans la region de paroi), voir [ 13. notCe R(xl, x 2 ) , et pour la contrainte de cisaillement de Rey-
nolds moyennCe en x3 notCe t(xl, x 2 ) :

Contrainte locale : tI3 = z (2,

XZ) . [t,$ (XI, x*, x3)
2.7
[y]’ [Log[Rs[?]’] + 5.25k - a] = k . Ilii,,,II (46)
t,, (XI,

La compressibilitd ne modifie pas les profils de vitesse. Les

profils de densite sont ici dCduits de la vitesse dans
‘t
I’approximation adiabatique isenthalpique ET = hT. Entrainement global : E = - E eq
L’entrainement n’est pas modifiC. L’effet de compressibilitC re- ‘te,

tenu pour le frottement et la dissipation globale s’Ccrit : ‘t

Dissipation globale : @ = - @cq
_I - te,

cfjoqremib/e = cf’ [I + 0.5 [ ~ I ] M z ]

Pour les rCgions laminaires enfin, les relations de fermeture at

ai =
8% + [8 - 8,] -
ax’ ,t
5 3
@
,, - E8
sont dauites des solutions auto-semblables, Le Balleur [7]. r 7

3.2. Modele de turbulence d’equilibre.

Lorsque l’hypothtse de turbulence d’kquilibre est

suffisante, l’entrainement exact E dCfini par (3a) a pu Ctre
26-6

i = 1, 6, j = 1 , 8 , (5a)

6 = [ l - z * ] 6, Z’G
[
max 0 , 2.406 a’- 0.5844
1
[ 11 , bi= , bh=
8, = 6 [ l - A * ( l - i i , , ! ) ] , A * = F ’ ( z ) d z =0.45

OeqEC f 1. ii; + 0.018 [1 - ii, 113

Dans ce modtle, 6, ii; reprksentent les Cchelles de longueur et
0

0
de vitesse effectives de la couche turbulente, telles
qu’analyskes par la modelisation ci-dessus des profils de vi- siny=si),,Iq, k=l,2
tesse moyenne, en Cvitant I’empirisme d’un modtle de type
Baldwin-Lomax. I(x316) et v, sont une longueur de melange et Rappelons que la vitesse normale B la paroi visqueuse est not6e
V,,. La vitesse normale B la paroi non-visqueuse est ici notee
une viscositt turbulente, donnCes par des modtles d’kquilibre.
+i’,,. Les termes b i sont des termes Cventuels de m6mque non-
XI, hz, h3 sont des coefficients caractbrisant les divers effets de
turbulence d’kquilibre, voir Le Balleur [7]. 11s sont proches de orthogonale. La pdsente mCthode numkrique hybride
1 pour les couches limites habituelles. hl est proche de 2 pour intCgraleAocale prend pour inconnues indkpendantes les gran-
deurs non-visqueuses de paroi, vitesse 9 ou nombre de Mach
les sillages habituels, & et S caractkrisant les effets de cour-
ruuit = .5 (?I) M’, vitesse normale la paroi +i’,,/q
bure et de turbulence extCrieure du modtle, prCcisCs dans [7].
(dCplacement), angle 6r entre vitesse et axe-s1 sur la surface
(SI, s’), pression totale PT, et enthalpie totale f i T , plus les in-
connues indkpendantes purement visqueuses, Cpaisseur de
4. METHODE NUMERIQUE VISQUEUSE. ’
couche visqueuse 6, paramttres de forme a = (6’/6)i~omprcssiblr,
et les paramttres de turbulence k , 2.
a’ = (6’/6))incOwrlssible,
La mCthode numtrique de couche visqueuse proposCe est Compte-tenu de la troncature 3D-approch6, les dCrivCes en
de nature hybride entre mkthodes-intkgrales et mCthodes- sz sont lites aux dCrivCes en S I par la relation (3c). I1 vient :
locales. Les grandeurs (vitesse, densitC,..), mCme modClisCes,
sont en effet discr6tisCes B chaque station selon la normale
avec un maillage auto-adaptatif. La mCthode numCrique
discrbtise donc des profils paramktriques selon x 3 , et un
systtme d’auations inttgrkes en x 3 dans l’espace (x’, x’). Les
schCrnas de discretisation, variables avec la densit6 locale du i = 1, 6 , j = 1, 8, (56)
maillage en X I , ont la capacitC d’&trenon-IinCairement impli- La discrCtisation en (SI, s’) de la prksente mCthode 3D est
cites en ( x ’ , x 3 ) . Ces schCmas sont x‘-marching, I’absence de la dCduite de la discretisation robuste en (SI, x 3 ) dCveloppCe
capacitt multi-marching “MZM” [24][25] Ctant ici compensCe depuis de nombreuses anntes, Le Balleur [7][9][1],
par la possibilitC de calcul en modes “direct “ et “inverse “. s ‘-marching et non-IinCairement implicite. Les schCmas
numCriques en direction s 1 sont de second-ordre et 31 2-noeuds
4.1. Methode hybride integrale/losale. de type “trapbzes-amont”. Les schCmas de second-ordre de
methode des trapkzes en direction S I sont cornmuds vers un
schCma plus robuste de premier-ordre dCcentr6-amont en SI, 18
A chaque station ( x ’ , x ’ ) , numkriquement, les profils de
oh le maillage devient grossier (grandes valeurs de As ‘16). Pour
vitesse paramCtriques 3D-turbulents des relations (4) sont
discrttisCs selon x 3 , entre x 3 = 2, et x 3 = 6, avec un maillage des valeurs plus Clevtes de AsS’/6,un sous-maillage en s 1 est
en x 3 / 6 qui est auto-adaptatif il 6 et B la forme du profil de vi- ajoutC (zones non-dCcolltes). Une technique de Newton est
tesse, caractCrisCe par le paramttre a’. La discrktisation auto- enfin introduite pour rksoudre ittrativement les pleines non-
adaptative utilise prksentement 37 noeuds, optimists en direc- IinCaritCs induites par Aj” , Aj2’ , bf , en utilisant une estimation
tion x 3 et dCpendants de a ’ , avec des schCmas implicites de approchCe numtrique des Jacobiens a A j i / a f , aAj2’1,3f, a b $ a f ,
traphzes pour lier les grandeurs locales et intkgrales. La densite avec f - (6, a’, a’), ce qui rend la discdtisation non-
est deduite de la vitesse par I’approximation adiabatique IinCairement implicite en (SI, x3).
isenthalpique KT = h T . Les equations et schCmas ci-dessus fournissent tgalement
De station en station dans l’espace ( x ’ , x’), le systtme une rksolution numCrique des sillages visqueux, avec leur
d’equations intCgrCes en x 3 , valable aussi dans le cas d’une pleine dissymCtrie, moyennant doublement du rang du systtme
pure mtthode-locale, est discrCtisC sur le maillage curviligne (rCunion des demi-sillages supkrieur et infkrieur, not& f de
( X I , x’) de surface de dkplacement, projection selon x 3 du mail- part et d’autre de la surface des minima de vitesse), Le Balleur
lage curviligne de paroi (SI, s’). Notons que les 2 maillages [7]. Ce calcul dissymCmque complet est ici effectuC en 2D. Le
liCs, de paroi (SI, s’) et de surface de dkplacement ( x ’ , x’), se traitement simplifi6 de la seule partie symCtrique des Cquations
diffkrencient en cas de dtcollement massif, et que I’intCgration de sillage [7], qui ramtne le calcul de sillage B un systtme de
en x 3 (normale B la couche visqueuse) n’est alors plus normale mCme rang que la couche limite, et qui est obtenu dans les re-
B la paroi, Le Balleur [1][14]. Les Quations en (XI, x’) lations (5a)(5b) avec C f = 0, F,, = 0, a: = a!.,
projetCes sur la paroi (SI, s’) s’Ccrivent : f j = .5 (fi + f!), est ici utilis6 en 3D.
 26-7

Enfin, comme des itkrations additionnelles sur le solveur 4.3. Singularit& 3D. CBnes caractbristiques (DirecVInverse).
numkrique visqueux sont nkcessaires pour le couplage, le
problkme non-linkaire visqueux complet (5a)(5b) n'est pas Les systkmes 3D-complet (5a)(5c) sont hyperboliques en
n%lU 2 chaque itbration de CouPlage, et des solutions espace, en mode direct et en mode inverse, et des domaines de
linkariskes, avec les mamces Aili ,A Y gelkeS a leurs valeurs dkpendance, dkfinis par des c6nes caractkristiques dans I'espace
l'itkration de couplage prkckdente, sont daliskes 21 la plupart (SI, sZ), doivent Ctre respectks par les mkthodes de marche en
des itkrations de couplage, B titre de technique d'accklkration espace, aussi bien en mode direct qu'en mode inverse.
de convergence.
Les systkmes 3D-approchks direcvinverse issus de
(5b)(5c), qui peuvent s'kcrire toujours sous la forme (5b), et
4.2. Modes Direct / Inverse. qui n'introduisent que la dkrivke en s', sont singuliers lorsque :

Les cinq inconnues non-visqueuses du problkme

numkrique visqueux (5a)(5b), notkes avec un signe chapeau,
(rii , an/q, kT , P T , &), doivent Ctre couplkes au quantitks En mode direct aussi bien qu'en mode inverse, I'examen de la
correspondantes du problkme non-visqueux, qui sont notkes relation (5d) montre que la condition de singularit6 du systkme
( m , W./q. hT,PT. a). 3D-approchk (5b)(5c) est smctement identique ? celle
i de re-
cherche d'une direction caractkristique h' du systkme 3D-
Avant de rkaliser %e couplage complet sur les 5 inconnues
complet (5a)(5c). Cette identitk entre kquations aux valeurs
non-visqueuses, ce qui est obtenu itkrativement par
propres pour les singularitks du systkme 3D-approch6, et pour
I'algorithme de couplage, des calculs visqueux dkcouplks sont
le c6ne caractkristique du systtme 3D-complet, dvkle que le
rkalisks, dans lesquels 4 inconnues non-visqueuses seulement
systkme 3D-approchk n'introduit aucune perte d'information au
sont couplkes. On impose alors au systkme (5a)(5b) l'enthalpie
niveau des domaines de dkpendance par rapport au systkme
totale (iT = hT), la pression totale GT = p ~ ) et
, la direction de
3D-complet, ces domaines de dkpendance ktant identiques. I1
la vitesse (& = a),plus une condition additionnelle B dCfinir. rkvkle d'autre part que le franchissement des singularitks du
Une commutation entre diffkrentes conditions addition- systtme 3D-approchk (5b)(5c) en mode direct s'identifie B la
nelles doit Ctre introduite (modes direct-inverses) afin de sur- naissance d'une influence-amont selon x' (pour la direction de
monter le problkme de singularitk de dkcollement, comme c'est troncature h' considkrke), et par conskquent B la nkcessitk de
dCjB le cas en 2D, Le Balleur [ 5 ] . En tridimensionnel, le commuter le calcul x'-marching vers le mode inverse.
problkme de singularitk s'identifie avec le calcul par marche en La modklisation issue des profils de vitesse (4a) peut Ctre
espace, et avec le respect des domaines de dkpendance ualiske pour analyser les singularitks et directions
imposks par les canes caractkristiques des systkmes hyper- caractkristiques, comme effectut B I'origine en 2D et 2D-
boliques (5a)(5c) ou (5b)(5c) (voir ci-dessous). En tcoulement instationnaire, Le Balleur [ 5 ] . Le calcul analytique des direc-
attachk, le systkme visqueux en mode "direct" peut Ctre rksolu tions caractkristiques a ktk :ffectuk avec le modkle (4a)
par marche en espace, mais selon les lignes de courant non- simplifiC par Cf = 0. D = 0, z = 0, comme deja utilisk en 2D
visqueuses. La condition additionnelle est alors le nombre de dans [5]. La comparaison du c6ne caractknstique en mode
Mach (rii = m ) . Le maintien d'une rksolution x'-marching , direct, qui s'kcrit en repkre orthogonal tangent aux lignes de
dkcollement inclus, demande qu'un mode "inverse" soit com- courant non-visqueuses comme l'enveloppe des directions :
mutk avant que le c6ne de dkpendance du mode direct
n'interdise ce mode (et donc bien avant le cas limite de
dkcollement-2D). La condition additionnelle imposke est alors
la vitesse de transpiration non-visqueuse (G,,/q = w , / q ) . MalgrC
la rksolution x '-marching, l'influence amont d'origine
-d=
r 2[
dr'
--a2 - a2
0.4014-a' ' 0.6417-a' ' 1.599-0'
- a2
1
, 0 , (direct)

et du c6ne en mode inverse, qui s'kcrit dans le mkme repkre :

visqueuse (dkcollement) est pleinement recouvrke B 1'Ctape de "2
couplage, oii ]'on rtalise simultankment (iT = hT, pT = p T ,
rfi = m , $,, / q = w, / q , 6r = a).
Au total, B chaque noeud ( i , j , 1) et itkration de couplage avec 0 < a' < A, A = 0.6417, 2A = 1.283, l-(1-A)ln = 0.4014,
N , le systkme discrktisk (5a)(5b) est fermk par 4 relations (5c), l+(1-A)ln = 1.599, a montrk que le mode inverse dduisait
I
discrktiskes par des schCmas centres pour ( a d / a s ' ) et l'ouverture du c6ne caractkristique par rapport au c6ne obtenu
Cventuellement dkcentrks-aval pour (am /as ') : en mode direct dans les mCmes conditions (aux grands
paramktres de forme a ' ) , et que le demi-angle maximal du
I E = 0 : Direct cBne caractkristique dans le plan (SI, sz) en mode inverse ktait
&= a , LT = hT , PT = P T , E = 1 : Inverse bomk B 90", au lieu de 180" en mode direct, Le Balleur, Giro-
droux [21]. La borne 90" en mode inverse n'est atteinte de plus
que pour une valeur i n h i e du paramktre de forme longitudinal

[::
N
Hi (soit a' = A). Ceci autorise, en mode inverse, la marche en
:[
E---
:]i,j,I
+ (1-E) --
:]:I,
=o (SC)
espace en toutes circonstances, mais en direction des lignes de
courant non-visqueuses.

1 La commutation d'un mode direct (E = 0) vers un mode

inverse ( E = 1) est prksentement connectke aux paramttres de
De ce fait, la mkthode 3D-approchCe (5b) peut Ctre
amknagke pour kviter de franchir une singularit6 en mode in-
forme du profil de vitesse. Elle est like tant6t comme en bidi- verse, ce qui kquivaudrait B violer le domaine de dkpendance

'
mensionnel au seul paramktre en direction des lignes de
! courant non-visqueuses (a' > 0.28 par exemple), Le Balleur
imposk par le c6ne caratkristique. Pour ce faire, la direction de
troncature h' est ici progressivement rapprochke de la perpen-
[7][11, tant6t aux 2 paramktres longitudinaux et transversaux, diculaire aux lignes de courant non-visqueuses, au fur et B me-
de facon & ce que le c6ne caratkristique du mode "direct" es- sure de I'accroissement d'ouverture du c6ne caratkristique en
timk d'aprks a ' , a 2 (voir ci-dessous) reste, t~chaque station x' mode inverse, calcul6 en chaque noeud B partir de a ' , a' et des
oii E = 0,positionnt en amont de la direction de troncature 1'. relations ci-dessus.
 26-8

5. METHODE NUMERIQUE NON-VISQUEUSE. 6.1. Thhrie numerique du couplage. Stabilite.

La mCthode numkrique, nouvelle, est une extension en La thtorie numCrique proposCe pour le couplage de sous-
tridimensionnel de la mCthpde de traitement de I'Quation du domaines fournit une estimation de prCconditionnement ori-
potentiel-complet, mise au point et introduite pour le ginale pour les opkrateurs numkriques de surface, "visqueux" et
dkrochage 2D, Le Balleur [l]. "non-visqueux", qui se trouvent introduits sur la surface-
frontitre ( i . j . 1) des champs tridimensionnels (i, j , k) ?i
Elle discrbtise l'kquation du potentiel dCveloppCe en coupler. Dans le probltme discret, ces opCrateurs numCriques
coordonnCes curvilignes, sur un maillage structun? i. j , k quel-

i
de surface ( i . j , 1) sont ceux qui pourraient Ctre obtenus par
conque, en transsonique : une Climination des noeuds interieurs ( i , j , k) sur les equations
discrttes des champs visqueux et non-visqueux.
I
Les opkrateurs de prkonditionnement proposCs donnent
accts une thtorie de stabilitC applicable ?i tout algorithme de
couplage explicite, par exemple aux algorithmes "direct" et "in-
verse" de point fixe. I1 donnent accts en outre au calcul des
champs de coefficients de relaxation nkcessaires pour stabiliser
m, n, I , a = 1,2,3
ces algorithmes.
La discdtisation i , j , k est effectuCe avec la mCtrique x1 = i,
La thCorie a enfin permis la naissance d'algonthmes de
x' = j , x 3 = k. Elle utilise pour les dCrivCes du potentiel des
schCmas diffbrences-finies centrCs en subsonique, 21 7 noeuds
couplage originaux plus ClaborCs, comme l'algorithme "Semi- I
inverse", Le Balleur [SI[ 1][ 151. L'algorithme "Semi-inverse" et
plus schCmas de dtrivCes secondes croisCes, complCtCs par un
dtcentrement de type Murman-Cole aux noeuds supersoniques. ses gCnCralisations rksultent en effet d'un couplage exact (donc
L'Cvaluation des termes mCtriques est effectuCe sur la mCme implicite), sur le nombre d'onde maximal, des operateurs
molCcule de discrCtisation que celle du potentiel, et avec les numCriques de prkconditionnement.
m h e s schCmas pour les dCrivCes de X"' et de @, ce qui per- 6.1.1. Prkonditionnement : "Operateur non-visqueux" (2.75D).
met de se rapprocher des formulations volumes finis.
L'opCrateur numkrique de surface "non-visqueux" du
La rksolution du systtme discret est faite par relaxation- couplage, dont 1'Cvaluation par Climination des noeuds
ligne SLOR, avec traitement implicite des conditons limites sur intCrieurs est en gCnCral impraticable, est estimt ici par une ,
la paroi, ainsi que des coupures de sillage. Le traitement du strathgie de petites perturbations lintariskes analytique, avec I
champ lointain (Dirichlet) est un comportement de type tourbil- dCcornposition de Fourier sur la surface (i, j . 1) et filtrage
lon plus source, tvolutif en envergure, couplC B la circulation. nurnkrique par la discrhtisation.
La non-linCaritC du potentiel est traitCe par une mCthode de
point fixe, les composantes contravariantes U"' et la vitesse q La thkorie, proposke p&&Iemment en bidimensionnel
Ctant dCduites du potentiel au moyen de schCmas centres. puis en quasi-3D (2.5D), Le Balleur [6][3], est ici Ctendue en
"2.75D". Elle considtre des perturbations stationnaires isentro-
Les conditions aux limites sur la paroi et sur la nappe de piques du champ de vitesse. en module, vitesse normale, direc-
sillage sont controlCes par les effets visqueux de dkplacement tion, autour de la solution convergke finale q,(sl. s'),
,
I
et de courbure. Le calcul sur maillage en C est effectuC en pro-
w,,,(sl, s'), a,(s'. s') sur la surface ( i , j , 1) :
longeant I'aile par une coupure (non-portante), numkriquement
traitCe comme la nappe de sillage. Les conditions aux limites,
discrCtisCes en demi-maille dans la dirextion x 3 = k, s'Ccrivent
sur la paroi : Fj (XI. x,.xJ)
4 -40

=[w:::]

'@
1

d iplacemeni Elle suppose des Quations de perturbation potentielles de type

axm ' Prandtl-Glauert autour de la solution localement figte
Sur la nappe de sillage, les conditions de "saut" (notCes < >) : q,, w,,,,a,. Les coordonntes curvilignes sont ici choisies avec
_ -I X I en direction de la vitesse non-perturMe, la surface ( X I , xz)

< w,,> = [g33] 2

g3m c
a@ >, diplacement Ctant une surface de courant (surface de dkplacement), et la
direction x 3 Ctant normale B ( X I , x 2 ) :

z+3!l=
axmaxn ax3ax3
utilisent I'approximation habituelle des techniques numkriques m , n = 1.2
"potentiel-complet" sur I'estimation du terme cq >. La thhrie analyse les perturbations Cmanant de la surface
( x ' , x') avec dtcomposition de Fourier sur la surface et filtrage
par la discrCtisation (i. j , 1). Les perturbations, de nombres
d'onde notCs (a,, %, a3),ont alors un comportement tant6t ex-
6. ALGORITHMES DE COUPLAGE. ponentiel tant6t harrnonique dans la direction x 3 :
x' +a,.x 2 + a,.2 )
/.(a,. I
Les algorithmes font appel B une thCorie numtrique ori- $'=$. e , az = O (en 2.750)
ginale du couplage qui a pu Cue sugg6rCe prCcCdemrnent, Le
Balleur [6][7][1][15], et qui est ici gtnCralisCe. Celle-ci traite le
probltme oh des rCsolutions par domaines doivent Ctre
coupltes par leurs conditions aux limites sur une surface com-
mune (s I , s') discrCtisCe en (i , j , 1). La thCorie est aussi appli-
cable pour coupler 2 sous-domaines non-visqueux dans un pur 1
calcul en fluide-parfait [11[151. Sur la surface ( X I , x'), en dksignant respectivement par U. v les I
 26-9

projections de la vitesse sur I'axe x 1 et sur sa normale :

. a3),l'opkrateur
Finalement, pour tout nombre d'onde (U',a
'
non-visqueux "2.75D" de prtconditionnement de la thCorie La relation peut Cue Ccrite & volontC sous la sous forme
numtrique du couplage s'knt : diffbrentielle ci-dessus, ou sous sa forme discrCtisCe. Les non-
u13 - I .P.q',
- (p complexe, l 2 = -1) linCaritts visqueuses "raides" (ou singularitks) sont incluses
dans les 3 coefficients pi, qui dCpendent fortement de i , j .
Compte tenu des relations (5c) qui imposent un couplage direct
sur les 2 grandeurs &,aT, le seul coefficient local important
pour la thCorie numkrique de couplage est p.;

6.2. Algorithme "Semi-inverse Massive-separation'' (2.75D).

Dans le cas oh la surface de dkplacement ne s'Ccarte pas
uop de la paroi, ce qui peut inclure des dkollements d'ttendue A chaque iteration de couplage, la vitesse normale & la
quelconque mais tels que w,/q reste modCd (en pratique paroi non-visqueuse w," (ou bien la discontinuit6 de vitesse
non-supCrieur B 0.4 environ), la thCorie de prkconditionnement normale non-visqueuse sur les sillages *,">), qui traduit I'effet
peut nCgliger la difftrence entre d3et w',, de dkplacement gCnCralisC, est modifi6e en chaque noeud de
U13 = w', couplage ( i , j , 1) de manikre explicite grace & I'algorithme
"Semi-inverse" de Le Balleur [6][1][15], Ctendu ici en
Pour tout nombre d'onde (al,a2.a3),on a alors, sur la paroi : tridimensionnel-approcht ("2.75D") :

6.1.2. Prkonditionnement non-visqueux "Massive-separation". O<OC2 (8),

Notons que si I'algorithme n'est pas tridimensionnel,
Dans le cas d'un dCcollement massif, la direction x 3 (nor-
male & la vitesse non-visqueuse et donc & la surface de l'approximation "2.75D" dCfinie ici est plus complkte que
dCplacement) n'est plus normale B la paroi. La vitesse normale I'approche quasi-3D (2.5D) qui Ctait dCja prCcisCe [3][1].
de perturbation d,,dCpend alors de q', et I'opCrateur non- Les coefficients a,!,j , a?, j , a;.j , dans l'algorithme
visqueux correspondant, donnC dans Le Balleur [ 11, devient : "Semi-inverse" de Le Balleur, sont calculCs en chaque noeud
w',, = q' sin w + ue3 cos w , sin w = w,,/q ( i . j ) de couplage. Ce calcul est tel que wf" corresponde & la
solution (IinCarisCe) de couplage exact sur le nombre d'onde
u'3= I .P.q' maximal [6][7][1] du maillage. Une surrelaxation de 2 est
w',,= ~ . [ p c o s w - I s i n y ) d = 1 . r . q ' alors la limite de stabilitt [6][7]. Pour I'algorithme "Semi-
Pour tout nombre d'onde (al,a
. cg), on a alors, sur la paroi :
' inverse Massive-separation", les coefficients s'tcrivent, Le Bal-
leur [1][15], en variables complexes :
[ wn - wno Ii8 = I . P' . [ 4 - qo ]
j, i, j , I 1 (6c 1
(p' complexe, I' = -1)
De plus, le nombre d'onde maximal en direction x 1 ou s 1
compte tenu du filmage opCrC par le maillage de discrktisation
est modifit?, Le Balleur [ l ] :
p=- a1 . P; (p complexe, I' = -1)
p"
p'= p cos w - I sin w
Dans ces expressions, I'extension en "2.75D" de I'algorithme
6.1.3. Prkonditionnement : "Operateur visqueux" (2.750). "Semi-inverse Massive-separation'' suggCr6 en 1989 [I][ 151 est
Compte tenu de la formulation numtrique si-marching du contenue dans le terme P donnC par la relation (6a). Le terme
problkme visqueux, il a CtC montrC qu'il est possible d'y p correspond & l'amplification complexe de I'algorithme de
effectuer, B chaque station ( i . j ,I), une Climination des vari- point fixe "direct" (p-' Ctant de ce fait l'amplification complexe
ables internes "purement visqueuses", Le Balleur [51[61[71[11, de l'algorithme de point fixe "inverse"). On notera que la re-
ce qui dduit le problkme visqueux & une seule relation connec- laxation o*,o3 ci-dessus correspond B la forme originelle de
tant les variables non-visqueuses de paroi. Cette relation (par- l'algorithme "Semi-inverse", Le Balleur [6], c'est & dire & une
fois dite fonction d'influence visqueuse) constitue prCcisCment relaxation optimale a-' = I+-', mais que 1'Ccriture ici en vari-
l'opkrateur de surface visqueux recherchC de la thCorie able complexe de o inclut tgalement I'analyse de Wigton, Holt
numkrique du couplage. Cette Climination des variables pure- [19], voir [22]. On remarquera enfin que l'algorithme "Semi-
ment visqueuses, qui peut inclure si nCcessaire les schCmas de inverse Massive-separation" conduit, mCme en subsonique, & ce
discrCtisation & la station (i , j ) (mCthodes-locales), Climine ici que les 2 termes o2et o3 soient actifs et non-nuls, Le Balleur
6,U ' , u2, E , 3 dans le syst2me (5a)(5b) et donne : r 11r 151.
 26-10

6.3. Algorithme “Equilibrage de nappe de sillage”. transforme une dpartition de noeuds X ( i ) existante en une
kpartition nouvelle X * ( i ) qui introduit un nombre quelconque
On utilise I’algorithme d’kquilibrage de sillage suggCrC en de resserrements locaux, rkglables indkpendamment en inten-
1981, Le Balleur [7][1], pour les profils d’ailes. Cet algor- sitC, largeur, et forme, par superposition (additive ou multipli-
ithme, issu des techniques numkriques du couplage visqueux- cative) de fontions de resserrement ClCmentaires de classe C-
non visqueux, peut Etre utilis6 tout aussi bien en calcul non- (fonctions exponentielles, Gaussiennes par exemple). La mono-
visqueux. La mCthode numCrique peut traiter 1’Cquilibrage 3D, tonie de la fonction X ’ ( i ) est assurCe en traitant la transforma-
en appliquant l’algorithme 2D sur chaque surface j = csr . tion sur les dCrivtes Cix’ldi, Ctyldi, positives, le maillage final
Dans le cas visqueux, l’algorithme d’kquilibrage de sillage X ’ ( i ) Ctant restituC par integration numCrique. Ceci introduit en
s’ajoute B l’algorithme d’effet de dkplacement, lequel assure outre une source de lissage favorable. Ce module de maillage
seulement les conditions correctes de “saut” de vitesse normale peut de plus, dans la transformation X * ( i ) , placer 4 noeuds B
non-visqueuse <w,,/q> sur la nappe (“saut” qui devient nul en des abscisses fixes ou dCterminCes.
fluide-parfait) :
(wn )+ + (wfi )- = (*n )+ + (*n )- (9a 1
L’algorithme d’kquilibrage effectue d’abord, avec les condi-
tions correctes de saut de vitesse normale, un calcul sur une
gComCtrie de nappe approchCe Z , N i , j . I’itCration N. Aprks
convergence de I’algorithme de couplage dans le cas visqueux,
ce calcul donne les vitesses normales non-visqueuses relatives
B cette gComCtrie de nappe, de part et d’autre de celle-ci
(wn/ q )+N, (wn
L’Cquilibrage est obtenu lorsque non seulement la relation
(9a) est satisfaite, mais lorsque aussi (*,,/q)+ = (w,,/q)+N,et par
consequent (G,,/q)-= (w,,/q)! (avec *“/q = 0 dans le cas non-
visqueux). La quantitb :
[3-3] i.j. 1
-
-

[
112 (*n/q)+ - (wn/q)+N-(*n/q)- + ( w n / q ) ~i ,]j ,
fournit I’erreur angulaire de pente de la nappe B chaque noeud
de couplage. L’algorithme d’kquilibrage relaxe alors cette pente
en chaque noeud, la nouvelle gComCme de nappe &ant dCduite
de la pente par intCgration selon i depuis le bord de fuite :

O<O<l,

Cet algorithme d’equilibrage a des propriCtCs similaires B une

mCthode de Newton, avec convergence quadratique.

7. MAILLAGE AUTO-ADAPTATIF.

On utilise pour le champ non-visqueux une mkthode de

gCnCration algkbrique de maillage originale mise au point en
1981 pour les profils d’ailes, Le Balleur [7]. La mCthode est ici
Ctendue au tridimensionnel en appliquant la technique 2D plan
par plan (j= csr), avec une frontibre externe identique dans
chaque plan, Fig. 3.
La mCthode construit, Fig. 3-4-5, par intCgration
numkrique selon k , un maillage algCbrique en C ob les lignes
i = csr sont des paraboles, en Climinant toutefois la technique
des paraboles de transformation conforme, de facon B pouvoir
rkpartir les noeuds A volontC dt uniformkment sur la frontikre
externe. Les lignes k = cst sont obtenues par contr6le de la dis-
hibution des noeuds sur chaque ligne i = csr entre la paroi et
la frontikre externe, Fig. 3-4-5.
La seconde originalit6 de la methode est d’introduire une
technique de contr6le monodimensionnelle de resserrements
multiples du maillage. Ce module de gCnCration de maillage Fig. 3 . Mkthode de gCnCration de maillage adaptatif3D.
(NACA4412, A = 6, M = .18, a = 17O, Re = 4.2 lo6)
 26-1 1

Une troisikme originalit6 de la mCthode, introduite plus Massive-separation” pour le couplage, la mCthode bidimension-
rkcemment, est de gCnCraliser l’emploi de ce mailleur nelle stationnaire (codes VISOS et VIS07, Viscous-Inviscid-
algCbrique A des gComCtries plus complexes (quelconques) par Solver-07) a donnC accks au calcul du dCcrochage et du post-
adjonction d’une technique de superposition de transformations dCcrochage. Les Fig. 6-12-13 reproduisent ici les resultats [l]
analytiques. Ces transformations, par distorsion du maillage de obtenus sur le profil NACA4412 expCrimentC par Hastings
base et avec minimisation d’aplatissement des mailles aux [33], avec la version la plus compltte de la mCthode. Dans
points singuliers, permettent ici par exemple le maillage en C
avec bord de fuite CmoussC (ellipse p. ex.), Fig. 5. L’originalitC
de la technique est de superposer des transformations de distor-
sion (3D) monotones A supports born&. Des superpositions de
2 distorsions ClCmentaires, translation et rotation, suffisent A
obtenir un large champ d’application.
L’auto-adaptation du maillage non-visqueux prend en
compte la gComCtrie exacte de la nappe de sillage,
itkrativement dCterminCe par I’algorithme d’kquilibrage. Des
raffinements mobiles du maillage (selon i , et Cventuellement
selon k), sont introduits pour adapter la maille locale As’ A
I’Cchelle 6 d’interaction visqueuse dans les zones d’interaction
forte telles que dCcollements et pieds d’ondes de choc, Fig. 3-
4-5, et sont dCplacCs de manikre itCrative. A chaque iteration
d’auto-adaptation du maillage, une interpolation bilinCaire par
cellule de la solution est effectuCe.

Fig. 4 . Maillage auto-adaptatif :champ non-visqueux.

(NACA4412, M = .18, a = 24O, Re = 4.2 10‘)

Fig. 5 . Mtthode de gtntration de maillage algtbrique.

(Corps tmousst, maillage en C,
raffinements auto-adaptatifs h la viscositt.)

8. RESULTATS 2D.

8.1. Dkrochage profond.

Grace au double -progrks

- du “RCfCrentiel de DCplacement” Fig. 6. Histoire de dtcrochage. Lignes de courant et
pour les Cquations visqueuses, et de I’algorithme “Semi-inverse Dressions DariCtales. iNACA4412. M = .18. Re = 4.2 ~.
~I ~
10‘)
26-12

celle-ci, le sillage est calculi5 avec sa pleine dissymCtrie et la

nappe de sillage non-visqueuse est mise en Quilibre exacte-
ment (code VIS07), sur le lieu des minima de vitesse. Ce cal-
CUI du sillage avec positionnement exact donne accts B la
pleine restitution des champs locaux moyens et tiuctuants,
lignes de courant, iso-vitesses, iso-turbulence, Fig. 6-10-11. Le
calcul restitue bien, Fig. 6, le passage d’une topologie 2 1 tour-
billon de bord de fuite au CLmax vers une topologie
d’Ccoulement moyen 3 2 larges tourbillons contra-rotatifs en
dCcrochage. Avec le modtle k-z 3 2 6quations proposC pour
les Ccoulements dCcollCs, dts que la transition est correctement
modClisCe [l], un bon accord calcul-expCrience est observC sur
la courbe de portance jusqu’B son maximum, Fig. 13, ainsi que
sur la distribution de pression & portance maximale, Fig. 12.
Dans 1’Ctat actuel, les codes VISOS, VIS07, permettent des cal-
culs B maillage auto-adaptatif robustes et routiniers jusqu’8 des
incidences 28’ ou 30°, Fig. 10-11. Fig. 9. Maillages auto-adaptatifs complets
d’interaction visqueux-non visqueux.
(Ellipse, M = .18, a = 00, Re = 4.2 10‘)
8.2. Maillage auto-adaptatif.
l’auto-adaptation aux effets visqueux, Fig. 4-7-8, Fig. 5-9. I1
Dans 1’Cvaluation de la simulation numkrique par interac- offre donc un nombre de degds de libertk accru pour l’auto-
tion visqueux-non visqueux, I’attention doit Ctre portCe non adaptation du maillage, par rapport aux techniques de
seulement sur la modClisation et les algorithmes visqueux, mais rksolution directe. La Fig. 7 montre & convergence le maillage
encore sur les avantages d’auto-adaptation automatique du “visqueux“ auto-adaptatif B I’tpaisseur de couche visqueuse et
maillage qu’elle apporte par rapport aux techniques de au profil de vitesse dans la couche, qui est introduit ici en
dsolution directe d’kquations de Navier-Stokes. La technique thkorie de “Formulation-DCficitaire”, et par consQuent en
d’interaction visqueux-non visqueux inclut en effet en elle- recouvrement du maillage “non-visqueux” auto-adaptatif de la
mCme le problkme d’auto-adaptation du maillage qui est Fig. 4, avec coincidence des noeuds sur la paroi et sur la nappe
nCcessaire B grand nombre de Reynolds, et dont I’impact sur la de sillage. Les maillages complets du calcul, Fig. 8 et Fig. 9,
simulation numCrique B nombre de Reynolds ClevC est sans montrent la totale auto-adaptation de maillage obtenue dans la
doute bien supCrieur au dilemne du choix entre Cquations prCsente simulation numCrique par interaction visqueux-non
complttes (toujours tronquCes par le filtre des maillages) et visqueux, tant pour la capture des sillages, bords de fuite,
Quations de couche-mince. points de dCcollement, que pour celle des couches limites et
Le duoublement numkrique ”visqueux - non visqueux“ couches de mklange. Ce maillage explique la facult6 offerte par
avec recouvrement des domaines de calcul de la prksente tech- cette mCthodologie numbrique de pouvoir augmenter
nique apporte I’avantage d’un traitement dCdoubl6 lui-aussi de indkfiniment, et sans viscositC numkrique majeure, le nombre
de Reynolds accessible. Les calculs ont ici Ctk effectuCs sur des
maillages 257x32 (champ non-visqueux) plus 257x37 (champ
visqueux). Un doublement du maillage visqueux dans la direc-
tion normale, en fin de convergence, n’aurait pas augment6
\

Fig. 7. Maillage auto-adaptatif :champ visqueux.

(NACA4412, M = .18, a = 24O, Re = 4.2 10‘)

Fig. 10. Ecoulement moyen :profils de vitesse moyenne et

lignes iso-vitesse. (NACA4412, M = .18, a = 28O, Re = 4.2 10‘)

Fig. 8. Maillages auto-adaptatifs complets

d interaction visqueux-non visqueux. Fig. 1I . Turbulence : profils d‘tnergie cinttique et lignes iso. I
(NACA4412, M = .18, a = 2 4 O , Re = 4.2 10‘) (NACA4412, M = .18,a = 28O, Re = 4.2 10‘)
 26-13

6.

9.

4.
- calcul
o o o o expkrience (Hastings)
3.

2.

1.

0.

-1.
0. .2 .4 .a , .a 1.o

Fig. 12. Pression a la paroi a portance maximale CLmax

(NACA4412, M = .18, a = 12S0, Re = 4.2 lo6,
xlr+ = ,014, xlr- = ,113, modkle 2 tq.)

0. LL I I I alpha
0. S. IO. IS. 20. 25. Fig. 14. Dtcollement mass$. Lignes de courant et iso-Mach.
(Cylindre circulaire, M = .18, Re = 4.2 lo6)
Fig. 13. Courbe de portance. Dtcrochage.
(NACA4412, M = .18, Re = 4.2 lo6, modkle 2 tq.)

9. RESULTATS 3D.
notablement le coQt du calcul. Notons enfin qu'un sous-
maillage unilatkral (selon i) du calcul visqueux se trouve intro- L'extension en tridimensionnel de la mCthode numkrique
duit automatiquement, de plus, dans les zones de faible interac- (code VIS11) conserve exactement, plan par plan, les mCmes
i tion visqueuse le nkcessitant. caracttristiques d'auto-adaptation du maillage aux effets
visqueux que celle de la Fig. 8. Le maillage moins riche utilist

,
ici en tridimensionnel est 131x21~20(champ non-visqueux)
1 8.3. Cylindre circulaire.

L'accbs des mkthodes d'interaction visqueux-non visqueux

plus 131x15~37(champ visqueux), avec sous-maillage addi-
tionnel visqueux selon i dans les zones de couche limite mince
attachke.
au calcul du dtcollement massif est enfin illustrt par la possi-

'
I bilitC de calcul du cas-limite du cyclindre circulaire, Fig. 14, au
moyen du code VIS05 - profil d'aile. La prtsente thtorie de 9.1. Voilure rectangulaire NACA4412.
"Formulation-Dtficitaire"et son "Rtfkrentiel de dtplacement"
kliminent ici les difficultts de couche-mince ou de rtfkrentiel Des calculs de validation numkrique de la mtthode ont
de couche-limite discontinu au (pseudo-) bord de fuite. d'abord ttt obtenus sur une aile NACA4412 rectangulaire
L'algorithme "Semi-inverse Massive separation" et la thtorie d'allongement 6 en atmosphbre illimitte, en conservant les
numkrique du couplage proposCe dtmontrent sur ce cas-test mCmes conditions de vitesse et de nombre de Reynolds que
leur application jusque dans le cas-limite stvkre oa le vecteur celles de Hastings. La Fig. 15 montre les lignes de frottement
vitesse "non-visqueux" B la paroi devient orthogonal B cette calculkes en turbulent aux grandes incidences sur l'extrados de
I paroi (et non plus tangent comme en couche limite). Un la voilure, et les lignes de force du champ de vitesse sur la
1 dtcollement turbulent B azimuth rkaliste de 1100 est ici prtdit
par le code VISOS, en calcul stationnaire.
nappe de proche sillage. Les lignes d'accumulation de
dtcollement et de recollement montrent l'apparition du
dkcollement. On peut remarquer en particulier l'importante
zone de decollement calculte B l'incidence 17', le dtcollement
de l'aile prenant ici naissance par l'emplanture (plan de
26-14

symCtrie). La Fig. 3 montre que la nappe de sillage est mise en

Cquilibre. La Fig. 16 fournit les distributions de pression
calculCes en envergure i l’incidence 17*, et montre I’apparition
d’un plateau de pression dans la zone dCcollCe. Les rCsultats
des Fig. 15-16 correspondent i des calculs sur des maillages oh
I’auto-adaptation est rCalisCe pour ce qui est du calcul visqueux
et de 1’6quilibrage de la nappe, mais qui sont dCpourvus
d’auto-adaptation du maillage de surface au dCcollement (cas-b
de la Fig. 17). Dans ces calculs sur maillage de surface orthog-
onal, I’approximation de troncature visqueuse est alors
sin h’ = 1.

Fig. 16. Distribution de pression h la paroi.

Aile rectangulaire NACA4412, allongement = 6 .
( M = .18, a = 17O, Re = 4 lo6, turbulent)

9.2. Calcul A maillage de surface auto-adaptatif.

La technique de maillage auto-adaptative mise au point

dans la prksente mCthode (code VISlI) permet aussi I’auto-
adaptation du maillage de surface, par rapport aux effets
visqueux de dtcollement, cas-a de la Fig. 17. L’auto-adaptation
du maillage de surface est ici asservie au passage en mode in-
verse du calcul visqueux, c’est 2i dire asservie au cbne
caratkristique et au domaine de dCpendance local du calcul de
couche limite couplCe.
Dans ces calculs, I’hypothtse d’une direction h’ de tron-
cature visqueuse parallkle aux lignes i = csr, qui est acceptable
lorsqu’elle est possible, et qui ne conduit pas alors 2i des
modifications majeures des solutions par rapport au calcul avec
sin h’ = 1, ne peut cependant Etre maintenue sur un maillage de
Fig. 15. Lignes de frottement. Extrados. Maillage b e . surface distordu par adaptation A un grand dCcollement. Le cal-
Aile rectangulaire NACA4412, allongement = 6. cul avec k’ paralltle aux Iignes i = csr sur un maillage con-
(M = .18, Re = 4 lo6, turbulent) vergt tel que celui de la Fig. 17 cas-a ne peut en effet Etre
 26-15
I

Fig. 19. Lignes de frottement. Maillage auto-adaptatif. Sin h' # 1.

Aile rectangulaire NACA4412, allongement = 6 .
( M = .18, a = 17', Re = 4 lo6, turbulent)
incidence 17', Fig. 18-19.On peut noter que le calcul avec
maillage de surface auto-adaptatif et sin h' = 1, Fig. 18, repro-
duit sensiblement le resultat sin h' = 1 obtenu sans maillage de
Fig. 17. Maillage de surface auto-adaptatif aux effets visqueux. surface adaptatif de la Fig. 15, et que la solution sur maillage
Aile rectangulaire NACA4412, allongernent = 6. de surface adaptatif avec sin h' calcult, Fig. 19, ne conduit
( M = .18, a = 1 7 O , Re = 4 lo6, turbulent) qu'h un lCger accroissement de la zone dCcoll6e.

effectuC sans violer les domaines de dependance et cBnes

caractCristiques de couche limite en mode inverse. Une direc- 9.3. Voilure en fleche AS27-V41.
tion 1' calculte en chaque noeud, en fonction de I'obliquitC du
maillage de surface et du c6ne caracttristique inverse local, est Une premikre validation quantitative de la mCthode de cal-
donc introduite. Elle Cvolue entre la direction i =cst et la direc- CUI tridimensionnelle VIS l l par comparaison calcul-experience
tion orthogonale aux lignes de courant non-visqueuses. Le cal- a CtC obtenue sur la voilure en flbche AEROSPATIALE V41,
cul confirme que I'impact du choix de h' sur les solutions qui a CtC expCrimentCe It grande tchelle sur une maquette
reste faible, le test Ctant ici effectuC sur le rCsultat sensible "avion" [34] dans les souffleries de I'ONERA, et dont les
qu'est le frottement, et dans le cas dCfavorable de la plus forte caractCristiques sont dCj& plus proches de celles des voilures
industrielles.
La Fig. 20 montre le maillage de surface du calcul et in-
dique la gComCtrie de la voilure. Le calcul est r6alis6 en tur-
bulent. La transition expkrimentale est dCclenchCe. La Fig. 21

Fig. 18. Lignes de frotternent. Maillage auto-adaptatif. Sin h'= 1

Aile rectangulaire NACA4412, allongement = 6.
(M = .18,a = 17', Re = 4 lo6, turbulent) Fig. 20. Aile AS27-V41 (allongernent=9.5). Maillage de surface.
26-16

a = 4O

Fig. 22. Comparaison calcul-exp2rience.

Distribution de pression. Aile AS27-V41.
Fig. 21. Lignes de frottement. Extrados. Aile AS27-V41.
(M = 2 4 , Renvergwe = 7.8 IO6, turbulent, modkle 2 tq.)
(M = 24, Renvergwe= 7.8 IO6, turbulent, rnod2le 2 Cq.)
 26-17

reproduit 1’Cvolution en incidence des lignes de frottement

calculCes sur l’extrados de la voilure, ainsi que les lignes de 11. REFERENCES.
force du champ de vitesse sur la nappe de proche sillage (mise
en Cquilibre), aux fortes portances. Les lignes d’accumulation
[I] LE BALLEUR J.C. - New possib es of Viscous-Inviscid numerical
de dkcollement et de recollement qui apparaissent, Fig.21, techniques for solvin viscous flow equations, with Massive se aration.
montrent le dCveloppement en incidence du dkcollement cal- Proceedings Fourth 8ymp. Numerical and Physical Aspects o?Aeoro-
culC, jusqu’au CLmax, avec le modtle turbulent k 2 Cquations dynamic Flows. Selected papers, chap. 4, p. 71-96, editor T. Cebeci,
SPRINGER-VERLAG 1990. (or ONERA TP 1989-24 Reprint)
k--7 proposC. L’incidence 11’ correspond sensiblement sur cette
[21 MELNIK R.E. - Turbulent interactions on airfoils at transonic speeds - Re-
aile k la limite d’opCrationnalit6 du code VISll (au stade cents developments - Agard Conference on Computation of viscous-inviscid
prkliminaire actuel). Les comparaisons calcul-expCrience obte- interaction, Colorado-Springs (1980). AGARD-CP-291, Paper 10. (1981).
nues sur les distributions de pression & la paroi pour ces [31 LE BALLEUR J.C. - Numerical flow calculation and viscous-inviscid in-
teraction techniques - Recent Advances in Numerical methods in Fluids,
configurations dtcollCes & forte portance, Fig.22, semblent Etre Vol. 3 : Computation methhods in viscous flows, p 419-450. W.Habashi ed-
trts satisfaisantes, de mCme que dans le cas non-dCcollt itor. Pineridge Press (1984).
I’incidence 4’. [4] LE BALLEUR IC. - Computation of flows includin strong viscous in-
teractions with coupling methods. - AGARD-CP-291. 8eneral Introduction.
Lecture 1, Colorado-Springs (1981). or ONERA TP 1980-121.
[5] LE BALLEUR J.C. - Viscous-inviscid flow matching : Analysis of the
problem including separation and shock waves. - La Recherche Aerospa-
tiale 1977-6, p.349-358 (Nov.1977). French, or English aansl. ESA-lT-476.
[6] LE BALLEUR J.C. - Viscous-inviscid flow matching : Numerical method
and applications to two-dimensional aansonic and supersonic flows. - La
Recherche Aerospatiale 1978-2, p. 67-76 (March 1978). French, or English
transl. ESA-TT-496.
[7] LE BALLEUR J.C. - Strong matching method for computing transonic
10. CONCLUSIONS. -
viscous flows includin wakes and separations. Lifting airfoils. La Re-
cherche Aerospatiale 1181-3, p. 21-45, English and French editions. (March
1981).
Une mCthode numCrique nouvelle d’interaction visqueux- [8] LE BALLEUR J.C., PEYRET R., VIVIAND H. - Numerical Studies in high Rey-
non visqueux transsonique pour le calcul des Ccoulements nolds number aerodynamics - Computers and Fluids. Vol. 8, no I . p. 1-30,
(March 1980)
dkcollCs sur les voilures aux fortes portances (code VIS11, [9] LE BALLEUR J.C. - Numerical viscid-inviscid interaction in steady and
Viscous-Inviscid-Solver-1l), a pu Ctre dtfinie. Une technique unsteady flows. - Proceed. 2nd Symp. Numerical and Physical Aspects of
3D-approchCe pour les Cquations visqueuses (2.75D-local), Aerodynamic flows, Long-Beach, (1983). chapt.13, p. 259-284 T. Cebeci
ed.Springer-verlag. 1984, (or ONERA-TP 1983-8).
adaptCe k l’allongement inCvitable des mailles en envergure, est
[IO] CARTER J.E. - A new boundary layer interaction technique for separated
proposCe. Le nouveau code VISll offre de surcroit une flows. - AIAA paper no 79-1450. 4th Comp. Fl. Dyn. Conference, (July
premibre approche de calcul tridimensionnel k maillage auto-
adaptatif aux Cchelles d’interaction visqueuse. - New quasi-simultaneous method to calculate interact-
- AIAA J. Vol. 19, no 1 (1981). p.79-85.
Des rksultats nouveaux ont CtC dCgagCs sur les singularitCS [I21 LE BALLEUR J.c.. GIRODROUX-LAVIGNE P. - A semi-implicit and un-
et c6nes caractkristiques de couche limite tridimensionnelle en steady numerical method of viscous-inviscid interaction for transonic
separated flows. - La Recherche Aerospatiale 1984-1. p.15-37, English and
mode inverse. Des rCsultats thCoriques nouveaux ont aussi CtC French editions, (Jan. 1984).
obtenus sur l’algorithme de couplage “Semi-inverse” de Le [I31 LE BALLEUR J.C.. GIRODROUX-LAVIGNE P. - A viscous-inviscid inlenc-
Balleur ainsi que sur sa theorie de stabilitC, en dCbouchant sur tion method for computing unsteady transonic separation. - Proceed. 3rd
une nouvelle extension “Semi-inverse Massive-separation Symp. Numerical and Physical Aspects of Aerodynamics Flows, Long-
Beach (jan.1985). T. Cebeci ed., Springer-Verlag (1986).
2.75D” de I’algorithme, compatible avec le calcul des voilures [14] LE BALLEUR J.C. - Viscous-Inviscid interaction solvers and computation
dCcolltes. of highly separated flows. - “Studies of Vortex Dominated Flows“, chap. 3,
p. 159-192, P r e . ICASE symp. NASA Langley Field, USA, (july 9-10,
Des comparaisons calcul-expkrience satisfaisantes sur le 1985). Hussaini and Salas ed., Springer-Verlag 1987. (or ONERA TP
champ de pression ont pu Ctre obtenues, avec le modtle tur- 1986-4).
bulent 2 Cquations k< proposC, tant sur le profil NACA4412 [15] LE BALLEUR J.C. - New possibilities of numerical viscous-inviscid
coupling, for viscous flows with massive separation, and for inviscid
au CLmax et au dCcmhage, que sur la voilure en fltche Zonal-methods. Proceedings GAWSMAI-IMA conference on Computa-
AS27-V41 de type avion civil, k grande portance et avec tional aeronautical dynamics, May 17-19, 1989. Antibes, , France,
SPRINGER-VERLAG to appear. (or ONERA TP 1989-166 Reprint)
dCcollement.
[la] COLES D.E. - Tle law of the wake in turbulent boundary layers. JFM
Ces rtsultats dCmonuent que les mCthodes d’interaction Vol 1, part 2 (1956).
visqueux-non visqueux donnent un plein accbs au calcul du [I71 MELNIK R.E.. MEAD H.R.. JAMESON A. - A Multi-grid method for the
computation of viscousJinviscid interaction on airfoils - AIAA-Paper 83-
dCcollement mdimensionnel. 0234 (1983).
[I81 MELNIK R.E.. BROOK J.W. - The computation of Viscidnnviscid interac-
tion on airfoils with separated flow. -
Proceed. Numerical and Physical As ts of Aerodynamics Flows 111.
Long-Beach (1985). Springer-Verlag 19Ep77-101, editor Cebeci.
[I91 WIGTON L.B.. HOLT M. - Viscous-inviscid interaction in transonic flow -
5th CFD, AIAA Paper 81-1003, PaleAlto (June 1981).
[20] LE BALLEUR J.C.. BLAISE D. - Computation Of separated intellla1 flows
Remerciements. L‘auieur adresse ici ses remerciements b la and shock wave-boundary layer interactions by viscous-inviscid interaction.
SociitC AEROSPATIALE, qui lui a aimablement communiqui les .- La Recherche Aerospatiale 1985-4, p. 211-227. English and French edi-
tions, (July 1985).
rCpartitions de pression expirimentales sur la maquette AS27-V41, [21] LE BALLEUR J.C..GIRODROUX-LAVIGNEP. - Calculation of fully three-
ainsi qu’au Service technique oficiel francais STPA du MinisQre de dimensional separated flows with an unsteady viscous-inviscid interaction
la Difense. qui a soutenu financieremen1 cette aptrimentation. method. Fifth Symp. Numerical and Physical Aspects of Aeorodynamic
Flows, California State University, Long-Beach (USA), January 13-15
I1 remercie aussi D.Blaise. P . Girodrou-Lmigne, H . Gassot. pour 1992. (ONERA TP 1992-01 Reprint)
leur concours prCcieu dam les iraiiemenis graphiques. [22] LE BALLEUR J.C. - Viscid-Inviscid coupling calculations for two- and
three-dimensional flows. Lecture series 1982-04 on Computational Fluid
Dynamics, Von Karman Institute. Belgium, (March 1982).
[23] LE BALLEUR J.C. - Progrh dans le calcul de I’interaction fluide parfait
- fluide visqueux . Proceedings AGARD-CP-351, Paper 1. Copenhague,
(1983).
26-18

[24] LE BALLEUR J.C. LAZAREFF M. - A Multi-Zonal-Marching integd

method for 3D boundary-layer with viscous-inviscid interaction - Proceed.
9th ICNMFD. June 1984. Saclay, France. Lecture Notes in Physics, 218,
p.351-356, Springer-Verlag 1985.
[ZS] LAZAREFF M.. LE BALLEUR J.C. - Development Of the MZM IIumerical
method for 3D boundary layer with interaction on complex configurations.
Proceedings 8th GAMM conference, Notes on Numerical Fluid Mechanics,
Vol. 29, VIEWEG. Braunschweig 1990. (or ONERA TP 1989-174)
[26] EDWARDS D..CARTER J. - Analysis of three-dimensional se arated
flow with the boundary-layer equations. - AIAA Paper 85-1498, July
1985.
[27] DAVIS R.L.. CARTER J.E.. HAFEZ M. - Three-dimensional Viscous flow
solutions with a vorticity-stream function formulation. - AIAA Journal. Vol.
27, no 7. p 892-900, July 1989.
[28] CEBECI T.. CHEN LT.. CHANG K.C.. PEAVEY C.C. - An interactive scheme
for three-dimensional transonic flows. - Proceed. 3rd Symp. Numerical and
Physical Aspects of Aerodynamics Flows, Long-Beach, USA (Jan. 1985).
T. Cebeci ed.. Springer-Verlag (1986).
[291 WIGTON H.. YOSHIHARA H. - Viscous-inviscid interactions with a
three-dimensional inverse boundary layer code. - Proceed. 2nd Symp. Nu-
merical and Physical Aspects of Aerodynamic flows, Long-Beach, USA,
January 1983.
[301 WAI J.C., BAILLIE J.C.. YOSHIHARA H. - Computation of turbulent
separated flows over wings - Proceed. 3rd Symp. Numerical and Physical
Aspects of Aerodynamics Flows, Long-Beach, USA, Jan. 1985.
[311 YOSHIHARA H.. WAI I. - Transonic turbulent separation on swept wings.
A return to the direct formulation. - AIAA Paper 84-0265, Reno, USA,
January 1984.
[32] STEGER J.L.van DALSEM W.R. Developments in the simulation of
separated flows usin finite difference methods. - Proceed. 3rd Symp. Nu-
merical and Physicaf Aspects of Aerodynamics Flows, Long-Beach, USA,
Jan. 1985.
[33] HASTING R.C.. WILLIAMS B.R. - Studies Of the flow filed near a
NACA4412 aerofoil at nearly maximum lift - Aeronautical Journal, p. 29-
44 (January 1987).
[34] PUJOL C.. DELORT M. - Wveloppement exploratoire aerodynamique.
Synthkse de I’exploitation des mesures sur la maquette AS27 dans la
soufflerie SlMa. Rapport AEROSPATIALE 445.077 / 83, Octobre 1983.
 27-1

FORTY YEARS OF HIGH-LIFT R&D -AN AIRCRAFT

MANUFACTURER'S EXPERIENCE

E. Obert
Fokker Aircraft 6.V.
P.O. Box 7600
1117 ZJ Schiphol
The Netherlands

SUMMARY
In the course of four decades a large amount of high- Po Static pressure of the undisturbed
lift applied research and development has taken place airflow
at the Fokker Company. In the 'fifties and 'sixties the q Dynamic pressure
F-27 and the F-28 were developed. In the 'eighties
these aircraft were developed further into the Fokker Re, Reynolds number related to the airfoil
50 and Fokker 100. In the seventies an extensive R&D section chord
programme was performed in preparation of a possible
successor to the F-28 leading to the F-29 project Re,- Reynolds number related to the wing mean
study. In each case two- and three- dimensional
windtunnel models were investigated in numerous aerodynamic chord
configurations. In the last decade these investigations S Wing area
have increasingly been preceded by theoretical a Airfoi 1 section angle-of-attack
investigations. Where data are available comparisons ("R Aircraft angle-of-attack with reference
have been made with flight test data. Of each to the fuselage centre line
development programme a detailed account is presented. 8, Elevator angle
Particular attention is paid to Reynolds-number effects 6, Flap angle
and the interconnection between the high-speed cruise 6s Slat angle
and low-speed high-lift design requirements.
I. INTRODUCTION

Wing aspect ratio When the Koninkl i jke Nederlandse Vliegtuigenfabriek

Wing span Fokker N.V. (Royal Dutch Aircraft Factory Fokker)
Airfoil section chord restarted design and production of its own aircraft
Wing mean aerodynamic chord after the Second World War good high-lift
characteristics were not among the prime goals pursued
Airfoil section drag coefficient in the design activities. Fokker's first designs, the
Aircraft drag coefficient F-25 Promotor, the S-11 Instructor, S-12, S-13 and S-14
Airfoi 1 sect ion 1 ift-coeff ic ient Mach-trainer were touring aircraft and military
Airfoil section maximum lift coefficient trainers where airframe-propul s ion integration and
Aircraft lift coefficient f 1 ight hand1 ing required much more attention than
Aircraft maximum lift coefficient maximum performance. For these designs split flaps were
good enough.
Aircraft maximum lift coefficient
based on minimum speed reached in the In 1950 however, a design study was started on a two-
stall manoeuvre (FAA definition) engined transport aircraft which was envisaged as a
Aircraft maximum lift coefficient based successor to the then ubiquitous Douglas DC-3 or C-47.
on the speed at which the "g-break" This was seen as a tremendous challenge. The C-47 was
occurs at that time easy to obtain at very reasonable prices
and had for its days quite reasonable field
('LmX'V min. / 1 , 4 4 performance. The new aircraft, the Fokker F-27
Friendship would be equipped with the not yet general1
Increase in section lift coefficient due accepted turboprop engine (the Rolls-Royce Dart RDa61
to flap deflection at a=O. in order to obtain lower operating costs and a
Aircraft lift coefficient for the significant increase in productivity through a higher
aircraft- 1 ess-ta i 1 cruising speed. Take-off power rose from 1200 hp for
Airfoil section pitching moment the C-47 to 1600 shp for the F-27 Mk100. The higher
coefficient cruising speed demanded a higher wing loading for
Increase in airfoil section pitchin optimum cruise performance. Wing loading increased from
moment coefficient due to ?lap 139 kg/m2 (28.4 lb/sq.ft) for the C-47 to 253 kg/m2
deflection at a-0 (51.7 lb/sq.ft). As the span of both aircraft was
Tailplane setting practically identical (29 m or 95 ft) the span loading
Lift-drag ratio at a speed 20 percent MTOW /b* increased from 15.11 kg/m2 (3.10 lb/sq.ft) for
the C-47 to 21.06 kg/m2 (4.32 lb/sq.ft).
above the stalling speed Finally the aircraft would be certified according to
Static pressure coefficient I
the U.S. certification regulations CAR 4b which were
-
P-Po considerably more stringent, both with respect to
performance and to flight handling, than was the
9 certification basis for the original DC-3.
Mach number
Local static pressure Nevertheless, the DC-3 field performance was to be
duplicated.
27-2

-
i
Thus a set of very demanding aerodynamic design NACA 64, 421 mod (root)
requirements was formulated for low-speed flight (high NACA 64, - 418 mod) (70% half-span
maximum lift in take-off and landing, high lift-drag NACA 64, - 415 mod) (tip)
ratio’s at take-off and landing conditions and
excellent low-speed flight characteristics both in all- The selection of these particular sect ons was entirely
engine and one-engine-out f 1 ight conditions). This based on a analysis of available windtunnel test data
special attention for low-speed flight, both with from open literature. The modifications consisted of
regard to aircraft performance and to flight handling straitening-out the cusps on the rear upper surface.
has ever since the original design of the F-27 been a
trade mark of Fokker designs as these have always
primarily been aimed at the short- and medium haul
market.

11. THE BEGINNING -DEVELOPING THE H I G H - L I F T

CHARACTERISTICS OF THE FOKKER F-27 FRIENDSHIP

When one looks into the open literature for data on

airfoil section high-lift characteristics available in
1950 one finds that most of the data is compiled in
four publications:
a. Theory of Wing Sections (Abbott and von Doenhoff,
ref. 1).
b. Summary of airfoil data, NACA TR 824 (Abbott, von
Doenhoff, Stivers, ref. 2).
c. Summary of section data on trailing-edge high lift
devices, NACA TR 938 (J.F. Cahill, ref. 3).
d. The Aerodynamic Characteristics of Flaps, ARC R&M
2622 (A.D. Young, ref. 4).
The data in these publications concern either tests
with efficient high-lift devices on airfoil sections
which were then considered obsolete such as the NACA 4-
and 5-digit series and the British RAF-sections or the,
at that time, modern NACA 6-series laminar flow
sections but at moderate relative thickness (10 to
12%). Data from systematic tests on a large series of
NACA 6-series airfoil sections equipped with a 0.20~
Lum .nu--.
NAc.4 -
P...m d UYI

Variation or maximum section lilt mcllicicnt with nirloil

thickncm ratio and camber for several NACA airfoil sections
split flap were available however. Also a very limited
set of data on thick NACA 6-series sections with
double-slotted flaps can be found. Fig. 1 Effect of airfoil section thickness on CLmX
(from ref. 1).
All test data indicated however that within the airfoil
classification system used at that time both older and
later section families showed the highest maximum lift Windtunnel tests for the analysis and optimization of
coefficients at a relative thickness of about 12% for the high-lift characteristics were performed on two
the clean section and of about 20% with a deflected mode 1 s :
trailing-edge flap. This is illustrated in figure 1
taken from ref. 1. a. A complete model at scale 1 : 15 measured on a
six-component fxternal mechanical balance at
Because of the desire to minimise induced drag both in Re, = 0.8 x 10 and M = 0.12
take-off and in cruise a wing of very high aspect ratio
was required for the F-27. An aspect ratio A=12 was b. A two-dimensional airfoil section model with
felt to be achievable provided the flutter requirements section chord C = 600 mm spanning the complete
could be met without significant weight penalties. This tunnel test section between the side walls (tunnel
could be safeguarded by providing maximum cross- width = 3.00 m). The airfoil section was NACA
sectional area in the wing torsion box. Therefore 64,-421 (mod). The model consisted of three parts
maximum airfoi 1 thickness was required over the greater of equal span. The outerparts were fixed to turning
part of the wing span. tables on the tunnel side walls. The centre part
was fitted to the outer parts through strain-gauge
Thus a happy coincidence occurred between the balances with which normal and tangential forces
requirements concerning wing weight and stiffness and and pitching moments could be measured. This
low-speed high-lift characteristics, albeit at the arrangement was supposed to minimise the effect of
expense of a somewhat higher profile drag. lift loss near the tunnel walls due to flow
separation in the corners between model surface and
The F-27 wing is defined by a 21% thick root section tunnel wall. Additionally drag was measured with a
and a 15% thick tip section. As a consequence at the wake rake at one rake position for the majority of
outer end of the.flapat 70% half-span the wing section tests. Only a limited number o f pressures were
.
relative thickness is still 18% Thus on average over
the wing part covered by the trailing-edge flap the
measured with a wandering static tube to obtain
data for loads analysis. Most tests were performed
relative wing thickness is 20% . at a Reynolds Number Re, = 2.3 x lo6.
The governing airfoil sections on the F-27 wing are: For the analysis of the flight handling characteristics
separate outer wing and tail surface models were used
to obtain detailed control surface hinge moment data.
 21-3

which made it superior to the Fowler flap. Figure 3

shows lift curves for the double slotted flaps with
fixed vanes and for the compound flap as measured on
the two-dimensional model.
The compound flap was then tested on the complete F-27
model at scale 1:15 and showed indeed a marked
improvement in the maximum lift coefficient for
landing. This is illustrated in figure 4.
location of the fixed rotation axis. A l s o the effect on
lift and drag was investigated of variations in the
3.01
shape of the lower shroud of the flap cavity on the

m Flap IU

r m Flap E

Fig. 2 Flap configurations tested on the F-27 2-D

model.
Representative CLmx - values recorded for the various
flap types are presented in the table below.

Flap no Flap type Flap setting Re, Gnl,

1 Double-slotted, 40 deg 2.2 x 10' 2.80
f 1 x 4 vane
I1 Double-slotted 50 deg 2.2 x 10' 3.13
fixed vane
I11 Conpound flap 40135 deg 2.2 x IO' 3.76
IV Fowler 45 deg 1.7 x 10' 3.68
Va Double-slotted. 55 deg 2.2 x 10' 3.37 Fig. 3 Lift curves for four flaps tested on the F-27
fixed vane 2-D model.
Vb Double-slotted, 50 deg 2.2 x 10' 3.27
fixed vane
When the first F-27 prototype started test flying in
November 1955 it quickly showed to have less than
The data obtained on the two-dimensional model led to satisfactory flight characteristics. In particular with
the conclusion that the use of the compound flap would large flap deflections and a fair degree of engine
be advantageous. This flap was a double-slotted flap on power longitudinal stability including the flare
i which the front and rear flap were of roughly equal characteristics in the landing were considered
chord and on which the rear flap could move relative to unsatisfactory. This was aggrevated by shortcomings in
1 the front flap. lateral and directional control.
This flap offered not only a very high maximum lift in After only a limited number of flights it was therefore
the landing, but, when the rear flap was not deflected decided to fix the rear flap to the front flap in the
thus closing the second slot, low drag for take-off neutral position and continue flight testing with the
conditions could be realised. Furthermore this flap single-slotted flap configuration. The maximum flap
I allowed the use of a fixed hinge and offered a
reasonable section pitching moment, two advantages
angle became thus 40 degrees.
27-4

F o k k c r F-22 stabilizer setting i, = -1 deg (stabilizer nose

JircrrfL - tars- tu.1 Power o f f .
down) and i, = + 3.5 deg with the flaps fully down.
C. An enlarged elevator with modified balance nose and
a spring tab.

Front. spar

Fig. 5 F-27 horizontal tailplane.

Modified leading-edge for production
aircraft.
Fig. 4 The F-27, windtunnel model. scale 1:151 -
Lift curves.
Although the improvement in C, x , compared to the
At one stage in the flight test programme in 1957 standard aircraft, as predicter by the windtunnel
stabilizer effectiveness with leading-edge roughness testdata could be realized, (figure 6) the flight
was to be investigated at extreme negative stabilizer handling characteristics were still considered
angles-of-attack. As the leading-edge de-icing system unsatisfactory and no further development work was
contained pneumatic boots, which function only when performed on this configuration.
some ice accretion has occurred, a limited amount of
leading-edge roughness must be tolerable. So, what did the development history of the F-27 teach
Leading-edge roughness was simulated for this test by us ? It tought us that for a medium-speed propeller
the application of bird-seed. At one condition with the aircraft design for high-lift is not so much a matter
flaps fully deflected the aircraft pitched forward at of coming up with an efficient flap system on a
the application of power reaching zero g and plunging suitable airfoil section but much more a matter of
into a vertical dive. keeping the high lift literally "under control". The
Retracting the flaps and closing the engine throttles combination of high-lift and high engine power leads at
saved the situation. take-off and landing speeds to very high propeller
thrust coefficients with associated strong adverse
As a result of this experience the inner wing flap slipstream effects on the flight characteristics in
extension mechanism was altered such that at each flap pitch, roll and yaw both in all-engine and single-
setting the deflection of the inner flaps is only two- engine flight. Concerning the longitudinal
third of the outer flap extension. Thus, the maximum characteristics a detailed analysis of these phenomenae
deflection of the inner flaps on the F-27 is 26 deg. is presented in ref. 5
Also, the stabilizer leading edge was cambered upward
as shown in figure 5. These two measures completely
elimininated the possibility of negative tailplane
stall over the complete range of flap settings,
operating speeds and power settings even at push-overs
Cl :;:-
3.2.
Fokker F-21 w i t h compound flap

up to 0.5 g.
3.0.
So a design which was originally intended to be
equipped with very effective double-slotted flaps (the 2.8.
"compound flap") was finally certified with single-
slotted flaps with reduced settings on the inboard 2.6.
panels. This resulted in a decrease in maximum lift Flap bmttingc
coefficient of 15 percent but was necessitated by the 2.4. Outboud Imbo&d
flight handling certification requirements, both for 26 li

/
all-engine and single-engine flight.
2.2.
As a part of design studies into possible further
development o f the F-27 flight tests were performed in 2.0.
1960 on a highly modified F-27 prototype. The
modifications consisted of: 1.8. V
30
a. The reactivated double-slotted compound flap as
used during the very first flight. On the inboard
flaps the aft flap was fixed in the neutral
position.
'
0 lo 2b io
-t---r---r--
6 .Front
T--

+totat

b. A variabel-incidence tailplane mechanically coupled Maximum lift from flight tests on the F-27
to the flaps such that with flaps retracted the prototype equipped with the compound flap.
 27-5

The basic wing sections were taken from open wing spanwise lift distribution occurred at a spanwise
literature, the various flap sections were drawn with position just outboard of the tailplane tips. In order
available French curves keeping practical experience by to obtain the flow separation pattern as described
others in mind and the range of flap gaps and overlaps above the airfoil section at 40 percent halfspan was
to be investigated in the windtunnel were again taken made thicker than the root and tip section.
from open literature with an eye on acceptable flap
bracket size and the associated flap extension The airfoil sections on the S-14 which had a straight
mechanism. No theoretical design tool was used neither wing were o f the British RAE EC-XX40 series and had the
was one available and pressure distributions were not following thicknesses:
considered for high-lift design .
Still, apart from the measures taken because of root 12.1%
aircraft control considerations, no modifications were 40% half-span 14.5%
required on the flap system on the actual aircraft. tip 12.0%
Also no tailoring of the stall characteristics was
required. The F-27 has perfect wing stalling As the S-14 had a negligible tuck-under tendency
characteristics without any protrusions such as between M = 0.78 and M = 0.82 requiring less than 10 lb
leading-edge stall strips or fences. change in stick-force over that small speed range (see
And a 1-g CLmx = 2.6 for an aircraft with single- ref. 6) the principle was assumed to work.
slotted flaps and CLmx = 3.05 for an aircraft with
double-slotted flaps are sti 1 1 respectable numbers even The same approach was followed in the initial design of
today. the F-28.
The original wing design of the F-28 had zero sweep at
Obtaining satisfactory characteristics on the basic the 40% chord line. Again, the section at 40 percent
F-27 over the complete flight envelope for the manually half-span was thicker than the root section. The
controlled elevator, ailerons and rudder took however initial windtunnel model demonstrated however a larger
the greater part of a two-year development and flight than expected pitch-down tendency at transonic Mach-
test programme. (1955 - 1957) numbers, possibly due to the high horizontal tailplane.
In subsequent design steps the wing sweep was therefore
increased to 16 degrees at the quarter-chord line
maintaining the concept of the spanwise thickness
111 HIGH LIFT COMBINED WITH JET SPEEDS - THE distribution as described above. In its final
DEVELOPMENT OF THE HIGH-LIFT CHARACTERISTICS configuration the high-speed pitching characteri st ics
OF THE F-28 Mk1000 were so much improved up to M, that no Mach-trim
compensator or other pitch-augmentation system was
required on the certified aircraft.
In 1961 design studies were initiated on a jet
transport aircraft with a capacity of 44 passengers, The governing sections on the F-28 wing have the
soon to be increased to 65 passengers, and cruise following relative thickness:
speeds up to M = 0.75. This aircraft was meant to
complement or succeed the F-27. As the aircraft was root 13.6%
intended for the lower end of the market and as in many 40% halfspan 14.0%
cases it would be the first time a possibly less 70% halfspan 10.5%
experienced airline would operate jet aircraft it was tip 10.0%
felt that optimum flight handling characteristics
coupled with maximum simplicity in the aircraft systems Since the development of the S-14 insight in airfoil
were mandatory. This feeling was strengthened by early section characteristics had increased in two
experience with the Boeing 707, Douglas DC-8 and significant area's:
Caravelle where the absence of slipstream effects as a
safety factor coupled with (on the first two mentioned a. Systematic analysis of a large number of windtunnel
aircraft) less-than-optimum control characteristics test data had indicated that .basically three types
lead to a number of serious incidents. of airfoil section stall could be distinguished:
(ref. 7).
Finally, the new aircraft was to have a field
performance comparable to that of the F-27. trailing-edge stall
leading-edge stall
The new aircraft, the Fokker F-28 Fellowship would not thin-airfoil stall
be bothered by adverse slipstream effects. Instead, a
new potential problem area was introduced: transonic Furthermore it was established that-the type of stall
aerodynamics .'
At Fokker's early experience had been gained with
was to a large extend determined by the leading-edge
shape. The governing parameter was the ordinate of a
transonic aerodynamics in the design and flight testing point on the upper surface at 1.25% from the leading-
of the S-14 Mach Trainer. edge (ref. 8). This is in effect the combined effect of
nose camber and leadina-edae radius.
In the design of this aircraft transonic tuck-under, a This coordinate could even be correlated with the
nose-down pitching tendency caused by flow separation maximum lift coefficient at a given Reynolds number
-
due to shockwave boundary - layer interaction and particularly for thinner sections with little camber.
exhibited also by many present day jet aircraft when
flying at transonic Mach-numbers, had been prevented,by b. Following research at NACA in the '30s and in
a novel design approach. Germany during World War Two the NACA modified
In this approach it was reasoned that the nose-down 4-digit airfoil sections were shown to have
pitching tendency stemming from the changing pressure superior highspeed characteristics when compared to
distribution on the wing once this flow separation standard NACA 4- and 5-digit series and the later
occurred could be compensated by a changing downwash laminar flow sections such as the NACA 6-series.
condition at the horizontal tailplane. The increased NACA modified 4-digit sections differ from the
downwash which was required for this compensating standard 4-digit sections in their location of
aircraft nose-up pitching moment would have to be maximum thickness (40% to 50% chord instead o f 30%)
caused by properly positioned additional trailing and thejr leading-edge parameter (= leading-edge
vortices from the wing. This required that the initial radius /chord).
flow separation with its associated local dip in the It turned out that at a given lift-coefficient and
21-6
e
relative thickness higher drag divergence Mach sufficiently extra drag, if required, would be '
numbers could be achieved when the leading-edge available from rear-fuselage speed brakes.
radius was larger than on the more conventional
airfo i 1 s. In order to determine the most suitable flap
configuration for the F-28 a two-dimensional model was
This was found to be caused by the way the initial area made based on an airfoil section which came very close
of supersonic flow on the section upper surface to the actual wing section at 60 percent halfspan. The
developed with increasing Mach number. Whereas on model spanned the height of the NLR low-Speed Tunnel
convential airfoil sections at cruise lift coefficients (2m) and was provided with a large number of pressure
and subsonic Mach-numbers the upper surface pressure tappings. The model could be provided with any of seven
distribution either shows a slight rearward slope from different fla s with a chord of 32% of the wing chord.
the leading-edge onwards or shows maximum super
velocities near their mid-chord point the sections with
(See figure 8f. The model chord with flap retracted was
0.gO m and most tests were performed at Re, = 2.8 x
1
larger leading-edge radii show a sharp leading-edge 1 0 . As the lift and the pitching moment would be
suction peak. This difference causes an almost
explosive development of the area with supersonic flow
determined via pressure integrations and the drag was
obtained via wake rake measurements the model was
1
on the conventional airfoil once the critical Mach- directly fixed to the upper and lower turning tables on
number is surpassed resulting in strong shockwaves and the windtunnel floor and ceiling. As there were at that
flow separation. On the NACA modified 4-digit series time no provisions to remove or re-energize the
the area with supersonic flow extends gradually from boundary layer on the tunnel floor and ceiling near the
the leading edge, often with a degree of isentropic model considerable flow separation occurred at high
recompression and ending in a weak shock leading to lift in the corners formed by the tunnel walls and the
higher drag divergence Machnumbers. model surface. This prevented the realisation of true I
two-dimensional flow to a considerable extend at high- 1
So, here was a situation were, contrary to earlier lift conditions resulting in considerable distortion of
days, low- and high-speed characteristics were not the upper part of the C, - versus -a curves for the 1
correlated to airfoil sections or section families as various configurations tested. A procedure had
therefore to be defined to estimate the maximum lift
I
a whole but specific characteristics of the geometry
and the pressure distribution could be correlated with coefficient for each test for the equivalent two-
specific dimensional flow condition. This was based on assuming
aerodynamic characteristics. These discoveries were the a near-1 inear curve up to the angle-of-attack for total
first step in the development of a process in which the flow break down. (See figure 9). Tuft investigations
designer obtained the possibilities to tailor airfoil which showed that in most cases leading-edge stall
sections to his own specific needs instead of picking occurred justified this approach. ,
an airfoil section from a catalogue.
A detailed comparison was made for the various flaps
In 1962 18 airfoil sections were thus defined by concerning maximum lift, lift-to-drag ratio's, pitching
Fokker and tested in a small trantonic tunnel at NLR at moment and sensitivity to flap gap flow disturbances
.
Reynolds numbers u p to 2.2 x 10. In figure 7 an due to flap bracketry and flap suspension mechanisms.
overview of these sections is presented. Most sections This led to the choice of the double-slotted flap with
have large leading-edge radii. Sections no 17, 18, 5 movable 'vane (flap no 6) as the most suitable
and 6 were used in defining the final F-28 wing. configuration for further development. Figures 10 and
11 shOw some testdata obtained in this analysis.

Fig. 7 Transonic airfoil sections tested in the NLR

High-speed Pilot Tunnel.

No leading-edge devices were envisaged for the F-28.

For the trail ing-edge flaps again requirements were
formulated for very high maximum lift for take-off and
landing configurations combined with high lift-drag
ratio's at take-off and landing conditions at 1.2VS and
1.3 V, respectively. In particular for the landing it
Fig. 8 Flap configurations investigated during the ,
F-28 development. I
was considered important to achieve low drag as
 21-1

as upper wing shroud trailing edge angles and

thicknesses and clearances with the flaps retracted,
sealing provisions and practical roller paths and flap
track shapes were considered of more importance.
Much attention was also paid again to minimizing drag,
in particular for the take-off flap positions. This
resulted in the adoption of movable flap shroud doors
(see figure 12).
Having defined the flap geometry with its associated
extension path two complete models, equipped with the
double-slotted flap, which spanned 70 percent of the
wing span, were tested. The larger one, at scale l:U,
in the NLR low-speed tunnel (LST) and the smaller one
at
scale 1:20 in the NLR High-speed Tunnel at M = 0.19 at
a tunnel pressure of 4 atm. These test2 were performed
at Reynolds numbers of Re, 7 1.4 x 10 in the LST and
Re, = 1.0, 1.4 and 2.8 x 10. in the HST.
Also a half model at scale 1:12 was investig%ted in the
Fig. 9 Correction procedure applied to the 2-D HST at Reynolds numbers up to Re, = 5 x 10 . On this
testdata during the F-28 development. model only the flow separation pattern at the stall
could be studied as a function of Reynolds number as no
force balance was fitted.
In the performance calculations for the F-28 it was
assumed that CLmx = 1.5 for the clean aircraft and CL,
= 2.5 for the aircraft in the landing configuration
would be realised, according to the definition of CLmaxs
and this value being obtained with the procedure as
described in the certification requirements,of CAR 4b.
Using the two-dimensional data with the classical
Weissinger extended lifting-line theory and using
limited available data on Reynolds-number effects on
maximum lift and on the difference between 1-g CLmx and
CAR 4b (minimum speed) C the conclusion was that
this was achievable. At th% time (1961/’62) this was
seen as a considerable achievement as the jet aircraft
then flying (Boeing 707, DC-8, Caravelle, Convair
A Coronado) reached C ,,-values of the order of 2, even
6 IO 30 40
Flap angle (dag) r lo eo 40 (dog)
FI ap angle
when equipped with teading-edge devices.

DOORS HlNOE TO FORM A

Fig. 10 F-28 - 2-D maximum lift.

Fig. 12 F-28 - Hinged lower surface flap shroud doors

for minimum drag.

When the LST- and HST-models were tested it was found

that the expected C, ,-values were realised, but the
characterisyics were unacceptable both in
$;Ainfoffering too little resistance to achieving
1 Fig. 11 F-28 - 2-D pitching moment coefficients. extreme angles-of-attack in post-stall conditions) and
in roll.
As the F-28 flap was assumed to be fitted on roller Concerning the latter the models showed tip stall at
tracks far more freedom was available to define optimum low Re-numbers and a sudden flow separation over t$e
intermediate flap positions. This lead to a much larger complete wing on the half-model at Re = 5 x 10 ,
test programme on the two-dimensional model for the F- Clearly, some stall control device(sf was (were)
28 than on .the corresponding F-27 model. Automated required to initiate stall on the inboard wing.
electronic data reduction helped much however to keep
the test period within acceptable limits. On both the complete models and on the half-model a
number of wing fences were tested at different/
1 Just as during the design of the F-27 no theoretical spanwise positions alone or in combination with inboard
’
tools were used in the design of the various flaps stall strips. Longitudinal stability at the stall was
studied for the F-28. Again French curves were used for clearly improved and in particular on the half-model at
, defining their geometry. Practical considerations such RE, = 5 x lo6 wing stall started clearly on the inboard
21-8

wing. All stall control devices lowered CLmax by We learned the following:
CLmX = -0.1 to -0.15, however. (Figure 13).
a. Wing geometry requirements have to be considered
So came the time that the stall was to be investigated integrally, for both high-speed and for low-speed
on the prototype aircraft in 1967. After an extensive flight, in particular when no leading-edge devices
flight test programme aimed at obtaining satisfactory are used.
handling characteristics combined with a minimum loss
in maximum lift a configuration was frozen with a b. Tayloring stalling characteristics both in roll and
single short leading-edge stall fence at Wing Station in pitch has to be considered as early as possible
3784 (mm from the fuselage centre line) combined with in the design as it is most likely to affect
a stall strip just inboard of the fence the size of a maximum lift and thus aircraft performance.
match stick. The latter is there to ascertain a
consistent flow separation behaviour. The stalling c. Good handling characteristics in combination with
characteristics are such that no stick-pusher is minimum lift loss can only be determined reliably
required. Figure 14 shows the various C, -values in windtunnel tests at high Reynolds numbers.
measured in the windtunnel and on the fuy? scale
aircraft. d. For reliable windtunnel testdata details of the
Refs. 9, 10 and 11 present more details on the flap mechanism have to be reproduced faithfully on
aerodynamic development of the Fokker F-28 Mk1000. the model.

IV LEADING-EDGE DEVICES

As part of continuing efforts to increase the

versatility of the F-27 a study was started in 1968 to
analyse the potential of equipping the F-27 wing with
slats.
A windtunnel test was performed on the original two-
dimensional model used in the F-27 development
programme. Apart from the original airfoil section two
modifications were tested:
---a Clean wing ----a
Clran wing
a. A modified leading-edge section with a larger
4 h n e e at Fenca at
wing STA mac wing STA 3784 leading-edge radius than on the basic section
combined with some leading-edge'camber.
b. A leading-edge section equipped with a slat.
The tests showed that a modified fixed leading-edge
showed only limited increase in,,C notwithstanding
the increase in leading edge radius whereas the slat
could improve CLmX by as much as ACLmX = 1.0 over the
complete flap range (Figure 15).

-
rlnd tunnel test
flap w t i n q 410
m 5 I 106
Effect of fence on progres- 4.0 -
sion of flow separation

Fig. 13 Effect of stall control devices on maximum

lift and flow separation pattern.

0 Basic airfoil
A Drooped n o s e
n Slat deflected

Sackion NACA 64,,-421(nod)

Flap angle (dag)

F i g . 15 Leading-edge modification and slat tested on

the F-27 2-D model.
Fig. 14 F-28 MKlOOO - maximum lift as a function of
Reynolds number.
In the late '60's windtunnel tests were performed in
the NLR-LST on a two-dimensional model equipped with a
slat and a single- or a double-slotted flap. Two slat
So, what did we learn, concerning high-lift shapes and three main component leading edge shapes
characteristics, in the development of the Fokker F-28 were investigated. The second single-slotted flap had,
Fellowship ? compared to the basic single slotted-flap a smaller
leading-edge radius and consequently a stronger upper
surface curvature.
 21-9

The double-slotted flaps differed in their design This considerably improved the credibility of the test
method. The first flap was defined entirely on an data and, incidentally had also proved the correctness
empirical basis, keeping in mind some inhouse of the correction procedure adopted for the F-28 test
formulated design rules and with an eye on seemingly data mentioned in chapter 111 (Figure 18).
succesful vane shapes from open literature (figure
16). The total flap chord was 30 percent of the section
chord. The other double-slotted flap was designed on a
theoretical basis. The total flap chord of this latter 42
flap was 35 percent of the section chord.
The airfoil section chord was 0.75 m Jeading to a
maximum test Reynolds number Re = 3 x 10 .

-
The various components investigated on this model,
indicated here as model A, are shown in figure 17. 25

From:
16

..
NACA RM S A 1 0
~ RM L56602
RHA58H12
6

NACA RM L52129

v Model A , F l a p 3 d - I Madel 5-2-Fta11 6

Fig. 16 Vanes shapes found from literature. Fig. 18 Effect of tunnel wall blowing on lift curves-
F-28 2-D model, Model 5-2.

The tests on the new 0.75 m chord model which consisted

of the now standard integrated-pressures tests for lift
and pitching moment and wake-rake measurements for drag
were performed on a number of different slat-main
component-vane-f lap configurations at many combinations
of gaps, overlaps and deflection angles.

.a-- ---
The tests.produced the following results:
a. The two single-slotted flaps produced for all
practical purposes identical maximum lift notwith-
standing the differences in geometry.
(Figure 19a).
b. Removing the "hook" on the slat lower surface did
not improve the slat's high lift characteristics.
(Figure 19b);
c. Decreasing the radius on the main component's
leading edge had no effect on maximum lift unless
the leading edge radius became extremely small.
(Figure 19c).
d. The three double-slotted flaps (as a third flap the
vane of the f i r s t flap was combined with the other
main flap) produced practically identical maximum
lift if the difference in extended chord was taken
into account.
(Figure 19d).
Fig. 17 High-lift devices tested.on the 2-D model,
Model A. e. The routine availability of detailed pressure
distributions allowed an extensive study of various
flow phenomenae such as the development of areas o f
Prior to the design of the model under consideration controlled separated flow in slat and flap shroud
NLR had installed a boundary layer blowing system in cavities, the movement of stagnation points as a.
their low-speed tunnel which produced near-two- function of configuration parameters. An analysis
dimensional flow conditions in the tunnel test section, of boundary layer conditions on a routine basis was
also at very high lift conditions. (Ref. 12).
27-10

of course not possible with the available test

data.

Fig. 19a Effect of single-slotted flap shape. Fig. 19d Effect of double-slotted flaps.

Fig. 19 Windtunnel test data from Model A - Lift

curves.

V DEVELOPMENT OF THE SLATTED VERSION OF THE

F-28 FELLOWSHIP. THE F-28 MK 6000

As the Maximum Take-off Weight of the F-28 Mk 1000/2000

had increased from its initial design value of 54000 lb
to 65000 lb and engine thrust had not changed, a need
-4S l a t 1 was seen to improve the aircraft's take-off and landing
performance.
Slat2 Slats seemed to offer one possibility.
Therefore, in 1970 a study was initiated to analyse the
6, various possible ways to improve the F-28 field
0 ' 5 10 1s eo
(dag)
25 performance.
This resulted in the definition of 5 slat shapes, slats
A to E, to be investigated on the two-dimensional F-28
Fig. 19b Effect of slat shape. model, Model 5-2. (Figure 20).
These slat shapes were again defined without any use of
theory but under consideration of some design rules
based on previous experience such as minimum slat
trailing-edge angle for stiffness, sufficiently long
Flap 2 lower surface to allow for stagnation point travel with
Noae 3 increasing angle-of-attack, etc.).
3.6

F-28

FT
- Model 5-2

r
z
stat A

Slat D I

Fig. 19c Effect of fixed leading-edge shape.

Fig. 20 Slat shapes investigated on the F-28 2-D

model.

Some significant test results from this investigation

are presented in figures 21 and 22.
The most important conclusions from this investigation
were:
 27-11

F - U 2-DModel 5.0 F-28 2-D Model

F-ed 2-0fiod.l
%la,

4.0

18
slat
angte
3.0 25 deg

2.0

1.0
slat slat
aap overlap
Fig. 21 F-28 2-D model - Lift curves for various slat (Kc)
24s
(9.4
-1.50
settings. --.n - -1.55
2.52 -110
0 ----A -0.60
J
0.5 1.0 1.5 2.0 2.5 -63.0
Rc,.lO
a. Increasing the slat chord from 16% c to 19.5% c did
not increase the maximum 1 ift coefficient. (Both
for slats D and E when compared with slat A). b. Effect of Reynolds number on CLMx.
b. For maximum lift the most attractive slat
deflection seems to be 25 deg. with a fairly wide Fig. 22 F-28 2-0 model.
gap and underlap.
c. Decreasing the slat angle and slat gap and underlap The testdata from these 2-D tests were considered of
lowered both the maximum lift and the drag. sufficient interest to continue investigations on a
three-dimensional model. The scale 1:12 F-28 LST-
d. At low lift coefficients their is no smooth flow model was modified in 1971 to incorporate slats and
through the slat gap. When this flap condition tested subsequently. Initially 7 slat positions were
occurs the lift coefficient drops rapidly and the investigated on this model, Model 8-5. Slat A from
drag increases very rapidly at decreasing the 2-D tests was adopted as the slat shape to be
angle-of-attack. investigated resulting in an unmodified wing shape
with the slat retracted. These initial tests produced
two surprises:
a. CLmx decreased when the slat angle was decreased
from 25 to 15 degrees. Improvement of the wing
root fairing shape improved CLmx significantly,
however. (Figures 23 and 24)
b. The drag increase due to slats at take-off
conditions had a dramatically high level
with the slat angle at 25 deg and was
significantly lower at slat angle = 15 deg.
(Figure 25).
In the end 25 slat positions were investigated
before it was accepted that additional measures had

I Slat
gap
L%C)
0 -1.1
0 -ai
a.i.1
A -46
Slat
overlap
(X9
2.M
2.52
t53
4.55
to be taken to arrive at acceptable aircraft lift-
drag ratiols.
It was clear that the large number of slat brackets
(rails on the real aircraft), simulated bleed air
pipes for de-icing, etc. had a large extra drag
contribution but this could not explain the large
2.5. 4 0 t0.5 1.m effects of slat angle.
Clearly some three-dimensional effects affected the
flow picture to a considerable extend but a true
explanation could not be found.
eo- Adopting finally the slat configuration given as
position no. 5 on Model 8-5 the additional measure
taken to obtain the desired lift-drag ratio’s
consisted of a wing span increase (0.75 m span
extension at each tip. This configuration was then
a. Effect of slat setting on CLMX. tested in the NLR bow-Speed Tunnel as Model 8-7
at Re, = 1.4 x 10 , in the High-speed Tunnel a2
Model 6-4 at Reynolds numbers up to Re = 2 . 8 x 10
and on the F-28,halfmodel Model 7-8 [which since
the development of the F-28 Mk 1000 had been
27-12

equipped with a 5-componeft balance) at Reynolds

numbers up to Re, = 5 x 10 . e.

Apart from C and C,/C, at take-off and landing

2?
2.2.
F-28 - Modal 8-5
#
conditions t h y windtunnel tests on these models 2.0.
centered on two other design requirements:
1.8.

1.6.

CL : F-28 -
3.0.
Model 8-5
1.4.

Tailplanr and Tallplans and

elevator in neutral
position. (I&.o)

2.5.
i P

slat gap ~ M X C Slat gap 1.99'4 c

Slat overlap -1.30% Slat overlrpOBo%c

Fig. 23 F-28 Model 8-5. - Effect of slat angle on

CLKiX' Fig. 25 Effect of slat angle on drag - F-28
Model 0-5.

a. securing satisfactory stalling characteristics,

both in roll and in pitch. Concerning the latter in
particular post-stall behaviour required attention.
b. Securing undisturbed engine operation at high
angles-of-attack.
Stalling characteristics received a great deal of
attention. Many different devices such as fences,
stall ing strips, stall promotors and spring loaded
closing plates, the latter partially or completely
closing the inner wing slat gap, were investigated
Original wing root at different Reynolds-numbers on the various
fairing. 1.5
models. Also various wing-root fairing shapes were
analysed. The best compromise between a high C,
-
Slats
deplected
and good stalling characteristics was thought to r e
obtained with the fence and stall strip as were
present on the Mk 1000/2000 and a small stall
stat angle tsdeg promotor when the slat was extended. Also the wing-
Slat g a p 1 9 5 Y . c root fairing which became exposed when the slat
S l a t over lap-Q8O%c extended had to be tailored carefully. The
05
possibility of the need for a stick-pusher to
Def unitive prevent a locked-in stalled condition was
win root fabring recognized however.
R e E -1.4'10
D e f i n i t i v e wing Proper engine operation at high angle-of-attack
root F a i r i n g . became questionable when it was noticed that at
that condition a strong vortex, springing from the
(deg) inboard slat end, entered the engine intake. Intake
flow analysis with the aid of total pressure rakes
showed that this vortex could be suppressed
Fig. 24 Effect of wing root fairing shape on CLmx. significantly by instal ling a Kruger-flap between
the fuselage and the slat.
 27-13

When the full-scale development of the slatted

version of the F-28 under the designation F-28
Mk 6000 was undertaken (for a description see
ref. 13) the following aerodynamic configuration
was adopted:
a. It was based on the F-28 Mk2000.
b. The wing span was increased by 0.75 m span
extension at each tip.
c. Wing slats were adopted with a single extended
position for both take-off and landing as close
as possible to slat position no. 5 on Model For &finition
8-518-7, given minor restrictions from
practical considerations.
d. A spring-loaded stall promotor on the inboard
wing would be added. This decreased the slat
ga with the slat in the extended position.
PFigure 26)
e. A wing-root Kruger flap was added. (Figure 26) Fig. 27 Effect of Reynolds number on drag due to
slats
f. The wing-root fairing at the leading edge was
tailored for optimum inboard wing separation
control near the stall. This affected the of-attack excursions at the stall with very course
pitching moment in the stall in a positive way. control inputs a stick-pusher was instal led.
(Figure 24). Figures 28 to 31 show comparisons of some
significant parameters as measured in the
g. A stick pusher would be developed. windtunnel, as estimated for the full-scale
aircraft and as derived from flight tests.
From the development of the high-lift
characteristics bf the F-28 Mk 6000 the following
lessons were drawn:
a. Again it became clear that small details on the
high-lift devices had to be reproduced
accurately.
W
b. Significant three-dimensional effects can occur
in the flow about wing slats on swept wings,
in particular concerning drag.

Slat
U etrvator in n.uhal
poaition.

Fig. 26 F-28 MK6000 - Stall control devices.

Further windtunnel tests at higher Reynolds numbers

on both a complete model (model 6-4) and on a half
model (Model 7-717-8) showed a marked effect of
Reynolds number on the drag due to high lift
devices, in particular at lift coefficients of
interest. (Figure 27)
When the prototype F-28 Mk 6000 was flight tested
we were in for a new surprise. The maximum lift
coefficient at full-scale conditions turned out to
be much higher than estimated. In fact it was
higher than considered to be useful.
Therefore in a very early stage of the flight test
programme the slats were slightly retracted on
their circular tracks from 15 deg to a slat angle
of 13 degrees in an effort to decrease drag at
take-off. The maximum lift turned out to be hardly
influenced by this modification.
Furthermore both stall promotor and Kruger flap
turned out to be unnecessary. Lateral
characteristics including aileron characteristics # D d i n i t i r m wing root fdiring
turned out to be excellent up to extremely high 0 5 10 1s 20 E6
angles-of-attack. Pitch stability in the stall at all (**a
extreme aft centre-of-gravity position was small
but positive. To prevent inadvertent large angle-
27-14

8% F - 2 8 Mk 6000
3.4 -
3.2. Slats
ert ended
Stick paher
3.0.
1s. 20.5 dag
2.8.
2.6.

24. /

SLltb deC1ect.d.
2.2.
0
P L r , 1g

Slat angle 15d.g

Slat gap 1 . 9 6 % ~
Slat ovcrlrp - 0 . ~ 0 % ~
Stall promotor Fdttad.
at f 2 % F

Flap
0 10 20 30 40 mgte (deg)

F:ig. 29 -
F-28 MK6000 Full scale lift curves and
maximum lift.

cLmax
3.41 P I scale
1

,3.3 -
Fig. 28 -
F-28 MK6000 Lift curves of the LST model,
3.2
Model 8-7 and the HST model, Model 6.-4. 3.1.
Tailplane and
elevator in neutral Scale
3.0. position.( i,,Je.o)

2.9.
Flap mgle
F-20 nk6ooo
28.
T a d p i ane and elevator
in n e u t r a l position 2.T.

.2.6. slat angle = l S d q scale

S l i t angle 13de Flight test: d a t a :
2.5.

F ig. 30 Effect of Reynolds number on maximum

lift. 1
c. Tailoring stalling characteristics of an
aircraft equipped with slats in combinat ion I
with a T-tail requires great care. Again
windtunnel tests at high Reynolds numbers seem
mandatory.
d. Significant Reynolds-numbsr effects may s t,i1 1
occur above Re, = 5 x 10 . I
 27-15

12.
11.
10.

9.
8.
4 Fokker F-28Mk 6000
?'. -0 AircraFt 'D'
4Alrcraft 'B'
6.

5i
f0 0.5 1.0 1.5 2.0
kltg Fig. 32 High-lift devices tested on airfoil section
NLR 7301.
Fig. 31 Lift-drag ratio's with extended High-Lift
devices .

V H I G H - L I F T CHARACTERISTICS OF ADVANCED TRANSONIC

AIRFOILS

a. AIRFOIL SECTIONS
In 1968 Nieuwland and Spee at NLR published their
results that proved both mathematically and
experimentally that transonic flow with complete
isentroDic recomDression existed as a stable flow
condition.
To define airfoil sections for the windtunnel
investigation a design method was developed at NLR
based on hodograph theory. With this method a series
of so-called quasi-elliptical airfoil sections were
developed. Initially only symmetrical airfoil
sections were developed. In later years also cambered
airfoils were designed with this method although this
was a cumbersome process. The first cambered airfoil
was NLR 7101 with 14.1 percent relative thickness
followed in 1973 by NLR 7301 with 16.5 percent
relative thickness. As these sections had very thick
leading edges it was decided to analyse the high-lift
characteristics of section NLR 7301 with a number of AirFoll sectlon NLR l 3 0 1
high-lift devices in what became in the end an
extensive test programme lasting from 1974 to 1977. t . . . . . . . .
Main flap -la
. . . ,
(dq)
The tests were performed in the NLR 3x2111Low Speed o I 90 (5 eo cs w 31 4(3 4s 30 5 1 60
Tunnel on a model, spanning the tunnel from floor to
ceiling, having a chord = 0.570 m giving a maximum
Reynolds no. = 2.6 x lo6. Wall blowing was applied.
The slat and flap configurations investigated are Fig. 33 Airfoil section NLR 7301. - some maximum lift
shown in figure 32. data.
Lift, pitching moment and drag were again obtained
through static pressure integration over the
component surfaces and through wake rake
measurements. In the latter 3 wake profiles were
measured for each data point. Some maximum lift data
are presented in figures 33 and 34.
Note that with the double-slotted flap with movable
vane a CLmX = 5.00 was reached when a slat was
fitted.
27-16

A i r f o i l section NLR 2301

C. 677- I

/--
:I
S l a t angle 15&g
Fowler f l a p
dePlrcttd 3 0 deg
4.2

.2.4 Slat angle Is dcg

El 10 0
0 lo 2:o -6
Re, x l 0
3.0

Fig. 34 Effect of Reynolds-number on maximum lift. -

section NLR 7301 with slat and flap.

The results obtained in these tests clearly indicated

that higher CL,,-values could be obtained than on Fig. 36 2-D models with high-lift devices tested
earlier sections investigated. The thick leading between 1978 and 1981.
edges to which this was attributed presented a
problem for high-speed flight however. Although the
pressure distribution adopted had in theory zero wave This configuration was investigated in 1978 and 1980.
drag at the design condition low-Reynolds-number Because of the large spanwise variations in drag
tests showed a considerable drag creep at Mach found in earlier tests the drag measurements on
numbers below the design Mach Number. Some tests were section NLR 7703 Mod were based on wake profiles
perfcpned at the Lockheed CFF facility up to Re, = 30 measured at 8 spanwise stations. In figure 37 some
x 10 which showed that at higher Re-numbers the maximum lift data are presented in comparison with
drag creep did decrease. Today, this can also be data from section NLR 7301.
calculated with Drela's programme ISES (ref. 14).
(See figure 35). A second modification with an even smaller leading
edge radius was also investigated with a slat in
A i r f o i l section 1980. This configuration, Model F-29-16 is
Ct G 0.4 illustrated in figure 36b.
50 !I

:I
Finally in 1981 a two-dimensional model, F-29-12 was
40 1 ---
I
T h e o r y (ISES)
Re-. 2 I( 10:
r. . l o x lo!
designed based on the same section as Model F-29-16
but with a 30 percent double-slotted flap. This model
was intended for high Reynolds-number investigations
at varying Mach-numbers in the NLR High-speed Tknnel
(HST). Rfynolds numbers between Re, = 1.8 x 10 and
10 7.0 x 10 could be obtained.
Lift, pitching moment and drag were obtained in the
same way as in the LST.
The airfoil section with its high-lift devices are
shown in figure 36c. Some test data are presented in
0 0.50 0.60 O.?O fl 0.80 figure 38.
Fig. 35 NLR 7301 - Effect of Reynolds number on drag
creep at high Mach numbers. Apart from overall data this test produced insight in
the effect of specific details in the flow such as
the effect of Mach-number and Reynolds-number on the
In 1977 however it was felt that a high peak Mach- maximum local Mach-number at the leading-edge suction
number (Mlocal= 1.30) in the supersonic flow region peak. As an example figure 39 is presented.
near the nose of the section would introduce the
unacceptable risk of a large transonic drag creep. Throughout the period described in this chapter slat
Therefore new sections were designed which produced and flap shapes were determined empirically based on
lower peak Mach-numbers near the leading-edge in the previous experience.
design condition. This led to lower leading-edge
radii. One such section was NLR 7703. In order to Finally, in the second half of the 1980's Fokker has
investigate the effect of this thinner leading edge participated in a GARTEUR high-lift research
a modification was performed on the LST-model programme.
changing the section to a quasi NLR 7703 section
indicated as NLR 7703 Mod.
Figure 36a shows the new section. The largeg chord (c
= 0.677 m) leads to a maximum Re, = 3.1~10 .
 27-17

A i r F o i l section NLR Ro5 r d .

4.4.

4.2.

4.0.

3.8.

S6

3A.

L2. 32.
N o slat
10.
8.0.
-
~/
28
tb.
26.
1.6. n a p retractad
14 _.
- 2.4

41at a n g l e 3sd.g

Fig. 37 Airfoil section NLR 7703 MOD with high-lift

devices. - Effect of Re-no on maximum lift.
\
Model F-29-12 z7
t.8
-.
U U -
Clean airfoil M a d u i F-29-12

'I
618 020 &?2 624 O h s 0.26
M
CLnax Fig. 39 Effect of Mach-number and Reynolds-number on
maximum lift.

No slat b. AIRCRAFT CONFIGURATIONS

Between 1975 and 1982 an extensive research and
development programme concerning a successor to the
3.2 F-28 was executed in the Netherlands. The programme
started with a windtunnel investigation both on two-
and on three-dimensional models to analyse the
3.0 aerodynamic characteristics and the associated
possible economic benefits of the newly developed
advanced transonic airfoils. Nieuwland .and Spee,
Garabedian, Korn and Bauer and later Jameson had,
together with a number of British researchers and
helped by the advances in digital computer technology
in the mid-seventies made it possible to design
airfoils with a considerable area of supersonic flow
on the upper surface.
1.4J,
0.5
, ,, I

0.8 1.0
I - I ' I
2.0 3.0 4.0 5.0 EO 10.0
I I ' " i
Transonic pressure distributions were not entirely
ReEU 10" new. Aircraft such as the Douglas DC-8, DC-9, DC-10,
Lockhead C-141 and L-1011, De Havilland Trident and
Vickers VC-10 Boeing 737 and 747 and Fokker F-28 and
Fig. 38 Model F-29-12 - Maximum lift. even the later versions of the Boeing 707 at high
weight had local areas of supersonic flow terminated
Model F-29-12
with a weak shock on their wing upper surface at
%in __8-\ slat
angle a&
Fla
cruise conditions (see ref. 15). The developments
just mentioned allowed however to extend the area of
supersonic flow to at least 50 percent o f the wing
0 retr retracted chord over most of the wing span in a controlled
design process eliminating much of the earlier cut-
and-try approach. This allowed either a thicker wing
-20. with higher aspect ratio or a thicker and lighter
- 18 . wing or a higher wing loading.
-16. As the Fokker company concentrated historically on
the short-to-medium haul market and field
-14. performance was felt to be as important as cruise
-12.
performance the possibility to increase wing
thickness was considered to be o f greater benefit
M .0.19 than increasing wing loading and thus design lift
coefficient. Thus the initial wing designs were based
on the experience obtained with airfoil sections such
as NLR 7301 of 16.5 percent relative thickness as
described in the first part o f
27-18

this paragraph. In the course of the aircraft To investigate low-speed characteristics three
development process which went from a highly modified complete low-speed models were produced for the NLR
F-28, the Super F-28 via the F-29 to the McDonnell Low-Speed Tunnfl (LST). These models, tested at
Douglas-Fokker MDF-100 a gradual shift in the balance Re, IJ 1.0 x 10 were:
between cruise and field performance led to wing
designs which became thinner with less blunt leading a. SKV-LST-1, with the wing shape SKV-3
edges. As an illustration figure 40 presents a three- b. SKV-LST-2, basically identical to
view drawing of the F-29 at a certain stage, project model SKV-LST-1 but with a
P-325 from January 1980. This trend in wing design modified inner wing resembling that
history found its parallel in the development of the o f wing SKV-5.
airfoil section models including their high-lift c. SKV-LST-3 based on wing shape SKV-6.
devices as mentioned before.
In figure 41 the three wing planforms are presented.
Figure 42 shows the overall shape of model SKV-LST-1.
F29 GENERAL ARRANGEMENT Q!=
All three models, tested in 1977, 1978 and 1979 could
be equipped with slats and double-slotted (tabbed
Fowler type) flaps, the latter spanning 75% of the
wing span. Slat and flap positions could be varied.
Furthermore aileron power, many different spoiler-
panel configurations, engine positions both on the
rear-fuselage and under the wing were analysed.
Stalling characteristics, both in pitch and roll were
also investi ated. (All three models
had a T-tailg.

Fig. 40 Fokker F-29. - Project P325.

Model SKV-LST-2
The following table i 1 lustrates the development
history as characterised by a number of design
parameters for each high-speed model wing.

Year Conplete
Windtunnel
mode 1
Aspect
Ratio
Wing
sweep
(t/c)
Kink
(51)
,,El C,
DES
1
1977 SKV-3

Fig. 41 Wing planforms of models SKV-LST-1, -2, -3.

The effect of Reynolds-no variation on the low speed

characteristics of model SKV-LST-3 was analysed by
testing the actual left-wing of this model at high
Reynolds-number as a halfmodel in the HST tunnel of
1981 F-29-4 NLR. This model, designated SKV-7 could also be
tested in combination with a blown nacelle. This
a1 lowed the analysis of engine-airframe integration

1 1
1982 F-29-19 effects including the measurements of ,static and
dynamic flap pressures.
Furthermore three half models were tested in the HST
to analyse Re - no effects. The test set-up of model SKV-7 with blown nacelle has
been described in ref. 16. In this test the nacelle
Mode;
desi nation
5; : ; Max t e s t
sha e Re-No a t M - Max.-test
0.19 Re-no a t M - 0.75
plus pylon and strut (through which high-pressure air
was fed) was physically isolated from the wing. The
wing was fitted on the 5-component balance. Engine
SKV-Sij 4.9 x 108 10 x 10'
strut and halfmodel were coupled however outside the
SKV-7 (-1 t o -6) SKV-3 3.6 x 10' tunnel test section so the combined structure could
F-29-10(-1 t o -6) F-29-2 5.0 x 10' 10 x 106
be set at varying angle-of attack. (Figure 43)
 a 27-19

3.5 Aircraft-less-toit
Wing-mounted engime nocollrs

-. Slats extended
(Slat lngk fOdeg)
Complete model
Modal SYV-LST-3
----A Half modal
Fig. 42 Model SKV-LST-1. Model S U V - 7 - 1

5 10 15 20 25 30
% (deg)

Fig. 44 Comparison of lift curves from halfmodel and

comp 1 ete mode 1 testing .

Finally in 1981 and 1982 the models F-29-1 (figure

45) and F-29-2 (the latter without slats) were also
tested in low-speed configurations, both with a
fuselage-mounted horizontal tail and with a T-tail.
The low-speed wing characteristics of model F-29-2
were checked at high Reynolds-numbers on the
halfmodel Model 10. In its later configuration this
model was also investigated with a slat and a wing
leading-edge shape closely resembling that of Model
MODEL SKV-7 WITH WAKE RAKE AND K U L I T E S ON FLAP F-29-5.
In figure 46 C tail-off is presented as a
function of Reynoflt'number and flap and slat setting
for a number of models.

Model F-29-1 /? P r o j e c t P-325 A3

XODEL SKV-7 WITH LOW TAIL

Fig. 43 Half model testing with blown nacelle.

Fokker has in general good experience with the use of

halfmodels (figure 44 is given as an example). For
accurate data an extensive matching procedure is
required however (see ref. 17).
Fig. 45 F-29 - Model 29-1.
27-20

Modal based on initial experience with the 2-D NLR 7301

4FZ940-2 Clean teadine-edde
...-.0 ~n-10-5 Droop leadingedge model. These initial 2-D tests suggested that the
optimum slat posit ion wou Id be:
slat angle 15 deg
0- slat gap 3.5% local wing chord
I slat overlap -1.5% local wing chord
As figure 47 shows this produced an impressive CLMX
= 3.64 for landing at Re, = 1 x lo6.
After more extensive tests on the 2-D model the
conclusion was reached that a better slat position
would be:
1 - M-0.19 . 1 M -0.19 slat angle 10 deg.
slat gap 2.5% local wing chord.
slat overlap 0
This slat position was investigated on model SKV-
Model
LST-2. Contrary to our expectations this latter
model, which had an identical geometry as model
SKV-LST-1 except for the inner wing, showed for all
configurations a much lower C, than the previous
model: for landing ,,C = 9 . 3 6 was obtained
(Figure 48). Due to the s5arper leading-edge C,
is even lower (CLmax= 3.26) on model SKV-LST-y.
(Figure 49).

4.0, M o d e l SKV-LST-1

cl
32- M, M o d e l F-29-10-6
Mode 1

20
4 NO Sl*
---0 Slats retracted

Fig. 46 Effect o f Re-no on maximum lift. - Various

halfmodels and complete models.
Fig. 47 Lift curves of model SKV-LST-1.
The investigations into the high-lift characteristics
of advanced transonic airfoils, both for 2-D and in A careful anabysis of the 2-D data showed that at
3-D configurations tought us the following: Re, > 1 x 10 the smaller slat gap gave indeed
higher increases in maximum lift than the large
1. Modern transonic airfoil sections can combine gap. At the lower Reynolds-numbers (and this is
good high-speed characteristics with high where the wing tip region, which stalled first,
maximum-lift for take-off and landing operated at in the windtunnel) the large gap
configurations. produced higher maximum 1 ift values (figure 50).
2. In particular for high design cruise Mach-numbers This means that, had the configuration of model
and lift coefficients a very strong interrelation SKV-LST-1 been investigated at higher Reynolds-
exists between these design cruise parameters and numbers the increase in CLmX had been considerably
the achievable high-1 ift characteristics. Leading- lower.
edge and trail ing-edge shapes depend heavily on
design cruise condition. Similar Reynolds-number effects were noted when a
"droop leading-edge" was investigated on model
3. Reynolds-number does not only affect overall F-29-10.5 (Figure 46).
characteristics such as lift, drag and pitching These results and data from tests with deflected
moment but also optimization procedures such as slats without a slat gap suggest that for sections
finding the optimum slot shape for slats and flaps. with relatively thick leading-edges and on wings
This is illustrated in the following example: with a moderate sweep angle these high-lift devices
Slat and flap configurations for take-off and are not very effective.
landing had been determined for Model SKV-LST-1
 27-21

Model SKV-LST-2
Aircraft -1~ss-tall slat
3.5

I p”a”” 1.4 -
1.2 -
1.0-
-
0 1.0 ’ -6 3.’0
2’o Re,xfO

---e S l a t s retractad
-0 Stat%extended
(stat angle iodeg)

I Y
0.6
a d o i l slat overlap
section angle
cdeal
2.5
Fig. 48 Lift curves of model SKV-LST-2. 3.5
0.2 22.5
2.5
Mode\ SKV-LST-3 0
Aircraft -less- t a i l
I5 Fuselrge-mountad nacelles

Fig. 50 Effect of Re-no on increase in maximum lift

due to slat extension.

4. Engines fitted under the wing can be effectively

used to control flow separation patterns at wing
stall. Provided that great attention is paid to
detai 1s of the pylon-slat-leading edge intersection
the loss in lift can be made negligible. (Figure
51). The latter used to be a strong argument for
rear-fuselage engine positions in the past.
5. A large number of, sometimes interrelated, design
parameters have to be considered when choosing and
---0 Slats retracted
I optimizing highlift configurations (See, for
--a Slats e x t e n d e d example figure 52 which shows the engine nacelle
(Slat dngk lodes) drag to be a function of both lift coefficient and
flap setting when the nacelle is mounted under the
wing.
6. As in the windtunnel development of the F-28
Mk 6000 a strong interrelation was found between
increases in maximum lift and in drag for various
slat configurations (Figure 53).
Fig. 49 Lift curves of model SKV-LST-3.

Fig. 51 Effect of slat cut-out on stall pattern and

maximum lift.
27-22

M o d e l SKV-LST-3 V I THE HIGH-LIFT DEVELOPMENT OF THE FOKKER 50 AND

FOKKER 100
50.
e a. THE FOKKER 50
P.5.
The design studies on the Fokker 50 (as on the Fokker
100) started in 1982. The Fokker 50 was to be
equipped with Pratt and Whitney PW 125 engines
20. producing 2500 shp in take-off and go-around (10
percent more than the most powerful Rolls-Royce Dart
engine on the F-27) and with new propellers. In order
t5. to obtain maximum insight in the stability and
control characteristics of the aircraft at the
highest propeller thrust coefficients a 1 : 5 scale
10. windtunnel model with a wing span of 5.40 m was built
and tested in the Dutch-German Windtunnel DNW in the
8 x Q m test section at Reynolds numbers up to Re,=3.2
.
x 10 The model was equipped with turbine-powered
s imu 1 ators driving var iable-pitch prope 1 1 ers through
a 2.7 : 1 reduction gear box.
0. In an aerodynamic sense wing and flap systems
: 2 b J W remained identical to those on the F-27. These tests
provided detailed insight in the flow conditions at
the tail at high thrust coefficients. Also tests with
Fig. 52 Under-wing engine nacelle drag. modified elevator, rudder and ailerons, the latter
now equipped with hornbalances showed that the
improvements sought could be realised. Flight tests
were performed in two stages. On an old F-27 bought
by Fokker for experimental purposes, the modified
control surfaces were investigated. On the two Fokker
50 prototypes the overall performance and f 1 ight
handling characteristics were analysed.
Overall the flight test results were in accordance
with predictions. Maximum lift coefficients and
stalling characteristics of the F-27 were repeated or
slightly improved upon, again without the use of any
stall control device. The only difference in this
respect (apart from different take-off flap settings)
O V SKV-LST-2 is the decrease in maximum flap deflection from 40
deg on the F-27 to 35 deg on the Fokker 50.
The flight handling characteristics in roll and yaw
were improved compared to the F-27 but to obtain
satisfactory longitudinal hand1 ing characteristics
over the desired c.g. range a "smart" bungee was
developed which applies a constant force in the
longitudinal control system the height of which is
dependent on engine power and flap setting.
As on the F-27 not performance but flight handling
70- provided the biggest challenge at high lift
2 conditions.
CL Model F-29-10-6
6.0. n I 0.19
6
b. THE FOKKER 100
5.0.
- Slat rctractad The aerodynamic development of the Fokker 100 was
-0- Slat angle 10deg described in ref. 18.
4.0. --A- slat angle fsdeg In order to obtain an early insight in the effect of
the proposed leading- and trailing edge modification
to the F-28 wing which would lead to the Fokker 100
wing, the old 2-0 F-28 model, Model 5 was modified
3.0. again and tested in the NLR Low Speed Tunnel in
Amsterdam as Model 5-3. Figure 54 shows the
improvement obtained compared to the F-28 both for
2.0. the clean section and for the high-lift
configurations. This figure also presents some
theoretical predictions. As in the mean time
t.0. theoretical prediction methods (covered in the next
paragraph) had improved considerably these were for
the first time used in the actual design process not
0
only for the 2-D analysis but also on the complete
50 100 I 5 0 & Z h 300 Jso 4bO 4bO C wing in the clean configuration. The analysis of the
ACDrfO spanwise distribution of leading-edge suction peak
with increasing angle of attack helped to predict the
Fig. 53 Effect of slat setting on low speed drag. initial stalling characteristics (Figure 55). The
agreement between predicted and actually achieved
 27-23

full-scale maximum lift is excellent. However, again V I 1 THE PREDICTION OF H I G H - L I F T CHARACTERISTICS

a lengthy flight test programme was required to - THE ROLE OF THEORY
obtain the best fence and stalling strip location for
an optimum balance between performance and f 1 ight
handling characteristics, the latter including tail Computational fluid dynamics, (and, in the days before
buffet in the stall. The certified CLm -values for the digital computer, aerodynamic theory) can be
all flap settings is presented in figure36. Note the divided into two working area's:
high ,C -values available for take-off

-
notwithstanding the absence of a slat. 1. The mathematical description and analysis of a I

r F o k k a r F-28
given situation concerning a body moving in a
fluid.
11. The determination of the limits of validity of the
description and analysis of this particular
Fokker 100- situation.
CY] The first area concerns for example the attached flow
36 F l i p angle 4 2 d e g about a wing section at a given angle-of-attack or the
stable leading-edge vortex flow on a sharp-edged
slender delta wing.
The second area concerns the boundary-layer separation
on a two-dimensional airfoil at reaching C or the
vortex bursting on the sharp-edged slender hgta.
Pressure distributions and boundary-layer conditions in
attached flow have been analysed since the dawn of
aviation. Analysis of controlled vortex flow started in
1940 by Jones at NACA.
06'
' (5 20
r(deg)
G 2 4

The determination of boundary layer separation on two-

Fig. 54 Fokker 100 - Improvement in 2-D maximum lift. and certainly on three-dimensional bodies on a
theoretical basis led to meaningful results only in
fairly recent times. This is notwithstanding the fact
that for aircraft design this matter is of equal
importance as the detailed analysis of an aircraft's
characteristics in attached flow.
The explanation lies o f course in the far greater
complexity of three-dimensional boundary- 1 ayers in
adverse pressure gradients than of potential flow. A
fundamental theoretical principle such as the
description of potential flow through sinks, sources,
dipoles and vortices is still lacking in boundary layer
theory notwithstanding the great effort being put in
I' turbulence modelling.
For wings, or even airfoil sections, equipped with
1 4 6 8 1.0
flaps and/or slats even the calculation of detailed
pressure distributions and local boundary- layer
me leadingedge sumon peak ar nign 7 conditions with attached flow has not yet reached the
angiw~arracrtor successive wing dSfgn%
desired degree of accuracy let alone their integration
into lift, drag and pitching moment.
Fig. 55 -
Fokker 100 Development of the spanwise
I

-
distribution of the leading edge suction
peak. For practical design purposes purely theoretical
methods for high-lift design contain therefore still
Clma* FAA' too high a degree of uncertainty. However, mixed with
CERTIFIED a proper degree of empiricism they become very useful
2.6
tools, if not in producing the final answer than at
least in guiding the way to the optimum design
solution. But, perhaps even more important, in the case
2.4 - of high lift devices theory has provided insight in the
interrelation between the various components of a
multi-component airfoil. The break -
through, in my
opinion, came with the publication of A.M.O. Smith's
2.2 . eaper:
Aerodynamics of High-Lift Airfoil Systems".
2.0 . in AGARD C.P. 102 in 1972 (ref. 19).
In this paper it was clearly shown that an airfoil with
a slat and a flap consisted actually of a combination
1.8 - of separate close-coupled airfoils with a strong mutual
interaction. Suddenly the relations between suction
. ~~~ . peaks, dumping velocities, upwash and downwash and the
0 10 20 30 Po 50 resulting pressure distributions and boundary layer
conditions fitted into a pattern.
We understand now, for example, why slats with small
leading-edge radii have to be deflected further than
Fig. 56 Fokker 100 - Certified CLmx. slats with large leading-edge radii and why small-chord
 27-24

flap vanes are ineffective. It is now also clear why

each slat has an optimum deflection. Too large slat
angles produce too high supervelocities on the main
component (figure 57).
The understanding of the physics led to an
understanding of the critical areas in the pressure
distribution.
As an example figure 58 illustrates that on the double-
slotted flap 3 a vane position A leads to a too low
vane loading and a too high flap loading. Consequently
the flow on the flap stalls. Vane position B gives a
much better loading division and consequently. .produces
a higher CLmx (see-figure 33).
But. as fiaure 58b illustrates, for practical DurDoses
an empirical design rule as presented in 1961' (the
tangent to vane and flap upper surface) leads to
s imi 1 ar concl us ions.
On single airfoils trailing-edge separation has been
studied on a theoretical basis (with some empiricism
such as a constant static pressure in the area of
separated flow) since the 1960's. At NLR C was
calculated as a function of Reynolds-number in 18% for
the airfoil section NLR 7301. Both a trailing-edge
separation criterion and a (very crude) leading-edge
criterion were considered.

BOUNDARY LAYER
-i5.0>
AND

=P FLOW CONTROL
Edited by PERGAMON PRESS
G. V. LACHMANN OXFORD ' LONDON ' NEW YORK ' PARIS
1961
-1ao.
THE DOUGLAS DOUBLE-SLOTTED FLAP
By W. H. Kunuun

-5.0 -

0.
* 1.0 A.

Fig. 57 Effect of slat deflection on peak

supervelocities on slat and mdin body.

-
b. \ %
Since 1985 a computerprogramme, ULTRAN V is available.
This programme although based on transonic small Fig. 58 Interaction between the circulation on the
perturbation theory coupled to a boundary layer theory vane and the flap on a double-slotted flap.
gives a very accurate description of the flow
conditions over the rear part of the airfoil section
including a separated boundary layer. This allows the determined. By comparing the data from computions with
determination of maximum lift due to trailing-edge both the programmes ULTRAN V and TRAFS the maximum.lift
separation. A transonic full-potential flow solver, the of single airfoils can be estimated. Some data are
programme TRAFS a1 lows, when corrections are applied presented in figure 59.
for boundary- 1 ayer effects through the concept of
displacement thickness, an accurate calculation of the In the early eighties all available 2-D test data on
leading-edge suction peak. By means of an empirical high-lift configurations available at Fokker's were
correlation between minimum static pressure as a analysed systematically. This led to the conclusion
function of Re- an Mach-number and leading-edge that on high-lift configurations CLmx could be
curvature the occurrence of leading-edge stall can be determined by either of 4 types of flow separation.
These are illustrated in figure 60.
 27-25

2.2, NACA 6716V

N o slat

l.8.
1.1. Maximum peak
s u c t i o n coefficient
1.6. -10 onthe main
airfoil component.

Slat

Maximum peak
Qmax. test suction coefficient
-10 - on the slat

Fig. 59 Maximum lift of clean airfoils. - Theory and

experiment.

Fig. 60 Four types of flow separation determining

maximum lift.

Isaqiaz
@ QWaI due to Leading-edge Cpm,sLat
5tau on slat

Modern theory can reasonably d l cope with trailing- CL;., OrChI

edge separation provided no confluent wakes and Slat I.CfUS

boundary layers occur.

%",M*,", C P l S M.i" 2
At NLR a computer programme VISWAKE was developed @ slwa,due to claw cp&a, Miin
(B. Oskam et al, ref. 20) which allows an accurate -rration 01 main conpol.nC CO-p.
computation of the pressure distribution and lift, drag
and pitching moment on multi-component airfoils as a
function of angle-of-attack including maximum 1 ift, Fig. 61 Semi-empirical procedure to determine
provided C, was determined by flow separation at the CLmx of a multi-component airfoil
trai 1 ing-e&e of any component. sect ion.
27-26

Of particular significance in this programme is the

modelling of the wake of each component including the
strong interaction with the pressure field around the
next airfoil component.
The large data base from windtunnel tests allowed the
determination of some empirical criteria for the
maximum height of the leading-edge suction peak before
leading-edge stall occurs on slat or main component as
a function of leading-edge curvature and dumping
velocities. Accurate determination of leading-edge peak
velocities at a given angle of attack is required of
course.

In figure 61 the character of the empirical criteria is

presented together with the development of the various Measured and calculated pressure
significant parameters on a particular configuration as Fig. 63
a function of angle-of-attack. The parameter which distribution.
first exceeds a limit determines maximum lift.
LIFT
Although the programme VISWAKE is a programme based on COEFFICIENT
theory for incompressible flow it produces nevertheless
very reasonable answers concerning maximum 1 ift (See
figure 62). In figures 63 and 64 some examples are
-
e
COMPUTED
MEASURED
aan.
n-
presented of pressure distribution, lift, drag and
pitching moment for a section with slat and double-
slotted flap. 4

40.

t,6 1
CALCULATED I
I I
I
II
I
A i r f o i l Sections with
0 8 1 i n m m a 1 a 1
slats and/or flaps INCIDENCE dt (DIG)
deflected. / .
LIFT As FUNCTION OF INCIDENCE

1 /
LIFT COEFFICIENT MAIN WING

CONTRIBUTION OF VARIOUS ELUIENTSTO L I R

Fig. 62 Calculated CLmx for airfoil sections Fig. 64 Lift on the various component of a multi-
component airfoil.
equ i pped with high- 1 ift devices.

In three-dimensional flow Ckvx-prediction is possible

at least for high-aspect-ra i o wings by combining 2-D
analysis with lifting line theory or with 3-D panel
methods.
Lift, drag and pitching moment below the stall for a
complete aircraft configuration are computed by
combining 2-0 data (in particular for lift and section
drag at higher flap angles this is taken from an
empirical data base) with a non-planar lifting surface
programme (NPLS) .
 I
27-27

An example of the panel distribution and some basic Fokker 5 0

aerodynamic data are presented in figure 65 for the
F-29. Estimated lift-drag ratio's for a trimmed Fokker Estimated l i f t -drag r a t i o ' s
L/D
50 are presented in figure 66.
181 Flaps retracted C.G. &k 20%MAC
Some examples have been presented of the use of theory 16 -
in the analysis and design of high lift configurations. 4 Estimatad From
A great progress has been made in the last two decades
in particular in determining CLmX for such 14 -
configurations. E s t i m a t e d with
12 -
Nevertheless, the accuracy of flow calculations is
still insufficient to determine the choice of an
optimum high-lift configuration for a given set of 10-
design requirements. Here the windtunnel will for the 1.4Vs 1.'3V, I vs
1.2
foreseable future still have the last say.
d

.8 1.0 1.2 1.4 1.6 1.8 2.E 2.2

'L trimmod
Fig. 66 Estimated lift-drag ratio's of the
Fokker 50
CONCLUSIONS

Forty years of high-lift research and development at

Fokker have produced a wealth of testdata combined with
extensive experience in the fields of computation and
windtunnel and f 1 ight testing.
These forty years have shown the following developments
muc f L l P and tought the following lessons:
ANGLE
1. Due to commercial pressures, in combination with
FLAP ANGLE technologies in the areas of computation and
wi ndtunnel test i ng techniques which were
continuously being developed, a steady improvement
in the accuracy of the prediction of full-scale
aircraft performance and flight handling
characteristics has been required and, in many
cases, has been achieved over the years.
2. In view of the above the need to perform windtunnel
tests at the highest possible, or affordable,
Reynolds-numbers can not be overemphasized, also in
the early stages of a design process when the
optimum configuration is still to be determined.
3. The role of computational fluid dynamics in the
design process has greatly increased. However, in
the design of high-lift configurations the actual
component geometry is still to a large degree
defined empirically.
Theory has, however, helped very much to understand
flow phenomenae and has provided guidelines for the
design of high-lift devices.
4. For improved understanding of details in the flow
physics and for application in the design of
moderately swept high-aspect ratio wings 2-D data,
both from theory and from windtunnel tests provide
still very useful data.
5. At present, a fair degree of empirical data has to
be fed into the computational methods in order to
make optimum use of theory. For this access to a
large empirical data base is a prerequisite.
Fig. 65 Aerodynamic characteristics estimated
with a Non-Planar Lifting Surface 6. In many cases the discrepancy between theoretical
and test data for a particular high-lift
Programme ( NPLS). configuration is of the same order as the difference
in aerodynamic characteristics between the different
configuration options one has to choose from for a
given set o f design parameters. In other words: the
final choice of the most suitable high-lift
configuration is still made on the basis of
windtunnel tests.
21-28

7. The above is true for 2-D flow. For a wing-body- 10. van Doorn, J.T.M., Han, S.O.T.H., Obert, E. ,
nacelle combination with slotted slats and part-span "Comparison of Fokker F-28 "Fellowship" Windtunnel
flaps the meaningful use of computational fluid -
and Flight Data. A summary, "in" Yearbook 1973 of
dynamics for design purposes is still limited. the Netherlands Association of Aeronautical
However, this should not deter, but be seen as a Engineers", 1973, paper 5.
challenge for, the CFD-community.
11. Anon, "Fokker F-28 Fellowship", Aircraft
Engineering, June/Ju ly 1967.
AKNOWLEDGEMENT 12. Vos, D.M., "Low-Speed Windtunnel Measurements on a
Two-dimensional Flapped Wing Model using Tunnel
The activities described in this paper were and are a Wall Boundary-Layer Control at the Wing-Wall
team effort. In particular the cooperation between the Junctions", NLR TR 70050 V.
code developers and windtunnel test people at the
National Aerospace Laboratory., NLR and the 13. Anon, "F-28 - Development of the Mk 5000/6000",
aerodynamicists at Fokker should be mentioned. May the Aircraft Engineering, October/November 1973.
following names be representative for this group
effort : 14. Drela, M. , Giles, M, "ISES, a Two-dimensional
Viscous Aerodynamic Design and Analysis Code, Paper
For NLR For Fokker AIAA no 87-0424, 1987.
H. Ti jdeman J.v. Hengst
J.W. Slooff B.J. Warrink 15. Obert, E., "The Supercritical Airfoil - An
B. Oskam J.N. Boer Evolution or a Revolution DGLR/GARTEUR 6
?'I
R. Houwink W.H.A.R. Willemse Symposium on Transonic Configurations, Bad
A. Elsenaar N. Voogt Harzburg, Germany, June 1978.
B.v.d. Berg D.F. Volkers
J.A. van Egmond Tj. Schuringa 16. (Nacelle blowing).
K.W. Muller P. de Boer
17. Boersen, S.J. , Elsenaar, A., "Half-model Testing in
But above all, Prof. J.H.D. Blom should be mentioned the NLR High-speed Tunnel HST: It's Technique and
who was at Fokker in charge of aerodynamics between Application", NLR MP 8303611, 1983.
(roughly) 1947 and 1975 who so much stimulated us to
look. into flow phenomenae and not only at overall 18. Obert, E., "The Aerodynamic Development of the
characteristics and who tought us integrated design Fokker loo", in "Proceedings of the 16th Congress
long before expressions such as concurrent engineering of the International Council of the Aeronautical
and design-bui Id teams became fashionable. Sciences (ICAS). August/September 1988.
19. Smith, A.M.O., "Aerodynamics o f High-Lift Airfoil
Systems", in "Fluid Dynamics of.Aircraft Stalling",
AGARD CP-102, 1972.
REFERENCES 20. Oskam, B., Laan D.J., Volkers, D.F. "Recent
Advances in computational Methods to Solve the
1. Abbott,LH.and von Ooenhoff,A.E, "Theory of Wing High- 1 ift Mu 1 ti -component Airf o i 1 Problem" NLR
Sections" MP 84042U. Also in "Improvement of Aerodynamic
performance through Boundary-layer Control and
2. Abbott,I.H, von DoenhoffA.E.and Stivers,L.S. ,"Summary High-lift Systems, AGARD CP365, 1984 Paper 3.
of Airfoil Data", NACA TR 824,
3. Cahill, J.F., "Summary of Section Data on Trailing-
edge High-lift Devices", NACA TR 938.
4. Young, A.D. , "The Aerodynamic Characteristics of
Flaps", ARC RM 2622,
5. Obert, E., "Low-speed Stability and Control
characteristics of Transport Aircraft with
Particular Reference to Tailplane Design" in
"Take-off and Landing", AGARD CP 160, April 1974,
Paper 10.
6. Blom, J.H.D., "Experience in Predicting Subsonic
Aircraft Character isti cs from Windtunnel Analysis" ,
in "Flight/Ground Testing Facilities Correlation"
AGARD CP167, June 1975, Paper 16.
7. Mc Cullough; G.B. and Gault, D.E., "Examples of
Three Representative Types of Airfoi 1-section Stall
at low Speed", NACA TN 2502, 1951.
8. Gault, D.E., "A correlation of Low-Speed Airfoil-
section stalling characteristics with Reynolds
Number and Airfoil Geometry", NACA TN 3963, 1957.
9. Schuringa, Tj., "Aerodynamics of Wing Stall of the
Fokker F-28", in "The Fluid Dynamics of Aircraft
Stalling", AGARD CP 102, April 1972, Paper 20.
 28-1

High-Lift Research:
Application to the Design of the ATR72 Flap
P. Capbern
AEROSPATIALEAvions
316. route de Bayonne
31060 Toulouse Cedex 03
(France)

I SUMMARY: l'atterrissage). et un gain de finesse au dhollage. qui

Due to slightly reduced Clmax objectives in landing se rbpercutent dzectement sur les performances
configuration for the ATR72. compared to the op&ationnelles, telles les longueurs de pistes
~ ATR42, the high-lift system of the ATR72 is made up minimales et la masse maximale au dbcollage. Ces
of a single dropped hinge flap, whereas the ATR42 performances drodynamiques ont bd obtenues g r k e B
was equipped with a double slotted vane type flap. l'introduction systbmatique. B partir de 1985. des
Elimination of the vane has had a beneficial effect in mbthodes numbriques en phase de conception, en
greatly simplifying the high-lift system for the complbment des m6thodes empiriques et
ATR72. exp6rimentales presque exclusivement utilisbes
jusqu'alors. Le volet ATR72 est en effet le premier qui
This simplification has been achieved while ensuring ait btb essentiellement c o n p par la voie numQique B
that Clmax levels are maintained or improved at same AEROSPATIALE.
flap deflection (up to the value required for landing),
and take-off ratio is improved. which has a direct Dautre part, de nouvelles rkglementations plus
repercussion on operational performance, such as sbvhres pour les avions B hblice ayant conduit B une
minimum runway lengths and maximum take-off restriction importante dam l'utilisation des grands
weight. This aerodynamic performance has been braquages de volet pour l'ATR42. l'effet de la
achieved thanks to the systematic introduction, since suppression du dbflecteur de volet sur les Czmax a btb
1985. of numerical methods in the design phase, in btudib par calcul pour e t avion Du fait des rdsultats
addition to the empirical and experimental methods prometteurs obtenus, des essais en soufflerie et en vol
used almost exclusively until then. The ATR72 flap ont bds conduits, confiiant la fiabilid des outils
is indeed the fust to be essentially designed with numbriques.
numerical methods at AEROSPATIALE.
Le ddveloppement de nouvelles mbthodes de
Beside this. more severe new regulations for turbo- conception et d'analyse s'est aussi poursuivi; il a
prop A/C. leading to a restriction in the use of large port6 d'une part sur l'extension de leur domaine
flap deflections for the ATR42. the effect of the d'utilisation et sur la qualitb des modblisations, et
elimination of the vane on Clmax has been d'autre part sur la r6duction du cycle de conception.
numerically investigated for this AJC. This study L'objectif btait double : meilleure optimisation des
having shown promising results, some wind-tunnel systhmes hypersustentateurs, et surtout reduction
and flight test verifications were conducted, which notable des coots d'btude associbs.
confirmed the reliability of the numerical tools.

The development of new design and analysis methods

has been pursued, it has involved, on one hand, an
1 - INTRODUCTION
extension of their field of use and the quality of the
modeling, and, on the other, a reduction in the design 1.1 General context
cycle time. The objective was twofold better Generally, the choice and design of the ATR72 flap are
optimization of high-lift systems, and, above all. an largely based on the information obtained from the
appreciable reduction in the associated design costs. ATR42.
Let us recall that the double-slotted high-lift system of the
ATR42 consists of a main flap at the front of which a vane
is attached (see fig. 1). the fixed assembly of these 2 items
RESUME

I Du fait d'objectifs de Czmax d'attemssage un peu

moins forts que pour l'ATR42, le systkme
hypersustentateur de l'ATR72 est constitub dun volet
simple B rotation, alors que l'ATR42 btait equip6 d'un
beiig rotated around a materialized pivot.

However, unlike the ATR42. the design critical phase for

the ATR72 is the take-off phase, the target Clmax on
I landing being slightly reduced.
sysBme B double-fente. avec un volet et un ddflecteur.
La suppression de ce demier a eu l'effet bQlefique de As maximum flap deflection must therefore be changed
simplifier nettement le systhme hypersustentateur
from 45" to approximately 35O. the upstream slot
(between the main body and the vane) is masked
pour l'ATR72.
throughout almost the whole deflection range preventing
I De plus, cette simplification a dd dalis6e tout en
full benefit from being drawn from the double-slotted
system.
assurant des niveaux au moins huivalents de Czmax B
iso-braquage (jusqu'h la valeur requise pour
28-2

This fact, plus the high manufacturing costs for the vane,
led to the idea of a single-slotted flap for the ATR72. the 2D viscous flows (1983/84)
single rotation kinematic system being conserved. A 2D viscous-inviscid strong-coupling method developed
at ONERA [l] in order to calculate incompressible flows
This choice was substantiated by the fact that the ATR72 around multi-component airfoils was industrialized and
flap would be the first one to benefit fkom the systematic validated at Akrospatiale [2].
use of the computational aerodynamic methods whereas
the ATR42 system was designed using methods which JD flows (1985/86)
were mainly empirical and experimental. This should A panel method was developed, f i s t in 2D then in 3D in
allow us to change from a double-slotted flap to a single- order to calculate the flow around high-lift configurations,
slotted flap without loosing the aerodynamic able to take into account very thin trailing edges [3].
performances (for a given flap deflection).
. . . (1985 to 1987)
Note that as the two aircraft have very similar The development of a numerical optimization method for
characteristic wing planfonns, with airfoils of the same high-lift systems was undertaken. This study has allowed a
family, the ATR42 is a very adequate reference for the fust stage in this field to be completed, leading to a code
design of the ATR72 flap. which could be used to optimize the overlap. Results were
also obtained in defining some objective functions for
Also, as the sweep of the wing is moderate, a mainly two- Clmax from Inviscid Flow methods, whose low computing
dimensional approach is justified. ti;;ie'is.particularly well adapted to optimization [4].
1.2 Objectives
These objectives were defined with respect to the ATR42 3 - ATR72 FLAP DESIGN
Clmax 2 Clmax + 2% for 6, from 15' to 3.1 Computational tools and methodology
72 42 The CFD codes used during the theoretical design phase
35' (8fdenoting the flap deflection) were as follows:
( L / D ) 7 2 at 1.2Vs 2 at 1.2 Vs for 6r = - Inviscid Flow method FP2D for multi-element airfoils
15' and 25O (L/Ddenoting the Lift-to-Drag ratio and Vs (this being the panel method mentioned in paragraph
denoting the stall speed) 2.2h
- Mono-element inverse transsonic method,
- Incompressible Viscous method for multi-component
1.3 Design parameters airfoils (this being the viscous-inviscid coupling method
The various parameters taken into account for the design mentioned in paragraph 2.2),
were the following: - Optimization method, under development (usable only
- relative flap chord
for overlap optimization).
- length of fixed trailing edge panel
- shape of flap At that time, none of these methods gave a reliable
- position of pivot indication of the drag. However. the coupling method,
gave a Clmax estimation. We therefore defined by the
The latter defines the position of the flap (gap, overlap) computational channel both the shape of the flap and the
for each deflection angle (see definitions on fig. 2). family of pivots meeting the Clmax targets. This allowed
us to devote the wind tunnel tests to the choice of pivot
offering the best lift-to-drag ratio on take-off.
-
2 STATE OF THE ART OF COMPUTATIONAL
3.2 Preliminary phase
METHODS IN 1985
The first parameter to be defined was the relative chord of
2.1 Difficulties to be solved the flap (which can be kept constant over the span as on
The flows around the high-lift configurations (fig. 3) have the ATR42). A value of 30%. same as that on the ATR42
various characteristics making them especially difficult to (including the vane), was a good trade-off between the
calculate: effect on the Clmax and the smctural constraints.
- presence of areas with separation followed by
reattachment on lower surface of the slats and the f i e d The choice of the fixed trailing edge panel end (84%) was
trailing edge panel also the result of a trade-off, a high value would lead to the
- turbulent separation on upper surface of highly deflected lowering of the pivot and an increase in the size of the
flaps fairings and a low value would lead to a reduction in the
- wake/boundary layer confluence phenomenon extended chord.
- frequent presence of laminar bubbles in the leading edge
areas 3.3 Computational phase
- possible presence of a supersonic area in the vicinity of
the slats leading edges. 3.3.1 Shape of flap
- 1astly;numerical difficulties specific to the panel The shape was defined in landing configuration for a given
methods used for the analysis and caused by the thin deflection (8, = 3S0), a f i e d gap and overlap.
trailing edges.
The principle can be summarized as follows (fig. 4):
2.2 Research undertaken - for a given initial shape and for an imposed lift level
The high-lift related research was conducted along 3 main close to the Clmax level considered, an FP2D calculation
lines, with a strong support from the French National is made in complete configuration mode then a calculation
Agency DRET (groupe VI). on the flap alone conserving its own lift coefficient.
_-
 28-3

- the pressure distribution for the complete configuration 4 - TECHNICAL AND FINANCIAL BENEFITS
is then modified empirically on the flap in order to reduce
the risk of separation. 4.1 Reduction In production costs related to
- the previous modification is transferred to the pressure simplifying the flap
distribution of the isolated flap to make the reverse The main benefit is the reduction in the flap manufacturing
calculation which is only possible for the mono-element costs by the deletion of the vane.
configuration.
- the shape obtained is modified in order to comply with This metallic part (high curvatures incompatible with the
the clean profile in the imposed portions. use of carbon) is subjected to a chemical milling operation
- 2 control calculations are then made on the complete which first implies the installation then the cutting out of
configuration: an Inviscid Flow calculation allows the a rubber mask before the sheet metal is etched in an acid
pressure distribution effectively obtained to be compared bath. These operations are made tricky by the high
with the target distribution; then, a Viscous Flow curvature on the vane which in addition has a shape which
calculation allows us to directly estimate if the modified varies spanwise for the outer elements.
shape leads to a gain in Clmax.
A significant reduction in manufacturing costs is therefore
If one of the two calculations is not satisfactory, the obtained, associated with a 7% weight saving, a rare
operations are repeated with a new modification to the combination worth mentioning.
pressure distribution. If, on the contrary, the new shape
leads to an improvement, it can be used as the initial 4.2 Operational gains
shape for a new iteration of the process and so on until The flight tests c o n f i i e d the Clmax gains (for a given
being close enough to the optimal shape. deflection angle) obtained in the wind tunnel for the
ATR72 the inorease in the Clmax with respect to the
3.3.2 Preliminary kinematic optimizdion "clean" configuration is greater by approximately 10
This consisted in an optimization of the overlap for points than that which was obtained for the ATR42 (these
several flap deflection values (Sf = 32.5", 35". 37.5') foi a beiig aerodynamic values at 1 g not taking eventual stick
fixed value of the gap (ranging from 1.5% to 3%). pusher limitations into account).
Remember that the gap cannot be optimized without
taking into account the confluence of the viscous layers As concerns the lift-todrag ratio at 1.13 Vslg (or 1.2
and therefore, to a greater extent, under Inviscid Flow Vsmin), for S, = 15". the relative gain provided by the
assumptions. single-slotted flap can be estimated at 4 %.

3.3.3 Sum-up of the computational phase This gain in lift-to-drag ratio affects the operational
This phase therefore allowed us: performances on take-off:
- to obtain a complete definition of the shape of the flap - the maximum weight of the 2nd segment (ISA at sea
which would have been very expensive to obtain in the level) changes from 21.3 T to 22 T; note that the low
wind tunnel value, which corresponds to an ATR72 with a double-
- to define a pivot envelope (see fig.5) guaranteeing that sloned flap is lower than the MTOW (21.5T) and would
the Clmax targets will be obtained in order to limit the therefore lead to a degraded nominal mission: - 100 NM or
wind tunnel tests to the choice of the pivot in order to -2 pax.
optimize the lift-todrag ratio on take-off. - the length of the "balanced" runway is reduced at iso-
weight by approximately 60 meters (that is 4 to 5 %) for
the take-off phase.
3.4 2D test results

2-D tests were made in the C.E.A.T S10 wind tunnel, 5 - ELIMINATION OF THE ATR42 VANE
assuming that the differences between the A/C in flight
and the 2-D wind tunnel test are the same as those 5.1 Introduction
measured for the ATR42. This successful design of the ATR72 flap was recently
exploited for the ATR42. A more severe regulation for
These tests therefore allowed us first of all to optimize the turbo-prop A/C [5] having reduced the maximum flap
kinematics. From the various pivots defined during the deflection eventually used for the ATR42 (from 45O to
theoretical phase, most met the landing Clmax target. In around 27"). the good performance of the ATR72 single-
compliance with the planned strategy. it was possible to slotted flap gave the idea to simplify the ATR42 flap
effectively choose the pivot on the 6f = 15O system, by deleting the vane without any change of the
characteristics (Cd at 1.2 Vs and Clmax, see fig.6), taking kinematics, thus strongly reducbg the weight of the
into account its vertical position which determines the high-lift system. This idea was justified by the fact that up
size of the flap fairings. to Sf =27", the vane remains entirely covered by the fixed
trailing edge (see fig.9), with no addition of extended
For the selected pivot, the comparison with the ATR42 chord. Therefore, the effect of the elimination of the vane
flap (fig. 7 and 8) shows: was numerically investigated for this A/C.
- a gain in the Clmax in compliance with targets for 15O<
S, <35O 5.2 Numerical investigation
This study was achieved with the new version of the 2-D
- a reduction of 25 % in the drag at 1.2 Vs for S, = 15". viscous-inviscid strong coupling method developed at
ONERA (see $6.3). performing calculations at the flight
Reynolds number. with and without the vane for a set of
flap deflections, varying from 15 to 45". and for a futed
28-4
e
angle of attack (a=14"). It is believed that such an 6 - CURRENT HIGH-LIFT RESEARCH
approach allows a conect evaluation of the effect on
Clmax, in the sense that the trailiig edge high lift system 6.1 ObJectlves
mainly affects the lift level, without any noticeable As high-lift is not a highest priority sector at
change in sax. The comparison of the calculated pressure AEROSPATIALE, research in this field remains limited and
concerns 3 subjects:
distributions for S, =27" is given on fig.10 and 11. It can - reduction in design costs
be observed that the direct contribution of the vane is - improved flow prediction
very limited, in the sense that the suction on its upper - capacity to process different types of aircraft
surface is compensated by an equivalent one on the lower
surface of the fixed trailing edge (see fig.10). Moreover, 6.2 Reduction in design costs: shape
the lift of the flap is slightly increased by the elimination optimization
of the vane, even though the separation point moves only
slighty forward. Without vane.the optimum flap 6.2.1 Motivations
deflection is 27" (see fig.12). with a rapid growth of the Although constantly concerned with reducing the
separated area for higher deflections. However, it must be computational times. whatever the calculation code, the
mentionned that this optimum is probably under- research entering specifically into this category mainly
estimated, due to a general tendency of the method to concerns the optimization methodology.
over-predict the separations. For high deflections the
configuration with vane becomes naturally more efficient, We have seen (paragraph 3.3.1) that although the shape
the deflection of the flow due to the vane reducing the design methodology based on the inverse calculations
separation on the flap. leads to satisfactory shapes, its implementation is
Beside that, it was checked that without vane, the flap gap especially long and laborious. What is more, from the
keeps acceptable values up to 6f mound 33". formulation point of view, it is based on an improvement
process, and not on shape optimization in the strict sense
5.3 Experimental validation of the term.
These promising results could allow a wind-tunnel test 7
be performed, in order to c o n f i i them, and also to Lastly, this method specific both to Clmax optimization
specify the optimum deflection more accurately. These and to Inviscid Flow approach does not offer good
tests were performed in the S10 wind-tunnel at CEAT development perspectives.
(Toulouse), with a 2-D low-speed model representing a
mid-section of the ATR42 wing, and consisted in balance 6.2.2 Characteristics of the OPTHYP method
measurements of global forces. These tests indicated an All the above reasons have led to the development of a
increase in Clmax due to the elimination of the vane, up to new method based on optimization in line with the
6f =36", the optimum flap deflection being 33" (see research already conducted in this field (see 8 2.2). In
fig.12). thus globally confiiing the predictions from particular, we have kept the idea of a criterion based on the
CFD. The slight underestimation of the optimum flap separation risk assessed under Inviscid Flow conditions.
deflection has already been adressed in the previous
paragraph; the failure of the code to predict the loss of lift The use of a viscous flow code would in fact today lead to
with the vane for moderate deflections is probably due to extremely prohibitive calculation times.
the lack of a modelling for viscous layers confluence, for
such a configuration, with a very small gap between the The new OFTHYP method recently made operational for
vane and the fiied trailing edge. The difference in level is shape optimization (Clmax) of high-lift devices has the
partly due to the lower Reynolds number in wind-tunnel following characteristics:
(2.3M to be compared with 4.7M in flight, at Vs - Analysis methods: FP2D panel code
condition). - Criterion: minimization of the separation risk (for a
f i e d C1 level), assessed by integrating Stratford
Some flight tests were Finally performed in order to coefficients,
validate the previous 2D results on the aircraft. Cl(ol)
- Shape generation (through the MICA2 geometry system,
developed at AEROSPATIALE by the Aerodynamics
curves are shown on fig.13, for several flap deflections, Design Section),
with a fair correlation with the wind-tunnel results. It can - Constraint: compliance with the areas imposed by the
however be noticed that the increase in Clmax predicted in clean airfoil.
wind-tunnel ,estimated 4 , 1 3 after 3D transposition. is
higher than in flight (4.08). It can be explained by the The method was validated on various test cases (ATR72
difference in Reynolds number (2.3M in wind-tunnel and single-slotted flap, upstream flap of a double-slotted
4.7M in flight), in the sense that the small gap already system designed for a laminar airfoil, A310 slat), leading
mentionned between the vane and the fixed trailing edge, to shapes with Clmax's of same order of magnitude as
causes probably less problems in flight, where viscous those of the reference shapes.
layers are thinner.
Apart from the basic geometrical data, the implementation
This operation is,a good example where a combined use of of this method only requires one preliminary Clmax
CFD and wind-tunnel can allow a reduced flight test estimation. This estimation can be obtained, if
campaign, with a very good level of confidence. and thus applicable, by viscous calculations on a preliminary
can lead to a reduction of A/C modification cost. shape but will already be known for all of the more
conventional cases.
 28-5

6.2.3 Benefits The test case selected (A310) has. among other things,
This new approach leads to a spectacular reduction in the allowed experience to be acquired on the calculation of
time required to obtain a shape when compared with the configurations with leading edge slats and on taking a
methodology based on inverse calculations: the time swept wing into account (simple sweep theory).
required is reduced from several weeks (or even several
months) to only several hours! Finally, note that the shape optimization code OPTHYF'
allows all types of flaps (single- and multiple-slotted) to
The benefit thus obtained is especially difficult to be processed and also leading-edge device (conventional
calculate as it depends to a great extent on the case slats, or kruger flaps).
considered but will be without doubt around 100 or 200
engineer hours per high-lift device. The capability of processing the high-lift systems of
various types of aircraft is therefore today largely
This gain in time can, if applicable, be used to take effective and this at moderate study costs.
additional parameters into account (flap chord, length of
fixed trailing edge panel or even spanwise shape
optimization, etc.) leading to a better optimization of the -
7 CONCLUSIONS
The development of the high-lift computational methods
high-lift system.
and their introduction into the design framework has
6.3 Improving flow predictions allowed benefits to be obtained at 3 different levels:
It is within this context that an improved version of the
2D high viscous-inviscid coupling method developed by - improvement in operational performances thanks to a
ONERA [6]was intrcduced and validated at better optimization of the high-lift system,
AEROSPATIALE in 1987 [7]. A comparison with wind-
tunnel results (see fig.14) for a single-slotted flap - improvement in the production process. mainly by
configuration shows a good correlation.'including high simplifying the high-lift systems,
flap deflections, and it can be noticed that stall seems
correctly predicted for this case. This method is still - reduction in the design costs by developing new and
undergoing development at ONERA. more efficient methodologies and by reducing the wind
tunnel tests.
For the three-dimensional aspect, support has been
granted to ONERA in order to implement strong coupling The capability of processing different types of
between the FP3D panel code. developed by configurations has been validated and can be used for new
AEROSPATIALE (see 8 2.2). with a 3D boundary layer projects.
method [8]. This method should allow the flows around the
engine installations to be calculated but could also be used
for the 3-D high-lift configurations. AKNOWLEDGEMENTS
The author wishes to aknowledge helpful dicussions with
As concerns the drag. the current version of the strong- JL, Chavanne, who also kindly provided the Flight Tests
coupling code could be submitted to a specific evaluation, data.
on the basis of available tests. If the results turn out to be It is also a pleasure to mention the importance of the
favorable, an optimization of the high-lift systems as contribution from CBonnet, J.Bousquet and C.Van de
regards drag could be considered. Kreeke in the work presented here.

6.4 Capacity of processing other types of

aircraft
The ATR72 high-lift system was relatively simple on the
one hand by its type (single-slotted flap), and on the other
hand by the low 3-dimensional character (low sweep, high
aspect ratio) of the wing. Nevertheless, various studies
have allowed the processing capacity to be extended to
other types of configurations:

Experience has been acquired both on the design of new

types of high-lift devices (double-slotted flaps , leading
edge Kruger flaps) and on specific problems posed by
laminar wings [9]. An example of pressure distributions
calculated by strong-coupling on a NLF airfoil equipped
with a double-slotted flap is shown on fig.15. Current
studies should allow an investigation to be made, in
particular, on the leading edge stall phenomenon; their
prediction by computational codes and by semi-empirical
criteria will be assessed.

Akospatiale took part in this high-lift research

programme, centered on the Mach and Reynolds effects
and on the 2DDD transpositions.
28-6

REFERENCES
[l] JC Le Balleur, M.NBron: "calculs d'6coulements
visqueux dBco11bs sur profils dailes par une approche de
couplage" (AGARD CP291, Pap.11, 1981)

[2] P.Capbern: "ElBments de validation des techniques de

mesure et des mdthodes de calcul en Abrodynamique des
systbmes hypersustentbs" (L'ABronautique et
1'Astronautique n0122, 1987)

[3] V.Rivoire, P.Eiche1: "MCthode danalyse tri-

dimensionnelle de systbmes hypersustentateurs"
(N.T. ABrospatiale 443528/87, Jui1.87)

[4] C.Bonnet: "MBthode de conception dun systbme

hypersustentateur bidimensionnel"
(N.T. Akrospatiale 443507/87, Mars 87)

[5] G.Cattaneo: "Evolution rbglementaire en matibre de

certification des avions civils en conditions givrantes"
(AGARD CP496, paper 4, DBc. 91)

[6] S.Henry, JC Le Balleur: "Dbveloppement des mBthodes

d'int6raction visqueux-non visqueux pour le calcul des
profils hypersustentes" (RTS ONERA n029/1736 AY,
1986)

[7] P.Capbern: ''Industrialisation et premibre bvaluation

dune nouvelle version de la mCthode ONERA de couplage
fort sur multi-profils" (N.T.ACrospatiale 443527/87.
Jui1.87)

[8] M. Lazareff, JC Le Balleur: "Development of the MZM

numerical method for 3D boundary layers with interaction
on complex configuration" (Vieweg, V29, Oct.90)

[9] P.Capbern: "Theoretical and experimental study of

high-lift device for a NLF airfoil" (First European Forum
on Laminar Flow Technology, Hamburg, March 92)
 a a 28-7

overlap
Double slotted system ,
vane flap

Figure 1 :ATR42 High lift device Figure 2 :Kinematics definitions

wake 1 boundary layer

supersonic thin trailing edges confluence

possible separation
separatiao + reattachment

possible laminar bubble

Figure 3 :Difficutlies to be solved :the flow around a high lift configuration

fixed position
61 = 35"
inviscidflow
i

KP KP

reduction of the
single airfoil multi airfoil
-
risk of separation

inverse single
airfoil calculation

constraint due to
imposed areas

Q
I
I
I
I
I
I
I
I
I
inviscid flow I
I
I
I
I
I
- 1
I
I
I
viscous flow I
estimationof
CI max
- - -J
Figure 4 :The shape of the flap optimized for CI max
28-8

Clmax
32.5"
flap deflection 35"
37.5"

-0
-.

pivot area
Q- (25 points)

Figure 5 :Definitionof a set of pivots Figure 6 :Optimisation of the lift to drag ratio
meeting the landing Clmax in take off configuration (61 = 15O)

Clmax
-- flap ATR42

" '
i.avm Free-
-I I I I
ACd
f

I -25%
--- ATR42
- ATR72
e

Figure 7: 2D W/T test (SlO CEAT)- Figure 8 :Wake drag in take off configuration
comparison ATR42 / ATR72 (6f = 15')

a I
I
XI I

ZI

Figure 9 : Configurations with /without vane (61 = 27 ")

 28-9

vane upper side

wing lower side

- KP
10

-
Figure 10 :Strong coupling calculation Configurationwith vane
(61 = 27O,a = 14O ,Re = 4.7 lo6)

-10
Kp 4

-
Figure 11 :Strong coupling calculation Configurationwithout vane
(61 = 27O,a = 14O,Re = 4.7 lo6)
28-10

-KP 4: -
"Take off" : sfl 7 O , sf2 15"
a = 12"
-
CI max

/*

4 B\
viscous inviscid

0
VI
-
calculation
(Re 4.7 M)
/-

experiment '
(Re E 2.3 M) / / ----.
/
- with
without
vane
vane

I
/'

I I I I
--
- - -f
-- -?a-
0 10 20 30 40 61
Figure 12 :Effect of the vane on Clmax 4 I\

ci

0
VI

0 5 10 15 a

-
Figure 13 :ATR42 Fligh test :effect of the vane Figure 15 : High lift device for an NLF airfoil :
on CI (a)curves Computation by ONERA strong
coupling method (Re = 2.25 lo6)
I

experiment
- - - -0- - - viscous inviscid calculation c Landing (61 =32"5)

Take off (61 =I 5")

A"!
I t
I
I
I
I
I
I

b I
0 6 10 I5 20 a 0 5 IO 15 20 a
Figure 14 :Viscous inviscid coupling method (ONERA)
Comparison calculation / experiment for ATR72 flap
 29-1 I
THE AERO-MECHANICAL DESIGN OF A NOVEL FOWLER FLAP MECHANISM

J.R. Mathews
AerodynamicsDepartment
Aircraft Division
Short Brothers plc
Belfast BT3 9DZ. UK

SUMMARY produce a genuinely optimum design the dependence of

theobjectivefunction upon all the design variables must be
The flow around a 2-dimensional wing and flap is re- known. Seldom, however, does the designer have time to
viewed using inviscid and viscous computational fluid determine all the functional relationships. As a conse-
dynamic techniques. In particular, the effect of flap gap is quence the design seldom achieves the global minimum
explored. The results indicate that for optimum aerody- because the objective function is not correctly specifiedor
namic performance at low flap angles, flap gaps in the its dependence upon all the design variables is not fully
region of 2 to 3% are required. A novel 4-barFowler flap appreciated. The design of high lift systems cannot be
mechanism is described which is shown to give these based solely upon maximum aerodynamic performance
required gaps for flap angles greater than 3 degrees. Such but is constrained by weight, cost, ease of manufacture,
a mechanism can be readily optimised for minimum flap maintainability,reliability and the mechanics of the flap
overlapat specifiedtake-off flap settings.A comparisonof movement.
a track and roller arrangement with the 4-bar mechanism
indicates significant advantages for the latter. High lift system design involvesthe application of special-
ised knowledge from a wide variety of fields e.g.

LIST OF SYMBOLS Aerodynamics

Structures
boundary layer parameter Systems
drag coefficient (2-D) Reliability
lift coefficient (2-D) Manufacturing
lift coefficient (3-D) Finance
pitching moment coefficient (2-D)
pressure coefficient For the aerodynamicist,high-lift system design arguably
lift to drag ratio involves one of the most complicated flows to predict
free stream Mach number accurately.Typically it may involvehigh speed modelling
Reynolds number difficulties (transonic flow around a leading edge slat)
normalised transpiration velocity along with complicated viscous flows (separatingbound-
coordinate along the surface ary layers and interacting wakes with boundary layers).
local flow velocity This paper reviews simple 2-dimensional flow considera-
free stream velocity tions (inviscid and viscous) and considers the application
maximum local flow velocity of a novel 4-bar mechanism (Patent pending) for flap
'local velocity on the upper surface deployment.
at 97% wing chord
inviscid velocity
viscous velocity (from inverse INVISCID FLOW AROUND A FLAPPED
boundary layer equation) AEROFOIL
transpiration velocity
panel length The incompressible, inviscid flow around a multi-element
incidence aerofoil may be examined using a simple panel method.
boundary layer displacementthickness The panel method used here follows that originally pro-
local density posed by Newling (Ref. 1). This method uses a piecewise
boundary layer momentum thickness constant source and piecewise linear continuousvorticity
distributions on each panel. The Kutta condition is satis-
fied by setting the vorticity to zero at the trailing edge. In
INTRODUCTION order to close the set of equations it is assumed that
opposite panels on the upper and lower surfaces have the
The design and optimisation of a high lift system to same source density. Similarly, opposite panel nodes are
minimise the direct operatingcost of an aircraft remains a assumed to have the same vorticity strength.
very complicated but essential task for the designer. To
29-2

The stability and convergenceof the method is indicated ally with decreasing gap, as indicated by the difference
on figure 1 where the effect of increasing the number of between the wing peak and dumping velocitiesin figure4.
panels is shown. The lift and drag coefficients are com- For a given incidence these inviscid trends suggest that for
puted using trapezoidal integration of the predicted pres- low flap angles there will be a range of gaps for which the
sure distribution.The contribution due to the trailing edge flow is attached on both sections,with the optimum being
thickness is ignored. (The NACA0012 section has a trail- the smaller gap to maximise the lift. As the flap angle
ing edge thickness of 0.25% of the chord.) Although the increases the range of gaps required to keep the flow
trailingedgecontribution to both the lift and drag integrals attached will reduce until it is no longer possible for the
is ignored, the asymptotic drag value appears correct. flow to be attached on both sections.

The accuracy of the method is indicated in figure 2 where ii) Due to the higher adverse pressuregradient on the wing
the panel method predictions are compared with an ana- one can also forecast that the stalling incidence of a wing
lytic solution by Williams (Ref. 2) for a 2 element aerofoil. and flap combinationwill be less than that of the basic wing
aerofoil.As the flap is brought closer to the wing, the wing
Using this panel method the effect of reducing the flap gap peak velocity increases rapidly resulting in a very severe
upon the inviscid,incompressibleflow around a 2 element adverse pressure gradient just downstream of the leading
aerofoil is shown in figure 4. (The definitions of flap edge. This will increase the likelihood of the wing aerofoil
overlap and gaps are shown in figure 3.) Both the wing and section exhibiting a leading edge stall.
flap sections are based on a NACA0012 section. The flap
(30% chord) is deflected to 30 degrees and set at zero iii) For a positive flap overlap, as the flap gap is reduced,
overlap. (It is assumed that for optimum aerodynamic the rapid increase in the peak velocity on the flap is likely
performance the flap overlap will be near zero.) Using this to lead to the premature separation of the boundary layer
geometry the following inviscid generalisations may be from the flap upper surface. This in tum will lead to the loss
made : of flap lift and may cause a levelling off or a reduction in
the total lift. Likewise, the high velocities and subsequent
i) As the flap gap is reduced, i.e. the flap brought closer to adversepressure gradientexperiencedby the lowerbound-
the main aerofoil,the flaplift reduces and the wing element ary layer on the wing in the cove region may also result in
lift increases at such a rate that the total lift increases. separation.

ii) Reducing the flap gap reduces the flap peak velocity, As a cautionary note, it is important to recognise that these
and increases both the wing peak velocity and the wing inviscid trends can only be expected to apply in the real
‘dumping’, or trailing edge, velocity. viscous flow if the effects of viscosity are confined to
relatively thin boundary layers in the immediatevicinity of
In general, for zero flap overlap, these trends are repeated the surfaces and their wakes. In this case, the outer inviscid
for lower flap angles with decreasing magnitude. flow will dominate and the interaction between the inner
viscous and outer inviscid parts will be comparatively
Figure 5 shows the effect of flap gap for the flap set at 15 weak. However, as the boundary layers approach separa-
degrees and 5% overlap. (For take-off flap angles, the tion the interaction between the outer and inner flows
constraintson the flap mechanism usually result in signifi- becomes increasingly strong till eventually the flow is
cant flap overlaps.) For large gaps these results follow the dominated by viscous effects.
previous trends for zero overlap. However, for gaps less
than 2% the velocities in the region between the flap and The inviscid panel method has been coupled with an
the wing get very large, as indicated by the maximum integral boundary layer and wake code, based on the RAE
velocity on the flap. This results in a rapid increase in the lag-entrainment method. In regions of adverse pressure
flap lift and reduction of wing lift with the net effect being gradient the code uses an inverse form of the boundary
an over all increase in lift. layer equations. Following the work of Le Balleur (Ref. 3)
and Lock and Williams (Ref. 4) the inverse boundary layer
equations are integrated to calculate a velocity from a
VISCOUS FLOW AROUND A FLAPPED specified viscous source strength distribution, or normal-
AEROFOIL ised transpiration velocity (S) :

The inviscid results indicated in figures 4 arid 5 suggest S=3

that in viscous flows the following may be expected to U
occur :
where v, is the transpiration velocity used in the inviscid
i) The reduction of the flap peak velocity will reduce the flow to simulate the displacement effect of the boundary
effectiveadverse pressure gradienton the upper surfaceof layer :
the flap. For the upper surface on the wing aerofoil, the
effective adverse pressure gradient increases very gradu-
 I
29-3
I
I From an initial guess of the S distribution an iterative was predicted ahead of the fixed position. In such cases
procedure follows, cycling between the inviscid and vis- Horton's method (Ref. 6) is used to predict the bubble
cous solutions. The choice of using S as the independent length and growth in the boundary layer momentum thick-
variable in the inverse boundary layer equations seems ness.
natural as it directly links the inviscid and viscous solu-
tions. In the direct part of the viscous calculation S is For gaps less than 1.3% the large velocity gradients, both
updated using equations (1) and (2). In the inverse region favoura6le and adverse, on the wing lower surface in the
the normalised transpiration velocity is updated from the region of the flap give rise to considerable difficulties in
I formula : the integrationof the boundary layer equations. The results
shown in figure 8 are similar to the previous inviscid
results in figure 5 except that the total lift coefficienthas a
maximum value for a gap of approximately 2%.

where B is a function of the boundary layer shape param- Figure 9a shows the predicted variationof L/D with gap for
eters and the local Mach number. Putting f= 1recovers Le SO. Again the results show a maximum for a gap of 2%.
Balleur's formula. It has been found, however, that in (Further analysis of this test geometry indicated that the
many cases the rate of convergencemay be accelerated by maximum lift coefficient is very insensitive to flap gap
increasing f. with the predicted value being approximately 2.9 for.the
range of gaps considered.) Figure 9b shows the predicted
According to equation (3) convergence will be obtained variation of drag with gap at a C,= 2.0, which corresponds
(i.e. A S=O) when W/Lp = constant. Although this is-a to 0.7C1,. The results clearly indicate that minimum drag
necessary condition for convergence it is not sufficient, occurs for a gap of approximately 2.5%. Although the
unless U"/U = 1. Fortunately the scheme does normally geometry is atypical, these results are in agreement with
'converge' with U" = U, although there are occasions wind tunnel measurements for a G A M - 2 aerofoil and
along the wake where 'Convergence' is obtained with the flap (Ref. 7).
constantintherange0.85 to 1.15. Decreasinghdecreases
the rate of convergenceand so the suitability of equation One important viscous effect not modelled by the above
(3) for updating the normalised transpiration velocity S on method is the viscous wake boundary layer interaction.
finely panelled geometries is questionable. When the wake from an upstream element grows into the
developing boundary layer on a downstream aerofoil, the
The Karman-Tsien compressibility correction is used to resulting interaction may lead to separation of the bound-
modify the incompressible Cp's predicted by the panel ary layer. Due to this adverse interaction, the previously
method. discussed inviscid trends with flap gap may be reversed,
effectivelylimitingthe minimumgap that can be used. One
Figure 6 compares the predicted force and pitching mo- may anticipatethat this wakeboundary layer interaction is
ment coefficientswith thoseobtainedfrom experiment for dependent upon geometricparameters such as the flap gap
a simple wing and flap (Ref. 5). The lower surface of the and the extent of the flap chord aft of the wing shroud.
wing aerofoilwas designedfor no separation. (To compute Figure 10 is an example of this effect as predicted using
the flow around a realistic flapped aerofoil it is necessary Irwin's model (Ref. 8) for the wake boundary layer inter-
to replace the physical cove shape on the wing lower action on the NACA0012 wing and flap configuration as
surface with a smooth fairing.) Both the lift and drag used in figures 5.8 and 9. There is clearly a rapid loss in lift
coefficientshave been predicted to an acceptable degreeof as the flap gap is reduced below 2.5%. In addition to this
1 accuracy. (The predicted drag is obtained by applying the loss of lift a rapid increase in drag occurs.

1 Squire and Young drag formula at the end of the wake.)

The maximum lift and stalling incidence are also ad-
equately predicted. DEVELOPMENT OF A 4-BAR FLAP LINKAGE
SYSTEM.
In figure 7 the predicted wing and flap pressure distribu-
I tionsarecomparedwith thosemeasuredin the wind tunnel. Conventional roller and track systems commonly used to
These results along with other investigations have demon- generate a Fowler-type flap deployment possess several
I strated that for similar types of configurations,the method inherent disadvantages e.g.
I predicts the pressuredistributionsand forcecoefficientsto
1 an acceptable engineering accuracy. Large number of parts
High weight
Using this viscous code, the analysis of the NACA0012 High cost
I wing and 30% flap set at 15 degrees and 5% overlap for Vulnerability to contamination
various flap gaps was repeated at a Reynolds number of 6 Wear
million and Mach number 0.15. Transition was fixed on High maintenance cost
both the flap and wing except where laminar separation
29-4

These aerodynamicresults, along with wind-tunnelinves- and 90% shroud, the maximum flap fairing width was
tigations, consistently indicate that for aerodynamic effi- approximately 10% and the depth was 13% of the local
ciency it is desirableto deploy the flap in such a manner so wing chord. These values are comparable to the fairing
as to minimise the flap overlap and provide a reasonably dimensions for a track and roller mechanism.
generous flap gap, typically in the region of 2 to 3%. A
novel 4-bar linkage, as shown in figure 11, has been A constrained optimisation method has been applied to
devised for flap deploymentas an alternative to the track determinethe optimum positionsof thejoints to maximise
and roller mechanism. Two swinging links (A and B) are the aerodynamic performance of the deployed flap. As a
used to join the flap ann,which is rigidly attached to the constraint, the landing position of the flap is pre-set in
flap,to a wing mounted beam.The proportions of the links terms of the flap angle, overlap and gap. (Previous inves-
a~ chosen to ‘programme’the flap translation and rotation tigations have indicated that the optimum overlap for
to give the slot gap and minimum overlap required for an landingis closeto zero,with severalinvestigationsindicat-
aerodynamically efficient arrangement. An example of ing that a slightly negative overlapmay be desirable.) The
slot gap and overlap with flap angle is shown in figure 12. stored, or nested, position of the flap is also known.
The initial chordwiseextension is relatively large for small
angular flap movement.The flap gap opensrapidly during Figure 14 shows a linkage designed by the optimisation
the first few degrees of movement from the stowed posi- process for a 28% flap chord, 90% shroud with the con-
tion. (In effect the flap drops, translates backwards and straints that the landing flap setting was 30 d e m with
then rotates.) zero overlap and a gap of 3.2%. Since it is known that the
optimum overlap for maximum L D for take-off is near
Figure 12also shows the locus of the instantaneouscentre zero the objective function for minimisation was taken as
of rotation with flap angle. For this particular geometric the flap overlap at a typical take-off flap setting of 10
arrangement, the centre of rotation initially moves up degrees. The optimiser indicated that to achieve the de-
wards to infinity, at which point the flap movement is sired minimum overlapat 10degreesthe kinematicsof the
purely translational. The centre then approaches from the optimisedlinkageresultinsmallnegativeflapanglesinthe
bottom left and indicates that as the flap approaches its early stages of the flap deployment. This negative angle,
maximum angle it is rotating about a progressively de- constrained to be less than 4 degrees. is unlikely to cause
creasingradius. The4-bar linkageis effectivelyequivalent handling difficulties for the short time that it exists. The
to that of a hinged flap having a continuously variable achievementof such a low overlap and reasonablegap for
pivot. As the instantaneouscentre of rotation is in general a flap angle of 10 degrees is particularly encouraging and
some distancefrom the flap the effectivetorsional stiffness indicates the aerodynamic advantages of adopting such a
of the system will be low. This therefore necessitates, in simple linkage system for flap deployment.
common with tracked flapped systems, the use of a coor-
dinated multi-point flap drive system. If a higher landingflapangleis stipulatedin theoptimisation
process with theother constraintsunchanged,theresulting
Flap linkage systems have been successfully used in the optimised’overlapfor the take-off flap setting is larger
past, e.g. on the Boeing B747SP and the Douglas DC-8 resulting in a lowerLD. This highlights the importanceof
aircraft. The main advantagesof the current4-bar mecha- balancing the landing flap requirements(essentially max-
nism over previous linkages is its simplicity and superior imisingC-) with the take-off requirements(maximising
aerodynamicefficiency at low flap angles due to the large L/D at an appropriate CJ.
gap and low overlap characteristic.

To preserveconstant % flapoverlapand gap along the span COMPARISON WITH A TRACK AND ROLLER
of a tapered wing, where both the flap leading edge and FLAP MECHANISM.
shroud are wing generators, it is necessary to scale the
linkage at each support in proportion to the local wing Figure 15 compares the flap overlaps and gaps for a track
chord. This produces a pseudo conical flap motion for and roller mechanism with a representative4-barlinkage.
which it is necessary to provide freedom for spanwise The superioroverlapcharacteristicsof the linkage mecha-
movement in someof the links. Figure 13showsschematic nism is immediately apparent along with the much larger
cross sections through the A and B links at the inner and gaps at low flap angles. One disadvantage of the linkage
outer supportsof a flapsegment. At the inboard supportthe relativeto the track and roller is that it would not be suitable
shaft that carries the A link has only rotational freedom as a variable camber device (Ref. 9).
within its bearing thus enabling it to react any spanwise
loads. Theremaining links have self-aligningbearings and The following table summarises the results of a study
are therefore free to accommodate the relative spanwise comparing the overall performance of both flap mecha-
movement. nisms for the same flap chord and wing shroud. The track
data were taken as the datum.
Figure 13 also gives an indication of the maximum flap
fairing width and depth required by the mechanism. A
detailed design study indicated that for a 25% flap chord
 29-5

6. H.P. Homn "A semiempirical theory for the growth

AsDect Track 4-bar Linkage and bursting of laminar separationbubbles."
ARC CP 1073 ,1967
Flap Weight Datum - 19%
7. W.H. Wentz "Wind tunnel test of the G A O - 2 airfoil
Direct Maintenance Cost Datum -40% with 20% aileron, 25% slotted flap, 30% Fowler flap and
10%slot-lip spoiler."
rake off Performance NASA CR-145139 ,1977
ISA, S/L
: BFL Datum -2% 8. H.P.A.H. Irwin "A calculation method for the two
:WAT Limit Datum 4% dimensional turbulent flow over a slotted flap."
ARC CP 1267,1974
ISA+28OC, 5000 f
: BFL Datum -5% 9. E. Greff "Aerodynamic design and integration of a
:WAT Limit Datum +7% variable camber wing for a new generation long/medium
range aircraft"
where BFL is the Balanced Field Length and WAT is the 16th ICAS Congress Jerusalem
Weight Altitude Temperature Limit. The results quoted Paper no. 88-2.2.4 , 1988
for the WAT limit cases refer to the change in the aircraft
payload + fuel weight.

CONCLUSIONS

The ability of a simple 2-dimensional panel method cou-

pled with an inverse formulation of the lag-entrainment
boundary layer equationsfor modellinghigh-liftflows has
been demonstrated with adequate accuracy. Using such a
tool, analysisof the flow around a simplewing and flaptest
geometry indicates that for optimum aerodynamic per-
formance flap gaps typically p a t e r than 2%are required.

A simple4-barflap supportmechanism has been designed

to position the flap with such agap whilst also giving a low
flap overlap. Relative to a flap track system. the 4-bar
mechanism results in significant improvements in the
overall airframe efficiency.

REFERENCES

1. J.C. Newling "An improved two dimensional multi-

element program ."
HSA-MAE-R-FDMMX)7.1977

2. B.R. Williams "Anexact test case for the plane potential

flow about two adjacent lifting aerofoils."
RAE TR 71197 ,1971

3. J.C. Le Balleur "Calculdesecoulementsa forte interac-

tion visqueuse au moyen de methodes de couplage."
AGARD CP291 ,1980

4. R.C. Lock and B.R.Williams "Viscous-Inviscidinterac-

tions in external aerodynamics."
Prog. Aerospace Sc. Vo1.24 ,1987

5. B. van den Berg "Boundary layer measurementson a

two-dimensional wing with flap."
NLR TR 790090,1979
29-6

0.61

Cl

0.6

059

0.008
cd
0.006

0.004

0.0m

I.#
k FIGURE 3 Defurition of flap overlap and gaps, measured as a 9b of
1.7s
the local wing chord.
1.7

1.65

1.6
0 0.m 0.01 0.01s 0.02 0.025 0.03
1 I (no.of p L )

FIGURE 1 Effect of the number of panels on the predicted inviscid,

incompressible mluticm for a NACAOO12. a = 5'.

1 I
--++-I4
'**
**

I
q&*
I
4

I
* *

0 2 4 6 8 10

R.p klid olp (%)

FIGURE 2 Exact and predicted pressure distributions for the FIGURE 4 2D inviscid. inunaprasiblc trcnds for a NACAOOl2
Williams flapped aerofoil (Ref. 2) Configuration A, wing and flap. Flap chord = 30%. deflected 30'.
Flap deflected 30'. overlap = 0, a= 0".
 29-7

FIGURE 5 2D inviscid, incompressibletrends for a NACAOO12 FIGURE 7 Predicted and measured pressure distributions for the
wing and flap. plap chord = 30% d e f l d 15'. NLR flapped aerofoil (Ref. 3).
Ovalap = 5%.a= 0".

0
cd
0

0 1 2 3 4 5 6
0 2 4 6
a(&& Flq-olp(Q)
Re Wlx106
P Mi@ 0.185 R B P ~ x I O ~ Mtnfn0.M

FIGURE 6 Lift.momcnt and drag cocfficimts for the NLR flapped NGURE 8 2D predicted viscous trends for a NACA0012
aaofoil (Ref. 3). wing and flap. Flap chord = 30%. deflected 15'.
overlap = 5%. a = 0'.
29-8

90-

UD <b

85..

80.

75 .I

o m 1-
Q
ora.
,
0.019. b

<' 0
FIGURE 11 The 4-bar linkage mechanism. 1
0.018.
0 1 2 3 4 5 6
F l . p M u l O.p(%)
Rs=&106 Min1=0.15

FIGURE 9 2D predicted lift and drag trends for a NACA0012

wing and flap. Flap chord = 30%. deflected 15'
Overlap = 5%.

<,
Cl
2
1.78.
P

0
1
1.76..

1.74.. ** 0 5 10 15 20 25 30 35
0

w lneb (&e)
1.72.

110
UD
105

lo0

95

90
0 1 2 3 4 5 6
I

= &lo6 M inl=0.15

FIGURE 10 Effect of wakebundary layer inmaclion on the lift FIGURE 12 Flap overlap and gap characteristicsfor the Cbar
and drag trends for a NACA0012 wing and flap. Flap linkage mechanism.
chord = 30%. deflected 15'. Overlap = 5%. a = 5'.
 29-9

10

A d d -

SECTION A-A

I
.. .
FIGURE 15 Comparison of mck wilh linkage.

SECTION B-B

FIGURE 13 Cross-sections through the links.

10 2

0 0
-5 0 5 10 IS 20 Y 30
W .neb

FIGURE 14 Opimised linkage for minimum overlap at a flap angle

of 10'.
 30-1

DESIGN, DEVELOPMENT AND FLIGHT EVALUATION OF

THE BOEING YC-14USB POWERED LIFT AIRCRAFT

TED C. NARK
BOEING DEFENSE & SPACE GROUP
Post office Box 3999
Seattle, Washington 98124-2499
M / S 11-Pc USA

SUMMARY gave the two companies, Boeing and Douglas, a

p a t deal of leeway in configuring the vehicle.
The Boeing YC-14 was designed to perform a There was one other objective which was
4OOnm mission caRying a 26000# payload and instrumental in configuring the lane. That
land in 2OOO feet on a semi-prepared landing strip. "pane was an
was, the cost to produce the airp
The high-lift system was one that had never been important issue.
flown befm, a upper surface blowing (USB)
concept utilizhg the "Coanda" effect The take-off
and landing perfarmance estimates developed from CO"
wind tunnel data were completely substantiates in
the night test gram. The uitical issue of The YC-14 configuration is shown in f i g m 2.
E
r
continuingei alandingortakeoffaftertheloss
of one ofthc two CF6 engines was also proven in
The fuselage was *to handle almost all ofthe
qui ment in the h y inventory. It was
the flight test program. The design details behind lJ
CO gmdto allow aerialdelivery ofboth tfoops
and equipment out of the aft cargo door and troops
the performance of the YC-14 are discussed and
some of the perfannancefeatmes of the airplane fromasidedoar'ustaftofthemaingear. The
are explained. d
most significant eature of the YC-14 was
probably the two -6 engines placed on top of the
wing.
I" The twin engine design was due to the ~ ~ s u lof
ts
The Boeing YC-14, and its competitor the Douglas an analysis that indicated such adesign was
YC-15, began life as elements of the Air Farce possible using an existing modem engine (a6)
AMST (AdvancedMedium STOL) program which and would be both cheaper and simpler to build
wasinitiatcdwithreleaseofanRFPin1971.The than afourengine configuration. In addition
YC-14 flew for the first time on August 9,1976. accident scatistics COIlf'irmedthat twia engine
airplanes have fewer engine shutdowns and were
Thecom titionforthe
P P uctioncontractwas
won by e Douglas C- 5. The mission was
substantially changed, particularly with respect to
safer to fly. In addition to the Air Force
requiFements, Boehg established an in-house
its STOLproperties, and the airplane is being built design philosophy, thee elements of which are
today by McDonnell-Douglas as the C-17. worth nothing here. F irst the ahplane must be
easy to fly on hnalapproach. It should fly like a
The basic design goals for the AMST mission conventional jet airplane to avoid adding an exm
were basically very simple and aze listed in Figm burden to the pilot training program and at the
1. Fundamentally it was to be a flying truck same time be safe to fly. Second the airplane must
capable of handling almost any of the fighting gear have engine out performance equal to wnvcntional
the Army had in its inventory and it had to deliver jet aim&, that is, be able to continue a STOL
a 27,000 pound payload to a battle field 400 take-offorlaa~gwithanaveragerniliearyczew
nautical miles away. It had to land on a 2000 ft. with an engine out. And thirdly, as mentioned
unprepared runway, off-loadthe payload and then earlier, it must be a configuration with a low cost
take-offwith another load andreturn to base. The of ownership.
specifications were very broadly spelled out and it
30-2

Low cost of ownership included both low unit fly- conventional jet such as the 727 (Figure 3.) With
away cost as well as low maintenance costs. A all engines operating (neglecting the effects of a
great deal of thought went into selecting the failed engine) the YC-14 can approach at 65 knots
materials and processes used in producing the YC- and land in about 1150 ft.
14. Parts and pieces were configured to lower the The design of the USB high lift system
manufacturing costs. For example the wing incorporating the USB flap was instrumental in
structure had a main spar that was straight and achieving a W/qS of almost 4, with an engine out.
continuous from tip-to-tip. The wing center
section was constant chord to allow this use of The high lift system was made up of a number of
constant airfoil sections. The same thing was done complimentary elements (Figure 4). It had full
for the vemcal tail, which was designed with a span (8 segments) two-position variable camber
constraint chord fin. The body structure was Kruger leading edge flaps, full span leading edge
composed of simple structural elements. blowing, using 8th and 14th stage engine bleed air,
and double slotted center and outbound flaps. The
At the time of the AMST competition there wefe two position double slotted flaps behind the engine
two serious powered lift concepts considered for were sealed with the engines operating and open
the design. The engine blown flap (EBF) and the behind a dead engine. Vortex generators, or
upper surface blown (USB) flap. As you can turning vanes on the wing behind the engine, were
imagine, there were many design issues to be used to assist in spreading the flow outboard. An
considered in selecting an approach, and as you USB/nozzle door on the outboard size of the
might also expect there was no clear cut winner. nozzle exit also assisted in spreading the flow
Douglas, after what must have been serious outboard.
considerations, selected the EBF concept with four
JT8D engines for the YC-15. Boeing on the other The entire high lift system was designed to
hand placed its bet on the USB concept, with two continue a STOL takeoff or landing with a critical
CF6 engines, for the YC-14. The USB engine imperative. The leading edge Kruger flap
configuration allowed the engines to be placed was designed with a tapered plan form and a
close together over the wing minimizing both the spanwise change in camber from root to tip. It
rolling moment and yawing moment associated represented a very unique solution of a difficult
with an engine out. The coanda effect, the design design problem. The leading edge Boundary
feature behind the USB flap, was shown by both Layer Control (BLC) System (Figure 5 ) was
NASA Langley tests and tests at Boeing, to designed to supply engine bleed air to the entire
effectively turn the flow through as much as 700 wing from whichever engine was functioning and
with less than 10% thrust loss. Another design to accomplish this with a pnuematic design, no
feature following from an over-the-wing design electrical or mechanical switching.
was the lessening of the concern over foreign
object damage when operating on the un-improved The aerodynamics of the leading edge blowing
runways required by the design requirements. The system was a unique design challenge. The wind
thrust reverser was designed to deflect all the tunnel testing of BLC was accomplished with the
thrust-- not just the primary flow-- forward and blowing orifice as a slot, producing a continuous
upward. This allowed the reversers to operate jet across the span. In addition the mass flow
effectively down to zero forward speed without across the wing was distributed in proportion to
any concern of re-ingesting the hot gases in the the local wing chord to approximate an uniform
inlet, and at the same time improved the braking blowing coefficient along the wing span. The slot
force associated with the landing gear. blowing design was considered impractical for the
full scale airplane. A design using small holes was
From the laws of physics, a 2000 foot landing developed. The design of the leading edge
distance equates to an approach speed of blowing nozzle and plenum system is shown in
somewhere between 80 and 90 knots. For Figure 6. With a series of tests of both the slot
conventional aircraft (non-powered liit)W/qS is nozzle at model scale and drilled holes nozzles at
defined as CL, but for a powered or propulsive lift full scale it was possible to correlate the
airplane a more suitable name is landing lift aerodynamicefficiency at model scale with the full
efficiency. Because of the thrust dependent lift scale performance of the airplane.
vector, W/qS can be quite a bit larger than the
aerodynamic CL. In the case of the YC-14 with an The leading edge BLC system was itself a coanda
engine out, W/qS is just about twice as large as a surface device. The small drilled holes created
 30-3

smail jets of air, which after impinging on the Philadelphia, PA to obtain the influence of the
coanda surface spread out and merged into an propulsive lift on the control system and develop
effective sheet of air, simulating the slot nozzle, as the engine out system for the take-off and landing
it was turned by the coanda surface. The spanwise configuration.
mass flow was varied by changing the size of the The test program was an invaluable asset in
drilled holes and the hole spacing in each plenum. establishing the design and performance of the
The hole spacing was selected so that if a hole critical engine inoperative configuration. A partial
were to plug up the two adjacent holes would list of the many design issues successfully resolved
spread together to create a thinner, but still a in the wind tunnel are listed in Figure 7. The last
continuous jet. The system in flight worked three items in the list were brought about by
flawlessly, if either engine failed the system problems uncovered in flight test and then wind
continued to function without interruption. tunnel tested to provide solutions that resulted in
substantial reduction in the high speed cruise drag.
The unpowered outboard flaps (Fig. 4) were of
conventional design and were supported by an
external hinge point to avoid the mechanical
complexities and maintenance issues of alternative
approaches. The USB flap had to be sealed to The wind tunnel was the source of the W/qs/max
function in the propulsive lift mode. With an data used in predicting the STOL perfoxmance of
engine out both the lift and drag penalties of such a the YC-14. Figure 8 compares the wind tunnel
configuration were too severe to be acceptable. A W/qs/mw data for flaps 60 USB 500 (nominal
mechanism, triggered by a single actuator, was landing flap configuration) with the corresponding
designed to convert the sealed USB flap with an flight test data. It is plotted against the jet thrust
engine operating to a double slotted flap behind the coefficient of the live engine. It is appropriate to
dead engine. This converted the high lift system point out at this time that the field length
on the dead engines wing to a double slotted flap, performance requirements established by the Air
with leading edge blowing, from the side of the Force were all to be met with a critical engine
nacelle to the aileron. To accomplish this inoperative. The performance with all engine
configuration change an engine out sensor was operations was especially outstanding. It could
designed to identify which engine had failed. land at 65 knots in about 1150 ft and take-off in a
ground roll of 1,100 feet.
With the decision to implement such a sensor came
a whole new approach to handling the continued The requirement to produce landing performance
approach or continued take-off with an engine out. with an engine out resulted in an intenxting
Model scale tests demonstrated that the turning of situation for a twin engine, powered lift airplane.
the gas jet by the coanda effect was enhanced as With an engine out on one wing, that wing has
the jet flow was less concentrated or more evenly basically no powered lift except leading edge
distributed over the coanda surface. In practice, blowing. The approach speed was essentially
thinning of the flow was enhanced on the YC-14 established by the lift obtainable on the wing with
with two design f e a m s . The first was a nozzle the dead engme. The key to the YC-14's success
door on the outboard side of each nacelle, and the was embodied in two unique design f e a m s .
second was the addition of turning vanes on the First the live engine could support the leading edge
wing just behind the engine exhaust plane. Both boundary layer control system on both sides of the
features were activated when the flap handle was airplane. Second with an engine out, the wing
placed in a landing flap detent. With an engine with a dead engine had a well designed double
failure the nozzle door on the dead engine was slotted flap from the side of body clear out to the
closed and the turning vanes stowed to reduce the aileron, a very effective high lift system. This
drag on the &ad engine wing. presented the configuration with one m m crucial
design challenge, them was just too much lift on
The high lift system was designed and developed as the live engine wing at the power required to
a result of a very comprehensive scale model test maintain the glide slope on one engine. In effect
program. A lot of the early developmental testing the powered lift had to be "killed" as the engine
of the high lift system elements and exhaust nozzle power came up or the airplane would roll over on
features were accomplished with a half model in a its side. Suffice to say this didn't happen. It is
9x9' non-return tunnel located in Seattle. The interesting to note that it was possible to build a
complete model (at the same scale, as the half twin engine high lift system with a trimmed
model) was tested in the 20x20 Vertol tunnel in
30-4

W/qs/max 4(Figure 8) with one engine moment associated with a single engine operation
operating. Also notice how well the W/qs/max in an approach configuration. Figure 10 shows
measured in the wind tunnel agrees with the flight that the increment in lift caused by the loss of an
test data. Figure 9 displays the same data only engine, which is responsible for the engine out
now is terms of minimum speed, Vmin. For a rolling moment, has an apparent moment arm
STOL airplane the minimum speed can be based equal to the distance of the nacelle from the center
on a control limit or an aerodynamic lift limit (wing line of the airplane. This implies that the engine
stall). On the YC-14 it was a wing stall that out rolling moment can be minimized by placing
determined the minimum speed. The minimum the nacelle as close to the fuselage as possible, a
speeds determined from flight test (data at the definite advantage for an over the wing engine
thrust required for approach on a 600 glide slope installation on a high wing configuration. On a
(TIW4.21},) is at least as good at the wind tunnel take-off, after the loss of an engine, control of the
data would predict for the same condition. In moments about the directional axis is also critical.
general the high lift performance of the YC-14 Again moving the engine as close to the fuselage
during flight test was always as good as predicted as possible had a significant impact on reducing
by the wind tunnel data base. the yawing moment coefficient associated with an
inoperative engine. Figure 11 illustrates that the 4
The take-off of a STOL airplane in 2000 ft. is engine YC-15 with JT8D-17 engines had a
dependent on two features, thrust, lots of it, and somewhat smaller engine out yawing moment
W/qs/max for the take-off flap configuration. coefficient with a critical outboard engine
Parametric studies conducted during the early inoperative to contend with than did the YC-14
design phase in the development of the YC- 14 with an engine inoperative. Also note the
indicated that a 0.2% T/W 5 0.30 was required to significant difference between both the YC-14 and
accelerate the airplane to a misonable lift-off YC-15 compared to a conventional twin engine
speed. Built into this take-off process was a 737. The control of the YC-14 engine out rolling
moment and yawing moment were a critical issue
W/qs/max=3.5. To put these numbers in in the developing the YC-14.
perspective the engine-out T/W for the YC-14 is
greater than the all engine T/W for a conventional Not only were the YC-14 rolling and yawing
jet transport and the YC-14 W/qs/MAX with an moments high, but the speeds at which they had to
engine out in the take off configuration is be controlled were much lower than conventional
substantially greater than the W/qs/max of a jet aircraft, The net result of having to produce a
conventional jet in the landing configuration. large moment coefficient at low q (low speed)
I' I'

is the requirement for large or multiple control

The key to the success of the YC-14 twinengine surfaces (or both) that move very rapidly and
STOL configuration was heavily dependent on the through large angles to p v i d e aircraft agility at
availability of the GE CF-6 engine. It represented STOL speeds. It was decided that double-hinged
almost a perfect match to the STOL performance in or anti-balance surfaces offer an attractive
both take-off and landing as well as the cruise compromise to overly large simply hinged surface.
performance in terms of thrust and fuel burned.
The flap configuration for take-off is rather A comparison of the differences in the vertical tail
unique. The USB flap behind the engine was of a 737 and YC-14 are shown in Figure 12. A
retracted to produce the most horizontal thrust for similar set of information comparing the YC-14
acceleration, and the outboard double slotted flaps horizontal stabilizer with that of the 727-100 is
were at 300 to produce an acceptable W/qs/max of shown in Figure 13. It is quite evident from a
3.5 with an engine out. review of the material presented in Figure 12 and
13 that a STOL control system, as represented by
An important part of the YC-14 mission the YC-14, is very diffemt from that for a
requirement was to continue either a STOL landing conventional jet transport represented by the 737.
or take-off after the loss of an engine. Placing the
engines over the wing to utilize the USB powered In d e r to achieve a 2000 foot landing distance by
lift concept and placing the wing on top of the military rules requires approach speeds on the
fuselage to minimize the influence of the ground order of 90 knots (Figure 14). But that alone is
on the lifting system had the benefit of allowing insufficient. A very effective retarding force is
the engines to placed very close to the fuselage also required, and the airplane once on the ground
thereby minimizing the thrust induced rolling must stay there and not bounce back into the air.
In addition the pilot must be able to control the
 30-5

airplane on the glide slope to a precise and positive redundant, electrical flight control system WCS).
touchdown. This embodies designing a Once a computer was added to the flight control
configuration with effective thrust reversers on the system the opportunities to exploit it became
engines, a robust, rugged set of brakes, a landing unlimited. The back side polar problem was
gear which is tuned to a high rate of descent at solved by incorporating a speed hold module into
touchdown (=lOFPS), precise glide path control, the EFCS. The pilot had a very simple task to
even in adverse weather conditions, and a lift master on approach. Use the stick to point the
dump system that "kills" the lift and puts the nose of the airplane where he wanted to land.
weight of the aircraft on the gear at touchdown. Everything else was taken care of for him by the
The YC-14 design addressed each of these issues EFCS. Servos where coupled to the throttle and
with a very positive approach. the USB flap. On approach a "speed hold signal
was developed and coupled to a conventional
The thrust reverser not only had an effective attitude hold mode. Once the target speed was set
with a dial indicator, the EFCS would modulate
reverse thrust component (a%), it also the throttle and USB flap to maintain speed. The
contributed a sizeable downward thrust component attitude hold mode would keep the wings level and
to increase the normal force on the gear and the pilot would control the pitch plane with the
thereby increase the braking force. The braking control wheel, his only task In the event of an
system had a coefficient of braking friction of engine failure during approach the pilot continued
, coupled with an anti-skid mechanism
~ 4 . 4 2and to control the pitch plane. The engine failure
was a very effective at bringing the airplane to a detection system coupled into the EFCS triggered a
stop on the ground. The landing gear was series of events which took pre-programmed
designed (Figure 15) to routinely land at a rate of measures to trim out the airplane about all three
sink at touchdown of 6 to 10 FPS. This coupled axis leaving the pilot to continue to control pitch
with a approach speed of approximately 86 knots plane dynamics to maintain flying the 60 glide
allows the YC-14 to fly down a 60 glide slope slope. His only task was to continue pointing the
(commercial transports routinely use 2-1/20-30) nose of the plane at the touchdown point.
and make a no flareor partial flare landing at
STOL weights on a very soft field without The maximum deviation from the 60 glide slope on
exceeding the design rate-of sink at touchdown. an engine-out go-around was less than 15 feet and
In actual practice, the airplane hits the ground, the less than 5 feet on a continued approach. An
gear compresses smoothly and the airplane is on engine out landing could be successfully
the ground to stay. A switch on the landing gear accomplished from an engine failure at any
then triggers a signal to the spoilers on both sides altitude. In addition, the thrust reversers alone
of the wing, from just outboard of the nacelle to would support landing on icy runways. A similar
just inbound of the aileron, to come up, directly approach was developed to handle an engine out
"dumping" the aerodynamic lift on to the gear and on take-off. A summary of some of its more
maximizing the braking effectiveness immediately significant EFCS modes is shown in Figure 17.
on touchdown. Having a computer in the control loop proved to be
an invaluable tool and in my opinion it proved
To make a precision landing the airplane must be itself on more than one occasion.
in control of the glide path in all weather
conditions. This is especially difficult for an As you can see developing a short field airplane
airplane with propulsive lift, which almost always around a high lift system is more involved than
fly on the "back side" of the polar (Figure 16). just designing the high lift system. Some of the
The YC-14 design was developed to give a elements involved in developing a successful short
military pilot the feeling on the front side the polar, field airplane m detailed in Figure 18. The
-- the way he was fundamentally uainedto fly. It landing performance of the YC-14, which had
designed into it the features described in Figure
was felt that in an emergency situation the pilot
could not be expected to remember a unique set of 18, is shown in Figure 19,. It represents the
flying qualities associated with a STOL airplane. statistics on the STOL landing dispersion. It
The decision was made early in the program that includes landings with all engines operating and a
there should be no reason to re& the pilots. The critical engine inoperative. It represents the data
STOL airplane would be made to fly like a from 108 landings. The piloting task was to hit an
conventional airplane. This was accomplished by aiming point on the runway. The measurement is
developing a computer controlled, triply the distance, in feet, between the pilot's aiming
30-6

point and where the plane actually touched down. flow channel to configure for both acceptable high
The average of all the errors was under 16 feet on lift and high speed properties A moveable leading
the long side, and 63%- of all the landing were edge device was first designed, but it was a
mechanical nightmare. After extensive wind
made within f 74 feet of the aim point. tunnel testing a compromise fixed leading edge
was developed and worked successfully. There
In summary the YC-14 met or exceeded all its were two areas of concern with the YC-14 high
STOL design requirements. The low speed speed drag uncovered in flight test. The up-swept
properties of the airplane were adequately aft body and the large gear pods. Both had drag
predicted with data from low speed wind tunnel levels that were:judged to be unacceptably high.
tests. The concept of operating a safe, twin- The gear pods were necessary because of the
engine, STOL airplane, was substantiated. requirements on the dimensions of the cargo
compartment and the high wing configuration, left
no volume in which to store the gear, hence an
external fairing or gear pod was created The up-
swept aft-body was a by-product of the
There was mofe to the success of the YC-14 requirement of loading and unloading rolling stock
airplane than its STOL capabilities. It could also through the aft cargo door.
fly at a mach number of 0.68, almost in keeping The flight test drag level of the original
with the commercial jet fleet of airplanes. This configuration was 12%higher than expected. As
presented the design with another challenge. The later found out in the wind tunnel, the flow around
first and most important was to "camouflage" the the gear rod was interacting with the cross flow
high lift system for high speed cruise. over the aft fuselage and the flow over the upswept
The upper surface contour of the USB flap was aft body was badly separated. The solution,
basically circular. The most cost effective design identified and solved in the wind tunnel, was
was determined to be achieved with an external composed of two parts. First the gear pod was
hinge point. This design also was consistent with redesigned to close the lines in two dimensions,
using the USB flap as a control element on like a wing shape, rather than in three dimensions
approach. The six hinge points on each wing were like a body of revolution. Second, strakes were
provided with fairings which are very evident in a added to the aft fuselage to assist in keeping the aft
front or back view of the YC-14. The engine body cross flow attached. Before the flight test
nozzle was specifically contoured to achieve a program ended the strakes and gear pod
widely spread jet, a high jet turning configuration, improvement were tested in flight and improved
but, not an acceptable high speed configuration. the cruise drag by 5%.
In cruise there was a scrubbing drag, it was
considered a part of the nozzle thrust coefficient of The control system as mentioned before had to be
0.97 and accounted for in all the pre-flight test very responsive and have a large authority to
performance estimates. A lot of effort was possess satisfactory handling qualities at STOL
expended in designing the nozzle that would both speeds. The YC-14 operated over a dynamic
enhance the exhaust jet turning at low speeds and pressure range of almost twice that of a
have minimum impact on cruise performance. The conventionaljet transport (Figure 21). This leads
empirically defined parameters are shown in to a situation where the control system tuned for
Figure 20. The problem was that nozzles with STOL flight speeds was too powerful for cruise
good turning almost always had unacceptable high flight. The elevator for cruise fight was only 1/3
cruise drag and good cruise nozzles of the available span. The high speed rudder is
characteristically did not turn the flow well at low only twethirds of the available span and pressure
speed. The net result is that a nozzle designed for limited to reduce the throw.
cruise can have a 10%improvement in cruise drag
and w t = 0.05 over a high turning nozzle
design, while a high turning nozzle can have a CONCLUSIONS
maximum lift improvement of AQ=1.3 over a
good cruise nozzle. The variable geometry nozzle The YC- 14 performed well, meeting or exceeding
went along way toward a compromise design. all the design goals. The pilots who flew it where
Another critical area with respect to cruise drag all impressed by its flying qualities. The airplane
was the flow between the nacelle, wing leading in general was an operational success, but the YC-
edge and fbselage. It was a particular difficult 15 won the prize.
 30-7

BOEING YC-14 DESIGN REQUIREMENT AT LOW COST

27,OOOLB (12,247KG) PAYLOAD

400 NAUTICAL MILE (741KM) RADIUS

2,OOOF" (609.6M) FIELD LENGTH CAPABILITY

11.7Fl' (357M) WIDE X 113FT (3.44M) HIGH

X47Fl' (14.33M) LONG BOX (REQUIRED)
53,000 LB (24,000 KG) OVERLOAD PAYLOAD (2.256)
2600 NAUTICAL MILE FERRY RANGE

TURBO-JET TRANSPORT CRUISE SPEEDS

FIGURE 1

USAF/Boeing Advanced Medium STOL

Transport Prototype - YC-14

WING AREA: ,i,762 sa FT ( i a . 7 sa MI

CARGO COMPARTMENT SIZE:
47 FT LENGTH (14.30M)
11.7 FT WIDTH (3.5MI
11.2 FT HEIGHT (MINI (3.4M)
ENGINES: CF MOD
-
48,300 LE SLS THRUST(INSTALLED)
4.3 BYPASS RATIO
1.65 FAN PRESSURE RATIO

129 FT I

(39.32M) i
A

I I
18 FT7 IN
15.66M)

FIGURE 2
30-8

Landing Lift Efficiency

e MAXIMUM PAYLOADS
e YC-14, STOL LANDING

4
YC-14

-
WIS
q 2 727
AT
LANDING

FIGURE 3

YC-14 HIGH LIFT SYSTEM

(OUTBOARD OF NACELLE/USB FLAP)
FLEXIBLE
PANEL U' U'

/ L.E. BLC

FLAP LINKAGE

VARIABLE CAMBER DOUBLE-SLOTTED

LEADING-EDGE FLAP EXTERNALLY HINGED
TRAILING-EDGE FLAP

FIGURE 4
 0 30-9

YC-14 BLC System

14TH-STAGE SHUT0 FF CHECK VALVE

AND CHECK VALVE

BLC SHUTOFF VALVE TEMPER ATU R E

BLC NOZZLE PLENUM

MANIFOLD
E-ICE VENTURI
PR ESSUR IZED
COMPARTMENT

FIGURE 5

YC-14 L.E. BLC Nozzle Pleiium

r BLC NOZZLE BLC MANtFOLD

.LED

/ DETAIL

FIGURE 6
30-10
e
ISSUES RESOLVED IN WIND TUNNEL

INCREMENTAL LIFT DUE TO JET EXHAUST

INFLUENCE OF GROUND PROXIMITY ON ALL FORCE AND
MOMENT COMPONENTS
SOLUTION TO A SIDE FORCE PROBLEM ASSOCIATED WITH AN
ENGINE OUT

AIRCRAFT CONFIGURATION FOR OPTIMUM CLIMB GRADIENT

WITH A CRITICAL ENGINE INOPERATIVE

APPROACH SPEED AND APPROACH CONFIGURATION FOR LANDING

WITH A CRITICAL ENGINE INOPERATIVE

REDESIGN OF GEAR POD IN THE PRESENCE AN UPSWEPT AFT BODY

REDESIGN OF NOZZLE DOOR ACTUATOR FAIRING

DESIGN OF AFT BODY STRAKES

FIGURE 7

YC-14
SINGLE ENGINEOUT MAXIMUM LIFT
FLIGHT TEST DATA

-2 .4 .6 .8 1.0 1.2 1.4

c j LIVEENG.

FIGURE 8
 I
30-11

YC-14
SINGLE ENGINEOUT MINIMUM SPEEDS
FLIGHT TEST DATA

USB = 50'
\
90-
\
'
sOn \
V MIN
80-
U
KEAS
70-
FlightTest
Data
f er
S.L.
60. STD.
I I
.os .10 .ls .20 .25

W
TG
- LiveEngine
FIGURE 9

Engine-Out Rolling Moment

0.24 ,-

APPROACH FLAPS
01" 0.20 -
I-
w
2 BALANCE DATA
4
LL.
0.16 -
U.

E
0
E 0.08 -
(3
-
z
-I

0 0.4 0.8 1.2 1.6 2.0

A CL - LIFT LOSS DUE TO ENGINE-OUT
FIGURE 10
30- 12

Engine Out Control

0.20-
z NOTE: TAKE-OFF THRUST
0
Y

0.15
U
W
0
0
0.10'

0.05

-
Y
bl2
FIGURE 11

YC-14
VERTICAL STABILIZER FEATURES

EFCS
FLAPS DOWN

EFCS
FLAPS UP & DOWN
r&
TYPICAL SECTION

% % RUDDER
CHORD
-
V
A @p=O Rudder Throw
RUDDER (1T O b )
YC-14
25 43f2 1.5 0.13 1.o 25O/5Oo
737-100 220 25 0.09 0.67 24'

FIGURE 12
 30- 13

YC-14
HORIZONTAL STABILIZER FEATURES

/ 30'

HIGH SPEED CONTROL

-
-hC
4
AR
-
%ELEVATOR
CHORD
-
"H
CL
TYPICAL SECTION

ELEVATOR THROW
TRAILING TRAILING
EDGE DOWN EDGE UP
YC-14 3.5" 5.00 38/19 1.6 1.60 20"/40" 30°/60"

I 727-100 35.0" 3.25 25 0.9 1.15 17" 26"

FIGURE 13

LANDING COMPARISONS

BEST OF CURRENT
JET "RANSPORTS
6" GLIDE SLOPE
AUTOMATIC GROUND SPOILERS
R/S @ TOUCHDOWN = 10 FT/SEC
EFFECllW THRUST RSVSRJERS

I 1

TOUCHDOWN SPEED KEAS -

FIGURE 14
30-14

Long Stroke Landing Gear

FIGURE 15
 30-15

YC-14
GLIDE PATH CONTROL
+20 - AFTROACH

ALL ENGINES
OPERATING
- KNOTS
AV 0 -
+10

-10 -
\./'" INCREASED POWER

FIGURE 16

ADVANCED ELECTRICAL FLIGHT CONTROL SYSTEM

SPEED HOLD MODE FOR APPROACH AND LANDING

LOAD LIMITER FOR USB FLAP

PROGRAM THE CONTROL SYSTEM WFTH SPEED

FULL-TIMEENGINE FAILURE DETECTION SYSTEM

AFTER AN ENGINE FAILURE

ADVANCES THROTTLE ON LIVE ENGINE

MATCHES USB FLAP TO OUTBOARD FLAP ON DEAD ENGINE SIDE

RE-TRIMS LIFT SYSTEM ON LIVE ENGINE SIDE

RETRIMS PITCH PLANE

FIGURE 17
30-16
e e
DESIGN FEATURES OF A SUCCESSFUL
STOL CONFIGURATION

-
SIMPLE TO FLY NO SPECIAL PILOT TRAINING

-
EFFECTIVE HIGH LIFT SYSTEM VAPP 90 KNOTS
WELL DESIGNED GEAR ALLOW HIGH IUS AT TOUCHDOWN
5,

EFFECTIVE THRUST REVERSER

EFFECTIVE GROUND SPOILERS
ELECTRICAL FLIGHT CONTROL SYSTEM

FIGURE 18

YC-14
STOL LANDING DISPERSION

2(

16
NUMBER OF
OCCURRENCES
l4 12
8 STANDARD

I: NORMAL
DISTRIBUTION

0
-1b -1h -io I
-40 0 ' -4

DISPERSION - FEET
FIGURE 19
 0 30-17

DEFINITION OF MAJOR
NOZZLE PARAM ET ERS
0 - KICKDOWNANGLE (NAC Q)
7 = BOATTAIL ANGLE (NAC $1
= CUTBACK ANGLE (INBD & OUTED)

-
$ = FLARE ANGLE (iNBD a o u T e o ~
W WIDTH
H = HEIGHT
AR, = EFFECTIVE FLOW AREA
cl2

P
FIGURE 20

Operating Dynamic Pressure Ratios

0 MAXIMUM PAYLOADS
0 YC-14, STOL LANDING
30 0 q- DYNAMIC PRESSURE

YC-14
20

MAX
LANDING

10

FIGURE 2 1
 31-1 I

HIGH-LIFT DESIGN FOR LARGE CIVIL AIRCRAFI'

A. Flaig, R. Hilbig
AERODYNAMICS DEPARTMENT
DEUTSCHE AIRBUS GMBH
"EFELDSTR. 13,2800 BREMEN, GERMANY

SUMMARY The primary consideration in the design of high-lift systems is

A general reflection of the high-lift system design process is the achievement of the required airfield performance, which
given in the first part of the presentation. First the objectives generally implies high maximum lift capabilities for landing
and constraints are reflected which drive the high-lift design and even high lift to drag ratios for takeoff combined with the
for civil transport aircraft. Further infomation is given on the constraint ofminimum deterioration ofthe cruise performance.
applied theoretical methods and the Deutsche Airbus wind- In general the required airfield performance for civil and
tunnel strategy. military transport aircraft is different due to their specific
An example of the high-lift system design process is given in operating conditions, Fig. 2. In comparison to civil operation,
the second part of the presentation. This deals with the military tactical operation has to consider take-off and landing
conversion of a singleslotted Fowler flap to a part span double- on short and unpaved runways even in hostile surroundings.
slotted flap, ahigh-lift system which was developedby Deutsche This requires lower approach speeds, higher decent rates and
Aifbus for the A321. better climb-out capabilities. Therefore, military transports
require significantly higher maximum lift capabilities (which
may be achievable only with powered high-lift systems) and
LIST OF SYMBOLS also higher thrust to weight ratios than typical for civil transport
CL lift coefficient aircraft.
C& maximum lift coefficient
CD drag coefficient
L/D lift to drag ratio MILITARY OPERArlON REOUIRES:

a angle of attack - higher CLmax for short T.O. and Landing MILITARY OPERATION

aEt maximum ground rotation angle - higher Thrust I Weight ratio for sleep climbOut
V ~ F lift-off speed
Vsmh minimum dynamic stall speed /
Vslg steady flight (lg) stall speed
ASA all speed aileron
TPS turbine power simulator
FAR Federal Aviation Requirements

1. INTRODUCTION Figure 2. Airfield Perfonnance

Within the European AIRBUS consortium, Deutsche Airbus
@A) is jointly responsible for the low speed aerodynamics of However, the high-lift design for civil transport aircraft is no
the AIRBUS types and variants, Fig. 1. less complicated, because safety and economic considerations
Typical Pclass seats play an more important role and encourage the designer to
=t achieve the required maximum lift and climb-out capability
with a high-lift system of minimum complexity ('design to

3oo
250.
t G
cost') and maximum reliability.
A general reflection on the high-lift system design process for
civil transport aircraft will be given within this paper. First, the
major design objectives and constrajnts will be reviewed.
200-
A321 Next, the DA wind-tunnel and model strategy will be explained.
Some comments will also be given on the applied theoretical
150 - G methods.
In the second part of the paper an example for the high-lift
loo - design is given. This deals with the conversion of the A320
Range (nm) smgleslotted fowler flap system to a doubleslotted flap system,
which was necessary to satisfy the increased requirements for
airfield performance of the A321, the stretched version of the
A320.
The DA aerodynamics worksharing part covers the whole
spectrum ofhigh-lift system development, beginning with aero 2. GENERAL, DESIGN ASPECTS
performance predictions for new projects, design and The mission of a civil aircraft is to transport a given payload
optimization of high-lift devices and preparation and analysis over a specific range with maximum efficiency.The efficiency
of wind-tunnel tests. can be quantified as direct operating costs (DOC), which
 31-2

incorporate costs for fuel, airframe, crew, maintenance, taxes, Furthermorethe design has to guarantee maximum fight safety
etc. Minimizing the DOC is the main aim in civil transport which means good handling qualities, moderate approachspeeds
aircraft layout. Due to the complex relations between the and 'normal', controllable stall characteristics.
aircraft configuration and the DOC parameters, tradeoff studies
Having selected the primary aircraft design parameters, e.g.
have to be done to determinethe optimum combination ofwing
wing loading, the required aerodynamic performances of the
planform, weight and thrust for a specific aircraft project.
high-lift system can be derived from airfield performance
The choice of specific cruise capabilities, such as speed, range, investigations for the most criticalconditions. The investigations
initial altitude, climb capabilitiesand buffet boundaries defines include take-off field length and landing distance calculation,
the targets for the aerodynamic design. With consideration of in accordance to the operating rules specified in FAR 25 for
the advanced available technology in aerodynamics and civil transport aircraft.
structures,the primary wing design parameters, such as loading,
span, sweep, aspect ratio, twist and thickness have to be
optimized. 3.1 Airfield Performance Requirements
When discussing the airfield performance requirements, a
Some of these primary wing design parameters strongly
separate view ofthe takeoff and landing situationis appropriate,
influence the efficiency of the high-lift system and have to be
because the requirements on the aerodynamic qualities of the
taken into consideration in the trade-off optimization process.
high-lift system are different.
A high wing loading, for example, increases the necessary
maximum lift capability to satisfy the required airfield
performance. This will increase the high-lift system complexity 3.1.1 Take-Off
and hence the weight and costs. Increasing the wing aspect In general the takeoff performance is characterized by the
ratio is beneficial for the high-lift efficiency, while increasing field length, which is a function of the wing loading, the thrust
the wing-sweep is disadvantageous. loading and the aerodynamic efficiency ofthe high-lift system.
Another important parameter in aircraft sizing is the thrust to The interrelationsbetween these parameters can be highlighted
when reflecting the general take-off procedure with
weight ratio. For civil transport aircraft the required thrust is
consideration of the specific FAR requirements for civil
usually determined by take-off field length considerations.
transport aircraft.
Due to the interrelations between wing planform and high-lift
As illustrated in Fig. 3, the take-off field length is defined as
system efficiency, the final design is generally a compromise
the total of ground roll distance which is required to accelerate
between optimum cruise efficiency and acceptable airfield
performance. This optimum is usually found in an iterative from the resting position to the lift-off speed V ~ Fplus , an
design process. airborne distance to overfly an obstacle height of 35ft.
According to the FAR, V ~ hasF to be 1.1 times the minimum
'unstick' speed V m , which is defined as the minimum speed
3. OBJECTIVES AND CONSTRAINTS OF at which the aircraft can safely take-off with one engine
HIGH-LIFT DESIGN inoperative.V m isnot onlya function ofthe aircraft maximum
The general objective in high-lift system design is to match the lift capability, because tail interferences with the ground can
airfield performance requirements in terms of approach speed, limit the usable angle of rotation which reduces the usable lift
take-off field length and climb rate. coefficient. Therefore, i f V m is increased, the required ground

Climb Rate 2 2.4%

1 Engine out
I
L

C,
: ,,,......._...'" '
...,;; ..-

Rotation - Llft-Off UIC In Climb Out

Ground Roll Distance Airborne

fO
f-ekaT; Fieled Length
1- .Segrnent+2.Segment -
Figure 3. Takeoff Procedure
 31-3

roll distance inevitably will be increased, too. Especially for Another aspect of flight safety during approach is the pilot’s
aircraft versionswith stretchedrear lbelage this V m limitation ground visibility, which is dependent on the pitch attitude of
can be of particular significance for the high-lift design. In that the helage and consequently angle of attack. Especially when
case the maximum usable angle of attack (not the maximum designing a high-lift system with slat or Krueger flap, care
achievable lift coefficient) will be the dominant parameter must be taken that their effect of increasing the usable angle of
which limits the usable lift of the take-off configuration. attack is not compromising the ground visibility. In this case
efficiency of the trailing edge flap must be improved to achieve
After lift-off, the next speed of interest is the take-off climb
the required approach lift coefficient at reduced angle ofattack.
speed V2, which must be achieved when the aircraft reaches
35ft height above the ground. V2 must be greater than 1.2 times Furthermore, the adverse effects of landing configuration L D
the minimum dynamic stall speed VSmb,and greater than 1.1 on the go-around climb capability and the approach engine
times the minimum control speed VMC.Vsmb is usually 0.94 rating has to be considered.A lower approachL/D, for example,
times the stall speed in a l g steady flight, a flight condition has to be compensated by higher engine thrust rating, which
which is comparable to wind tunnel conditions and therefore typically causes better engine response characteristics for flight
related to CL., Hence, V2 must be greater than 1.13 times path corrections or transition from approach to go-around
Vslg, or in terms of lift coefficients: CLv2 must be below condition. This item can be a partial problem for aircraft with
C&/l. 132. single-slotted flap high-lift devices, because their typical
After take-off, the second segment climb begins when the approach L/D may be ‘too good’. On the other hand, a lower
undercarriageis retracted. According to the FAR, the minimum
L/D degrades the go-around climb rate, which must be greater
second segment climb-rate with one engine inoperative must than 3.2% with all engines operative. Therefore the limitation
be greater than 2.4% for aircraft with two engines (3.0% for of the landing weight is the usual way to provide the required
aircraft with four engines) and the minimum speed must be V2. climb capability.

Approximately the climb rate is a function of the thrust to In principle, the required maximum lift capability for the
weight and the lift to drag (L/D) ratios as defined by the landing configuration determines the complexity of the high-
following equation: lift system, in particular the number of slots (or elements) of
the trailing edge devices. The degrading effect of wing sweep
on the maximum lift efficiency necessitates an increase in the
Thrust
CR= - complexity of the high-lift system.
( Weight UD
The general trend of maximum lift efficiency versus system
This means, for a given thrust to weight ratio, the climb rate is complexity is illustrated in Fig. 4.
directly related to the L/D.
A typical envelope of L/D versus CL is shown in the small
diagram within Fig. 3. The steps in the envelope separate the
different take-off configurations.The right end of each envelope
segment indicates the maximum usable liff coefficient for
second segment climb of the specific configuration (usually
CL
J.,, 132).
It is obvious, that a higher lift coefficient, as achievable with a
higher flap setting for example, decreases the L/D and ~ UNPOWflfO HIGH-LIFI SYSIEYS -PGWfIfD HIGH-LIFT SlSTfYS -
0
consequently climb rate, while the ground-roll distance is COMPLEXITY OF THE HIGH-LIFT SYSTEM
reduced.
Figure 4. Lift efficiency versus high-lift system complexity
Therefore the aerodynamic optimization of the take-off
configuration is aimed at finding the best compromisebetween
lift capability and LJD efficiency to satisfy the requirements Firstly an example is given of some transport aircraft
for both the takeoff field length and the climb rate. demonstrating the limits of conventional unpowered high-lift
devices, and secondly of some experimental aircraft
demonstratingthe efficiency of powered high-lift devices. It is
3.1.2 Landing obvious, that the maximum lift limit for unpowered high-lift
The final approach of civil transport aircraf is performed on a systems (on a aircraft with typical 25deg wing sweep) is in the
glide slope of 3 deg. The typical approach speeds of major jet order of 3, while powered high-lift systems with additional
powered aircraft lie between 13Okts and 15Okts. Due to the active boundary layer control may achieve maximum lift
evident correlation between approach speed and accident rate, coeficients up to 7.
the landing performance design is mainly concerned with
achievingmoderate approach speedswithin the aforementioned
range. In order to provide some future stretching potential, the 3.2 High-Lift Design Constraints
design approach speed for a new project is normally more The cruise wing design determines a lot of important design
orientated towards the lower value. parameters for the high-lift devices, such as chord and thickness
The FAR requires that the approach speed has to be 1.3 times distribution, aspect ratio, trailingedge kink location, etc. Only
higher than the minimum dynamic stall speed Vsmin, which the type of the high-lift devices, the shape, the spanwise
has been previously defmed as 0.94 Vsle. This l g stall speed extension and the settings can be chosen by the high-lift
and the design wing loading for landing determine the maximum designer, but with consideration of some constraints, which
lift coefficient which has to be achieved by the high-lift system. will be discussed next.
31-4

3.2.1 General Constraints Also influencedby the relative flap thicknessis the aerodynamic
Usually the chordwise extension of the high-lift devices is efficiency.Usually the flap efficiency is limited by the onset of
limited by the location of the front-spar and rear-spar flow separation on the upper surface. Therefore the main
respectively, which can not be changed due to considerations objective in flap shape design is to increase the angle of flap
of wing stiffness (twist, bending) and internal fuel volume, deflection at which flow separation occurs. This is usually
Fig. 5 .

,.,’
-
Slat TraillngEdge Gap
done by controlling the pressure peak at the nose and the
following pressure gradientsby the adequate shape design. It is
apparent that a thin flap offers less possibilities to modify the
shape and thereby the pressure distribution in the intended
manner. Unfortunately this can cause some loss in efficiency.
,..’. (
i.....S
A321 Airfoil Section
Slat Chord Flap Chord
3.2.2 Constraints in Kinematic Design
&
Thin Rear Section

...... .....
We at Airbus prefer slats as leading-edge devices instead of
Krueger flaps due to their modest support complexity. In a
& KNeger Flap
A33O/A340 Airfoil Section typical design, e.g. as represented on A320 and A340, the slats
are supported by circular tracks, which are directly actuated by
rotary drives, Fig. 6.
Figure 5 . General Constraints on High-Lift Design
Track

Especially the required fuel capacity for a long-range aircraft

can be of particular significance in the wing sizing. Moreover,
the inner wing flap chord of a typical low set wing aircraft is Un
limited by the required storage space for the retracted main
undercarriage.
After the chordwise extension of the leading edge and trailing
edge devices has been fned, the next design item is the
optimization of their shapes.
The typical leading edge devices of today’s transport aircraft
are slats and Krueger flaps. In the case of a slat, the profile of Figure 6. Support of High-Lift Devices
upper and lower surface is defined by the cruise wing nose
shape. Therefore only the shape of the slat inner side and the
nose of the futed-wing can be optimized. Usually the radius of each slat track is different in spanwise
direction due to wing taper. However, with small spanwise
A Krueger flap with a folded nose or flexible shape, as an changes of the slat setting and some variation of the track‘s
example, generally offers greater design freedom to achieve an vertical attachment to the front-spar, the radius of several
ideal upper surface shape, and thus gains a little in LID and tracks can be made identical, which simplifies manufacturing
C b m But, trade-off studies carried out in the past for A320 and saves cost.
and A340 have shown, that this advantage for the Krueger flap
is compromised by a more complex and heavier support It is apparent that both the aerodynamic efficiency of the
structure than required for a slat. trailing-edge devices as well as the complexity of the support
system depend on the number of flap elements employed and
For future projects the trend of preferring slats instead of on the type of the flap support itself.
Krueger flaps may change, because the envisaged wings with
laminar flow technology can not be realized with any small The structureof a track type.flap support system, as an example,
step and gap on the upper surface, as are usually present behind is generally more complex than that of a dropped-hinge or a
the trailing edge of a retracted slat. For this reason, Krueger linkage support system.
flaps will offer some advantages, because they are typically But typically, the track type kinematics offer more design
retracted in a storage bay on the lower side, which is a less keedom to realize the optimum relation between flap Fowler
sensitive region of the wing. movement and flap deflection for both take-off and landing
The crucial point in the chordwise layout of the trailing-edge configurations. This is the reason why track type flap support
devices is their relative thickness, which is dictated by the rear systems are inevitable for the utilization of trailing edge flaps
shape of the cruise wing. for ‘variable camber technology’, which may be envisaged for
cruise performance optimization of future projects such as
The typical state-of-the-art cruise wing airfoils, such as FLA or Airbus Ultra-High-Capacity-Aircrafi.
represented on the Airbus A340/A330, are characterized by a
relative thin rear-end shape. The resulting small flap thickness However, a good compromise between aerodynamic efficiency
causes some problems for the flap design. It is obvious, that the and total system complexity can be achieved - in our view- by
smaller stiffness ofa relatively thin flap has to be compensated a flap system with minimum elements combined with a support
system of higher complexity. This was demonstrated by the
by heavier structure and possibly more spanwise support
advanced high-lift systems of the A320 and A340, where
stations, which both increase weight and costs.
typical airfield performance requirements on today’s short and
Moreover the realization of a multi-element flap can fail if the even long-range aircraft are satisfied by a track supported
thickness of the single elements is insufficient for production. single-slotted fowler flap system.
 31-5

4. HIGE-LDT SYSTEM DESIGN Our experience has shown, that the evalnatedtrmds,in tams
The task of the high-lift system designer is to achieve the ofliftefficiency,profiledragandboundarylayerstability,due
auodynamic targets by choosing the type of the high-lift to shape and setting are replesrmted conectly. Problems occur
device, defining its shape and spanwise extension, and onlywiththeclosecoupledarrangement oftwo singleelaaents,
optimizhg its settings. e.g. slatandfixsdwingwithsmallslatdeflections,~sethe
applied theory does not incorporate a satisfactory solution for
In general, the lift characteristics of a two-dimensionalwing the direct confluence of two boundary layers. In generalthisis
section depend on its effective c a m k and on the boundary
not significant,because the shape optimization is usnaUy done
layer contml. The camk is hacased by deflection of a for the landing configurationwith maximum deflection of all
bnilingedge flap, and the attachment of the boundary layer on
devices.
the tlap is controlled by the setting and shape of the flap.
It has been found, i b t the optimizing of the two-dimensional
In addition the onset of boundary layer separation on the main
shape at deliberately choaen spanwise d o n s of the finite
prome is delayed with a leading-edge flap, such as IGneger or
wingis suf?icient for the three-dimensional shapedesign of the
slat,whichmcreasesthe usable angle ofattack and subsequently
high-lift devices, while the fmal optimization of the seltlngs
the m.w.imum lift capability.
must take into account the three-dimensionalflow.
When applyingthe twodimensional design to a definite wing,
the lift efficiencyisprincipally decreasedby three-dimensional
flow effects due to finite span, sweep, nacelle interaction, 4.1.2 Tkree-dimensionalMetkoda
s p w i s e limitation and cut-outs of the high-lift devices. Tbe available methods for the theoretical consideration of the
three-dimensional flow is mainly based on panel methods,
The optimization of the three-dimensionalhi@-lift design is
such as VSMRO, which is applied by Dentache Airbus in the
mainly aimed at minimizingthese effects.
high-lift &sign procedure.
The general problem when applying panel methods is the
4.1 Theoretical Methods expensivepreparation of the panel grid, which is nmssary to
Athoughthe applicationofadvaacedtheoreticalmethodstends representthecomplexgeometryofawingwithextendedhigh-
to play an even more importantrule in the aerodynamicdesign liftdevices.Th~fore,theapplicationb~ti~withinthe
of ahma?,thek use in the high-lift field is virtually limited to iterative design process, e.g. for &g optimization, but is
two-dimmsionalpmblems.Becauseofthecomplexflowmund nseful to a d y effects on the spanwise lift disbibutionand the
a three-dimensional wing in high-lift configurations, the lift curye slope in the linear regime.
neceseityofwind-tunneltestingfor configuraon optimization
and especially for stall investigationis unquestioned. Athoughthepelmethodincorgoratesappro~ate solutiom
for the thedm ' ensional boundary layer consideration(which
obviously impruvm the lift c w e slope roplegentation), they
are insufficientfor prediction of the stall characteristics or the
4.1.1 Two-dimensional Mefkods total drag, because significant intmactiom between bomdary
Usually the optimizationof the two-dimensional d e w of the layer and wake/vortex flow are not considend.
high-lift devices is iteratively done on the h i s of calculated
prebsnre distxibntion. Due to the significaut effect of the
boundary layer flow on the high-lift efficiency, the applied
4.2 Wlad-Tunnel Strategy
theoretical methods must incorporateviscous eExts.
In spite of the fact that any wind-tunnel teat results need some
ThemethodusedbyD~tscheAirbus(1)isbasedonapotential interpretation to be transfared to the real flight condition,
flow theory,which is combined with an inverse boundary layer whichisnndoubtely oneofmemost di!licultitemsinhi@-lift
and wake flow iteration. This method in~orporatessolutions design, themostreliableresults cm be generallyobtainedwith
forthe followingtypicalflow problems ofmulti-elementhigh- large scale models in large and pressudzed high Reynolds
lift airfoils,Fig. 7 nmber tunneb. Due to the expense of such tests, the normal
practiccistocarryoutmostoftbedev~o~ent~insmaller
-transition &om laminar to W e n t boundary lay= tunnels at lowerReynoldsnum- and tbm use large tlnmels
- &- of flow separationregions, which can be for verification of the achieved optimization, or for the
either short separation bubbles or large region investigation of special itanS.
separations &om one or more elem-
- conflnence of wake and boundary layer flow, even for 4.2.1 Deutsche Air& Low-Speed Wind Tunnel
multi-elemmt conkiguations. Deutsche Airbns is in a position to cany out most high-lift
developnentwodcinitsawnlow-spedwindtatBremen,
thetest section ofwhichmeasws2.lmto 2.1~1, Fig. 8. This is
a sitnation which o& Deutsck Airbus the advantage of
more test flexibilitym the developmentprocess.
Wemostlyusehalf-m~lswithaspanofapFwximately1.3m.
Shwh I B.L. In gened, this s p n is sufficient for a correct geometrical
... representation of the threadimensionalwing, inclnding the
high-lift devi-. Typically, our half-model8 are equipped with
through-flow nacellsa, but turbine p o d simulators (TI'S),
WING FLAP blown nac8lles or pmpfan simulators can also be employed to
simnlatepropulsion effects. The availabilityof state-of-the-art
Figure 7. Flow on High-Lift Airfoil equipment for surface flow-visualization, wake-flow
314

Usually th8 tests am pafonned at a Mach n u m k of 0.2,

achiavingaReyn0ldsn~of2.6*10~.Thes~oftypically
employed complete models is in the order of 5m. This size
gmmttees a highly W ed rsprcssntation of the airaaft
Bwmetcy including nndorcaniage, moveable surfaces, traclcs
and fairings.
Generallythemodelis supportedby amoveable sting. which is
commted d k d y to the fuselage intend balance. The six
component strain gauge balance has demonstrated high
sensitivity and long time repatabsty.
Therefore, withrespect to the excellent flowqualityintermsof
low bubulence, this -el is primarily used for low-spmd
pmformsnce investigationd. with special regard to second
sepmtclimbIJD. These WtsincludeinWvejetintafemm
investigstiomby means of turbine powered simulatom (TPS).
Furthermore the DNW funnel is especially suitable for ground
effect investigations, even with jet simulation by TF’S, due to
its moveable model support and effective sepamtion of the
tunnel Uoor boundary laycr, which additionally is combined
with a moving belt gmund plane.
Theinvestigationofwingdownwash,tailefficien~andlataal
stability completes the standard test program which Deutschs
Airbus Csnies out in theDNW.

Figure 8. Deutsc6e Airbus Low-Speed T~mnel 4.23 Oh‘ER.4 FI Tunnd

v i ~ o n , w a k e - ~ ~ ~ m a l y s i s d p r e s s u r e ~ b u t The
i o neffects of Mach and Rcynolda nun& variation on UD,
measurements offers excellent conditions for high-lift maximum lift coefficient and stall charaoteristics are
development WO&. mveatigated on a second low-apeed complete model in the,
The didvantageof the low Reynolds n m k , which is m the French ONERA F1 tunnel, Fig. 10. This tunnel can be
orderof1.3*1O6,isgenerallycomplmsafedbyverificationof presslldzadupto4bars.TberebyReynolds~bersof6*1O6
theevaluatedmodificationtrendsathi~Reynoldsnum~. areaohievableatO.2Machmmiberwiththe~i~yemployed
This IS Usuauy &ne by testing some confi@mtions with the models oft- ‘pan.
samehalf-modelintheONERAF1humel, whichisapreasurized
funnel allowmg Reynolds numbers of 6*106. The practice of
‘backAo-back’ Wting the same model in both lwmels for
several di-t projectahas led to compndmsive experience
in the interpretation of half-model results obtained at low
Reynoldsnumbers.

4.1.2 German-Dutch Wind-Tunnel (DNW)

Another wind-tunnel employed by DA in the high-lift design
process is the DNW funnel, which is a large atmospheric low-
speed -el, sitnatedin theNetherlands, with a test section of
8m by 6m, Fig. 9.

Figure 10. ONERA F1 Tunuel

4.3 Hlgh-Lift Design Process

As outlined in Fig. 11, the high-liff system design pmce8s can
be split in three successive phases: pre-development,
developmentand preflight

4.3.1 Prc-Development Phase

The aim of the pre-development phase is to demons!mte a
high-lift concept which achieves the required airfield
pformamx of anew project. This is Usuauy done in an
Figure 9. Gennan-Dutch Wind-Tunnel iterative &signprocess.
 31-7

Within the iteration cycle several alternative concepts are liftmg-surface theory, which were modified to incorporate ow
develop& on the mis,,fae bsedesign. comprehensivedata base o f empirical investigations.
First, the aerodynamic characteristics ofthe differentconcepts ne estimated aerodynamic of the most
aretheoreticallyestimated.Themethodsappliedforthispurpose high-liflconceptsarethenwedto establishapreliminaryaero-
were developedby DA and are basedmainly on litling-line and data-base for airfield performance calculations. At any stage of

PROJECT DEFINITION

START OF PREDEVELOPMENT. .............. ...............................................

OEFlNlTlON OF SPANWISE EXTENSION LIFT EFFICIENCY

WING DESIGN AND CHORDWI

AERO DATA BASE

AERO DATA BASE

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

TAKE.OFF AND UNOING

.DESIGN FREEZE .... ................................................. ....................................................

TAKE.OFF AND U N D I N Q
PERFORMANCE CALCULATION

-FLIGHT TEST

Figure 11. Scheme of High-LiflDesign Process

31-8

the development phase the results of these calculations are

directly compared to the required airfield performance to
demonstrate the achieved design progress.
After the initial design cycle has led to a realiible high-lift L.E.: SLAT (243
T.E. :SSF ( 32.6)
concept, the next step is to design in W t e r detail. At this
+ ASA dmop
stage. more use will be made of wind-tunnel testing, usually
starting with a Swept-Constant-Chord-Half-model (SCCH).
Thismodel isusell forpr*optimizing settings and prelimhag
check of the aerodynamicloading on the high-lift devices.
The next step is half-model testing with correct representation
ofthethredimensiod wing,thenacelleandthewing-fwelage
juncture. The half-model is provided withvariable brackets for
setting optimization and is also employed for flow-visualization,
preliminary perfonnance testing, pressure distribution L.E.: SLAT (29.' I 25.4 )
measurements and wakeflow analysis. T.E.: UB DSF ( 41')
OIB SSF (31 .E*)
The optimized design is then tested on a large scale complete
model in the DNW facility. Finally, the resulk? of the complete
model tests are used to demonstrate the achieved aerodynamic
performance of the pre-developed high-lift concept.

4.3.1 Dwelopmeni Phose L.E.: SLAT (21')

Usually the fmal design is initiated by the official go-ahead for
anew project.
Ibe task of the high-lift designer is then to adapt the p m
-->
developed high-liftconcept to the geometry of the f d cruise
wingdesign,wbichhastobedoneinclosecooperationwiththe
structureand kinematic designer. L.E.: SLAT (20.11'I 24')
Am0 E.: SSF (32')
The fmther development of the high-lift system is mainly + FLAPERON
driven by improvingthe C b a in the landing and the LID in
the takeoff configurations. This is done by the iterative
optimization of the senings and the geometric details, such as
slak/pylonjuncture, nacelle strakes etc. Within this work, the
wind-tunnel testing is intensified andexecutedinthepreviously
mentioned sequence ofhalf-model and completemodeltesting.
The obtained aerodynamicperformance in terms of lift slope,
drag and C b m are fmt scaled to !light conditions and then
trimmed toagreedreferenceconditions.Theappliedmethodas Y
developed by Deutsche Airbus (2) incorporatestrends between Figure 12. Anangementd of High-Lift Devices
wind-tunnel and flight test, which have been derived bom
previous projects.
limitations of the high-lifl devices, Fig. 12.
These interferencesare dominated by the effects of wakes and
4.3.3 Pre-Flight Phase vortices which influence wing stalling characteristics,induced
After design freeze the complete models are updated to the drag, ilap vibration and tail buffet. Typically such wakes and
final aircraR confignmtionand then tested in the DNW tunnel vortices are shed from the edges of the extended high-lift
and ONERA F1 hmnel, respectively. Both tests generally deal devices, their tracks and the nacelle-pylon-wing juncture.
with f d performance check-out. The DNW test is aimed
Therefore,the efforts in threedimensional design optimization
more towardsthetakeoffpaformancewithRlgineinterference
fmtly are concentrated on these details to minimize their
and ground effect investigations,while the FI test is primarily
&himental effectson the aerodynamiccharacteristics. Secondly
aimedatevaluatingReynaldsnumbertrendsandmaximum lift
the beneficialeffects ofattachedvortex flow areutilized on the
performance.
wing boundary layer and thereby on the stall behavior, e.g. by
The combined results, obtained bom two differentmodels in application of nacelle strakes.
two excellentwind-tunnels and then scaled to flight conditions
This considerationis of siiyticant importance since the close
with the experienceofpriorprojects, represent a pre-flight data
coupling between high-bypass nacelles and the wing (e.g.
base of high confidence. A320andA340)dominatesthewingstallandC~.Typi~y,
the close nacelleking arrangementrequiresa slat cutout at the
pylon, which subsequently initiates premature stall further
4.4 Items of Deslgn Optlmiutlon downstream, Fig. 13.
An essential part of the design optimization is aimed at
minimizing the three-dimensionalinterferenceeffectson high- Nacelle strakes, which shed a vortex at higher angles of attack,
lift performance, which are usually caused by the nacelle are employed to feed energy into the boundary layer flow
installation, wing-fuselage juncture, cutouts and spanwise which delays separation and improves C&=.
 31-9

WCEUE S m E
Figurel5.AirbusA321Airliner
-
Figure 13. WakeNortex Boundary Layer Interaction
The major goal for the aerodynamic design of the A321 is
The beneficial effect of vortex boundary layer interaction can dehedbythe requirementthataimeldandcruiseprfomance
also be utilized to delay flow separation at the wing root shouldbecomparablewiththatoftheA320.I t i s a p t t h a t
!miling edge reaon by adequate design ofthe slat-end h l a g e an optimized design as represented by the high-lifI system and
juncture. the cruise wing of the A320 needs some modifications to
master a 13% weight increase with optimum efficiency.
The optimization of the wakdvortex flow effectsrequire flow
field investigations.UsuaUy this is done by application of the Typically the design efforts on stWched aircraft versions are
Crowder wake ima@ng system, a 6ve-port-probe pressure drivenbydemandsofminimumc~g~tothebaselineaircraft
survey system and smokdaser vortex visualizationtechnique, Therefore the necessary wing structure modifmtions for the
Fig. 14. A321 are restricted to the war part of the f m h g edge fhp.
Thesemodifidom includeaspmvk-gchord actensia
which increases the wing area by approximately 2.6%. and
replacement of the s@e-slotted flap by a double-slotted flap.

5.1 Requirements of the High-Lift System

The requid efficiency of the A321 high-lift system was
drivenbythe handofhaving c o m p a r a b l e a i r I i e l d p d o ~
withthatoftheA320.
In terms of MI, a 13% increase in weight necessitates an
in- in the operational lifI coefficientsof the same order.
Furthermore the rear fuselage insert reduces the maximum
Figure 14. Wake-Flow behind A340 Wing in T.O. allowable ground rotation angle for takwff and landing flare
by approximately2 degrees, Fig. 16.
Summarizing the required lifl capability demands 13% more
It is apparent, that the nacelle/wing interference problem will
be significantly increased by the introduction of very high- lifl at 2' less incidence.
bypass engines ('Super-Fan') to hturs projects. This problem Especially the reduction of the usable incidence requires a
highlights the need for three-dimensional flow field significant increase in the lifl coefficientat constant incidence
-
investigations especially with engine jet simulation to - (CLo). This can usually be achieved by increasing the effective
demonskate the flow-interactionsand then to minimize their airfoil camber, e.g. by further dekting of the traihg-edge
penalidng effects. flap. However, this is not a practical solution for the A320's
singleslotted fowler flap, because a deflection beyond the
present 40° for landing will cause flow-separationon the flap
5. FEATURES OF THE A321 EIGE-LIFT and subsequently a loss in lift
SYSTEM
Ageneralreflectionofthehigh-liftdesignpmcesswasgivenin
the pr&%ding sections of this paper. One example of this
process. the development of the high-lifI system for the Airbus
A321, will bs -bed next.
In 1989 Airbus Industde reacted to the increasing market
demand for a 180 to 200 seat short to medium range airliner,
and decided to close the gap between the existing 150 seat
A320 and the 220 seat A310 by astretchedversion ofthe A320,
the A321, which will fmt fly in March 1993, Fig. 15.
In comparison to the baseline airrmlt, the A321 fuselage is
stretched by 8 framesahead of the wing and 5 frames behind,
which in-s the payload capacity by 36 Sears and 3 LD3 0
' a
containers. This stretching cause8 an 13% increase of the
maximum takwff and landing weight Figure 16. Lifl Requirement on Stretched Version Aimaft
31-10

The Sltwative was to replace the singleslotted flap by a theeil"oftheshnud~thismlative.ly smalliucomprakm

doubleslotted flap, which allows higher deflections without to the total tab efficiency. Therefore, inclubg consideration
flow ssparations. of shroud stiffness, a shroud length of 10% of the total flap
chord was W y chosen, Fig. 17.
With a double-slotted tlap wncepl, the required lift efficiency
m terms of CLo and C& may be achiwable with any
conriguration. But, the doubleslotted tlap will cause more
5.2.2 Three-Dimensional Layout
profileandinduceddragatcon~tliftInadditionthefricti0~
Frmtheaerodynamicpintofviewthe spanwiseanangement
drag of the fuselage insats will fuahm inaeasc total drag.
Thus,thecompromisedLD oftheA321 wasacrucialpointin of the A320 trailing edge devices reprpxmts au optimum
takeoff paformame consideration, and the major design
design,because the spanwise flap extension is not divided by
my cut-out and even the intersstion between inboard and
effortswere conwdrated on takeoff L/D optimizatiou.
outboard flap is sealed at all settings. Thisassures a continuous
With consideration of the previously given approximation for spauwise lift dwhibution and minor penalties due to wake and
the climb rate, it is apparent that the requued second segment vomx flaw. The spanwise lift diahibutionis furthes improved
climb perfomance can only be satisfied ifthe weight and drag by aileron dm~ping.
rise is compensated by an increased takeoffpower. This was
offered by thrust enhanced versions of the A320 power-plants, The fht design intention was to rralize a continuous double
slonedflapalongthefullflapspan.Unfommately,thisconcept
the CFM cFM56-5B1 and the IAE V2500-A5, which both
had to be changed due to structural constrain@, because the
offerapproximately16% moretake-offpower.The availability
inboard tab was partly hid by the engine's fan jet, and the
ofthess @e vwions were prereqUsite for the feasibility of
outboard tah end became to small for manufachue. Therefore
the A321.
the tab span was finally limited on the I/B flap W e e n the
fuselaseandtrack 2 and on the O/B flap W e e n the kink and
(rack 3, as illustrated in Fig. 18.
5.2 Elgh-Llft System Derlgn
The layout ofthe doubleslotted flap system was dominated by r
the demand for minimum structural changes. Inphcicular, the
nosepat, the track-support and the actuation system of the
A320flapbadtobeprpservedwithwlyminorchsngesn~
to compensate for higher loadinga This restriction generally
excluh the application of a vane flap type doubleslottedflap
system, as represented by the inboard flap of the A310.
Foramstelythe relatively thick rearprofile oftheA320 allowed
an auxiliary fowler flap (tab)to be incorporated into the rear
part of the main flap.

5.2.1 Two-dimensional Layout

Affa pre-development wind-tunnel testing of the principle
flaphb concept had demonstrated satisfactory M?efficiency,
the finaldesignwasinitiatedbyoptimizing the two-dimensional
m g e m e n t of tlap and tab. Figure 18. Wing Planform of A321
Therefore studies were carried out to investigate the effect of
5.23 Kinematic Layout
theratio oftab chord to the total flap chord and the lengthoftbe
The A321 main flap element is supported by a track guided
&mud on the profile drag, lift efficiencyand stall sensitivity.
carriage and actuated by a rotary drive system. The new tab
These studies, based on viscous pressure distribution element is supported by a four hinge linlrage system, which is
&uhtions,iudicatedfirstly, thattheoptimumtabchcndwm i n d i i y actuated by a drive rod, Fig. 19. This md couples the
in the order of 40% of the total chord and secondly, Ihat the movement ofthe. main flap carriage with that ofthe tab linkage
~udlen~shouldbeaslargeasposSible.IngeneraZhoweve, SUPPOL

A320 Single-Slotted Flap Section

----

A321 Double-Slotted Flap Section

Figure 17. Comparison ofA320 andA321 Flap Section Figure 19. A321 Flaplhb Support
 31-11

satisfiesthe aemd~amicrequbmentsinterms ofgap, overlap

and tab/flap deflection for all conEgurations,Fig. 20.

5.1.4 Setting Optimization

In general, the profile drag at constant lift increases with
profile camber, or flapltab deflection, respectively. Therefore
theoptimumL'Dfor each take-offconfiguration can be obtained
with the minimum flap/tabdeflectionthat achievesthe required
lift.
?be determination of the minimum tlapltab deflections was
based on the diagram shown in Fig. 21. The @id demonstrats
the lift coefficients which can be obtained with different
combinationsof flap to tab deflections at a constant incidence
of 9.7",which repremts the operational limit of the ground
rotation angle. The required lift coefficients for the three take
off codigurations are marked on the left axis. Additionallythe
function of the designed Ilapto-tab gearing is given as a
&Flap function of the grid. By means of this curve, the feasible flap
Figure 20. Flaplrap Settings and tab settings were determined which achieve the required
lift Coefficients.
In comparison to an alternatively investigated track support, Wind-tunnel investigations have demonstrated that the WD
the linkage system is simpler, lighter and smaller, and can be could be further improved by modification of the spanwise l i
integrated completely into the main track fairings. distributionthmugha differentialdeflection ofthe IEIand O/B
tab. This aspeft was realized by the fmal kinematic design,
A disadvantage of the linkage system is the restricted function which provides a maximum deflection of 20" for the tab
of the tab movement relative to the tlap, because it can be
and 25" forthe O/B tab, while the main flap's deflectionis 35'
defmed for only two settings, usually for one intermediate
on I/B and O B .
takeoff configurationand for the !anding configuration.
The intention ofthe initialkinematicdesign was to keepthetab
gap sealed in the takeoff configurationsfor optimum L'D,and 5.3 Performance Status
open with a large slot for landing to achieve high CLmm. The aerodynamic performance, achievedwith the A321 'S high-
However, ow intention could not be realized due to restricted lift system during high Reynolds number wind-tunnel tests, is
tab kinematics. Therefore, takboff seaing variations were given in Fig. 22 for the lift and in Fig. 23 for the L'D.
intensivelyinvestigatedduringwind-tunnel testing. The results Especiallythe increments of CLo and C h m given relative to
suggestedthat a small gap of 0.5% of total flap chord provided the A320's singleslotted flap demonstrate the superior lift
little better WD than a sealed gap, and further, that gaps up to capability of the part span double-slotted flap system.
1% could be tolerated. This increased gap margin relaxed the Additionally the lift increments for a singleslotted flap are
kinematic constraints and allowed a kinematic design which given, the chord of which was identical to the doubleslotted

TO

Figure 21, Flapnab Setting Optimization

31-12

flap with retracted tab. Its marginal performance, especially The second part of the paper dealt with the design and
the W i n g maximum lilt at deflections higher than 2 5 O optimization of the A321's trailing-edge high-liftsystem.
demonshates that the required lifl capability for the A32 1 can
only be satisfiedwith a double-slofted flap It was demonstrated that the significantly increased lift
requiremenrsforthestretcbedversionaircraffcouldbesatisfied
by a conversion of tbe basic single-slotted fowler flap to a part
span double-slotted flap, even with consideration of the
'minimum change' philosophy.
The forecasted growth of the world-wide civil air transport is
obviously limited by the traffic - slot - capacity of the major
airports, with some of them having already reached their
operating limits. lhis situation can be eased by improving the
approach and take-offproceduresto in- the airport traffic
density and by using smaller airports even for long range
operation for trafficdecentralization.
Therefore, the airfield performance of fiture civil transport
Figure 22. A321 Lift Efficiency aircraft need to be increased in terms of decent and climb
capability, field length and maneuverability, thereby tending
towards the airfield performance ofmilitary transport aircraft.
The envelope of LID versus CL for all take-off canfignrations
is illustrated in Fig. 23 for A320 and A321. In comparisonto
the A320 it is shown that the operational lift regime for all References
A321 configurations could be extended with only minor
deterioration in LID. When considering the total amount of 1. G. Dargel, H. Jacob, "Berecbmmgsverfabren tiirviskose
drag increase due to fuselage extension, auxiliary Iab-hinge Klappenpmfilstriimungen - Rechenpmgramm HILI",
fairings, tab-edge vortices and increased effective camber, it is ZKP, Fliigel kontrollierter Strijmung, Ergebnisbmicht
evident that the LID efficiencyof the double-slotted flap alone Nr.9.LFK85105
is slightly better than that of the A320's single-slottedflap. 2. B. Hahann, F.4. Debbeler, H. Gielen, 'Takeoff h a g
~
PredictionforAIRBUSA300-600 andA310 Compared
'C with Flight Test Results", Journal of Aircraft, 25, 12,
December 1988, pp1088-1096
3. R. Hilbig, "Aktuelle Entwicklungen in der
Aerodynamik",DGLR 91-155, September 1991
4. M.Win,A.Haig, "Modificationder AIRBUS A320
Fliigelbinterkante ZUIOptimieruugder Hugleitstungen
fILr die gestmkte Variante A321", DGLR 90-074,
October 1990, pp1275-1283
5. D. Schwetzler, "EffectofSlat 1 InnerEndModification
on Wing Root How", Deutsche Airbus, EFI2-13B1,
Figure 23. L'D Comparison of A320 and A321 199I, (unpublished)

6 CONCLUSIONS
A general reflection on the high-lift design process was given
in the 6rst part of this paper.
It was discussed, that the application of theoretical methods in
the high-lilt design process, at present, is sufficientonly for the
two-dimensional design cases such as shape optimization.
In spite of the advances in three-dimensional computational
methods, theiruse for maximnmliftandtotal drag prediction is
still insufficient due to the unconsidered complex inmactions
between boundary layer flow and wakehortex flow on a wing
in high-lift configuration. Therefore the necessity of wind-
tlmnel testing is unquestioned for the high-lift development as
well as for performanceprediction.
TheDeutscheAirbuswind-tunnelshategywasdesnibed,which
is aimed at carrying out the majority developmenttests in OUT
own low-speed tunnel and to employ the large tnnnels for
verification, fmal performancecheck-outs and investigationof
spezial items.
 32-3

4.6 Repdda' Number Eticets on Advanced

AIIfOU.9
- double-slotted flaps#also operable togsther as
The effects of Reynolds number were also pa l fdd-dPa spau,camberedleadingulgeslat
ed for the 16% thick advanced airfoil and - slipstr~amh & t h BIL eight b u d
L Z t h with and wihut the leadvgedge slat. pmpellu drivenby an 80 hp air motor
Typicallift results for a case without a f with the - w m p r d air supply to the motor via internal
flap location optimized for 6 million Reynolds wingducts
number, are shown in 6gure U Tests made at a
reduced Reynolds' number of 2 million for the same
- the ability to return the motor d u s t back
thr~@the wing and thence &-balance with
location s h o w e d a h inmaximumlift performance lowiatcrfercnec
of about 5%, and there was negligible change install - a detachable outer wing panel m tbat the
incidence for this tion. Atthelowerflap model canbe installed in t h e m 15mx 15m
anglesthe smaller liff losses tramonk tunnelfor cnrisc testing.
due to Reynolds' number effects (2-5%), in wntrast - a removable tail lane for measurements of
to the behavior found with the commuter airfoii downwa~hand ta$plane/clnrator eftectiveness.
Typical results for tests with a leading edge. slat are The IAR 2m x3m low speed wind humel in Ottawa
Shawn in figure 14 for a eonfiguration with double- was used for the half model t d u g . This tunnel can
slotted flaps detlected 15/15 degrees. In this case ate up to 035 Mach number. For the majority
the tests at lower Reynolds' numkr showed
m
negligllle changes in below stalling angles. Thc
XL
uacd,
high lift ~ a c numb
testing a h of
to Rcynolds number of
carreaponding a
was
0 2
15
maximum lift was penalized lass than 2% for this million based on the mean chord of the
case, while the stall incidence was found to be
essentially unaltered.
limited amount of testing was
speeds on the wmmutcI
Tat
fledomled A
anng to investigate
OWCI

Reynolds n u m b eflcds.
The airfoil section testing without slats indicates that
larger half models may be dcsirablc for flap 52 Dewlopmeat d M d to Floor salinp
development work, in order to diminish the Early on in the half model high lift te
corrections nceded for Reynolds' number effects.
In the case.of models ha leading edge slats, the
realized that the fusclaec drag Lcvc were
unacceptably high due to excc88ivc
yawas
e bctwccn
3
scale of the current halfm el appears adequate for
preliminary development work.
the model and the tunncl floor. %itwas
accepted tbat such instalMona do not normally
provldc amrate drag measurements, there was
5. HALFMODELTESTING Wuld af€& lift also, so it was
wusidercd desirab to make improvements.
w-mthatl=Tj
5.l ~ L u t M o d e l F c l l t o n s
The experimental investigation used a research The 2m x 3m hume1 bas a plywood floor and a
model in the IAR 2m x 3m low-speedwind tunneL turntable supported on a steel beam structure.
The model was tested with various sings during this Models werc provided with elearanws of the order
development program. An iastallation using a h@h 6mm to allow for wa- in the floor and for ita
mounted wing with commuter airfoils is shown m lifthg under a e r o d p d c loads. Tests of the
figure 15, and its features were hruy dcscn'bcd in demonstrated the mapitude ofthe
reference 2 Tests have also studied a low wing with 17 show that M y the drag
some results reported in reference 2 A photopph was about three times eater thanpredictions. The
ation tested with the advanced a u f d gap between the m& and floor were reduced to
isshownm
of the w%e 16. rmnimums and a skirf was addcd made from
adhesive backed aluminium foil The underfloor
A half model was used to nwdmk the Reynolds' ga around the model mounting block were also
numbers and model size for a given humel, while J u c e d to minimums just avoiding fouling to
substantially reducing costs for the model and reduee venting from outside. These modi6cations
testing relative to an e+nt com lete model.
However, half models mtroduffi d a d t i e s with
sealing between the hLpelagc and the tunnel walL
enabled the fuse
2 drag to be reduced to abut
twice predicted uc11. These tests were all made
V-shaped, floor mounted, vortex
Sealing is needed to diminish leakage and the
resulting l i f t - h and anomalies in the drag and
with
enerator
a % of the model in order to thin the
%nundary layer about the modct
pitching moment measurements. Such issues will be
discussed in more detail later in the papcr. Some follow-on work has investigated cruise drag
with half models in the IAR 15m x 15m Transonic
The wind tunnel model in figure 16 had a wing of WmdTuMeL InthattunneltheinstaUationcanbe
l5Zm a d M Mlefall asped ratio of made with mueh tightcr c l e a r m as the walls are
twelve. An advand airfoil of 16% maximum mort rigid a n d h wavy,and the balance stiffmst is
thid.scss to chord ratio was used at the model high80 de-m m low. In order to improve the
centre line and one of U% at the wing tip. Some instaUation it was decidcd to use a non-met& 6Uer
features of the model included : atand-ofQ plate between the model and Wau. see
- variable span h p s extending to 60 or 80% clS. The.gapsbctwecnthefuselage,plateand
span with plain sealed ailerons outboard hrnnclwallwere small, ofthe order 1to 15- and
32-4

they were sealed using s ring loaded tenon strips. behavior to the atfoil section data disenapcd
~nthiscaac there is no L w p p t h for lca~agepast previously. A summary of the effect of slats on
the balance as it is internally sealed and the tunnel is maximum lift ooefficients is presented in tigure 22.
within a plenum. Static pressure rails are mounted These data show increments in
on the tunnel porous walls and they provide data for
corrections including model bwyancy effects. For
this installation the measured drag for the isolated
--r ding to at least 3% over the
slat-0 cases.

was about 80% of estimates. Tests on The slat increased stall augh by more than ten
wing/ lage/nacek contigwations reported in degrees out to values about 25-30 depes. Such
reference 5 showed good correspondence with extreme mgh will require special attention to the
lift/drag polars from k a f t tests. relative to current propeller driven
are satisfactory without needing
A new model and mounting -cm is now under or any leading edge d h
development for the IAR 15m x 15m wind tunnel,
to achieve further improvements in drag Also to be noted, the undersurface atall behavior
measurements. Thir u8cll complete models induced by the slat at negative mgh of incidence
mounted on a slim, centre-line plate. Again, static found in ZD tests, persisted into the wing behavior.
pressure rails are used on the walls for deriving Such behavior is not unexpected on an unswept,
tunnel corrections. Tests of a fuselage in isolation high aspect ratio while with swe t planfonns
have shown drag values similar to estimates,
although the tare corrections of the mount have yet
J
where slats are ~ m a l l yused such ects may be
obscured. Certification requires that ad aircraft be
to be established, figure 17. flown beyond Vny! (1.8 x Vs) and to near zero "g"
conditions, so safecxfursiws into this re nwillbe
From the fore oinp work it a an that the half
model approac i shU requires K er de& ment
necessary. Although it may be easible to Pl
demonstrate satisfactory 5ight in the spoiled region,
in the IAR 2m x 3m wind tunnel installationL o re the option of retracting the slats could be exercised
it can be used reliably for parameters &er than liR. to ease 1 - a control.
53 HallModel Tat Rcsults The gains in field performance from leading e*
slats will not be as &Sicant as the increases m
5.3.1 AifoiUQlap choraaerixtics With SI& c?f maximum lift they conferred in these. experiments.
The lift data from half model tests with commuter When aircraft land they normally have angles of
and advanced airfoils are shown for cases with flaps inadence in the range W O to five degrees, as
retracted, and with 25 degrees deflection, in figure dictated by 5ap angle and the requirement to
19. The data shown are for cases without a n a d e . approach at 13 Vs. During the landing flare the
The half model results exhibited similar features to aucraft pulls "g", there is a small speed loss of
the airfoil section data dixvsscd previously. The maybe 5-10%, and incidencc increases to near the
stall with the advanced airfoils occurred about three maximum permitted by tail strike, say to values of 8-
to four degrees sooner and miuimum lift was 12 degrccs. Thus, the most important factor
reduced accordingly. The reduction in stall angles governlog landing performance is the range of la
observed with the advanced wing may lead to some eocfIieient available in the range of usable incidenw
restriction on the largest 5ap deflection usable for between zero and about 12 degrees, rather than the
l a n d y , a r i s i i from the stall margins needed in absolute value of C As a result, unless the
groun effect at touchdown attitudes during usable range of lift increased in concert with
maximum effort ladings. This situation may well using slats, the will not augment landing
dictate the use a€leading edge den- particularly performance s d c a n f i y . % means that to my
on wings with thinner sections or less tolerant exploit slats, aircraft will need more powerful 5ap
designs than those. used here. systems.
The variation of maximum lift with main 5ap angle 5.4 Compuisona nlthAerodgnamle PRdletioua
obtained for cases with single and doubledotted ThC simple nature of a high aspect ratio, unswept
5a are summarized in figure 20. At larger flap wing e n a b h lift to be adequately predicted using
dek&ions the advanced wing achieved maximum lifting line methods. Comparisons are presented in
lift weffirients about 10% less than the commuter figure 23 showing predicted and experimental lift
*-
requued Aand,- %asm o f c Pm
h p s are
figure, about 15
degrees deflection on the second 5ap element was
curves for the commuter wing with 5aps retracted
and also with 25 degrees of deflection. The method
acwunts for the non-linearbehavior of the lift with
needed to remex the deGency, excepting the incidence across the wing span. The agreement is
wntribution from trim loares needed for complete n e r d y good, although sometimes difliculty is
configurations. Eund in predicting maximum lift reliably at large
5ap angles and semi-empirical tixes to the input
5.3.2AitfoiWlap Characteristics With Skn-On data are used to force solutions.
Exgcrimental lift curve data for tests with and
wi out l e a 9 edge slats using the wing with
advanced alfolls, are shown in 6gure 21. The.
model's lift charadcrirtics again exhibited similar
The lifting line procedure also
pitching moment contriiution. x
edicts the wing
tunnel data can
be used to obtain au approximation to this
wntribution by subtrdng the isolated l i d a g e REFEREhTEs
v a l w from valucl for thc wmplctc model. "lis,of 1. Poole, RJD, and Teeling, P,
course, leaved the intuaciion effects of wing on 'Airfoil8for Light Tram@ Aka&,
h e l g c still prcscnt in the net result. The variation SAE 81076,April 1981
of predicted pitching mom- with incidence and
those dcrind from the experiment& arc presented 2. m B v
in iigure 24 for thc eommutcr wing with flaps "R&D in the Evolution ofthe Daah 8",
retracted, whilc figure 25 shows data for 25 degrees CASI Joumd, VoL 30,September 1984
of flap. Thc agreement is r d l e with only a
small residual diffcrcnce in pitching momeats of 3. B, Poole,RJ.D, Jon+ DJ.,
about ACM= -0.04, which did not vary strongly with
inadace within thc usable range of ineidcncc. &r
ThickSu raiticalAufodswith
LowDragandNahu LaminarFbW,
J o d of Aka& Vol. 24,Junc 1987
6. CONCLUSIONS
The paper has d c a a i i m a r c h on high lift 4. Khalid, M. and Jones, DJ.,
wm- earlier wmmuter airfoils of Transonic Natural Laminar
ai&& sutt.blc for airad auidq at Dcvclopment at NAE'
.
spccaS a h Mach 0.7 This work is still on-go*
Based on thc results presented hen the foU-
"-65. May 1990
d u s i o n s arc ma& 5. m%-
My Digncy,J.R., Poolc,R J.D.,
a) Thc half model approach to high lift Doub t P a d M~thodfor Halt-Modcl
dewlopmeat was found satisfaeory for wind-Tunncl cmcctioq
prclirmnary work. It still requires further Journal of Aka&,Vol. 24, May 1981
development on model mounting and in
articular 011 fusclaec/hmncl floor scaling to
reliable for drag and pitching moments as
%U as lift.
b) The relatively thick advanccd airfoils of this
. study were- able to achieve C
performanee within 5 to 10% of &E
wmmuter airfoils when using similar flap
systems. This was demonstrated using two-
dimensid airfoil scetion tests and also with
halfmodels.
e) Leading edge slats added wasiderably to
by extension of the stall in- out
c%iYucs of 25-30 degrecs, giving gaim in
to
C h a x > 30% at "flap arlgh.
Howcvcr, in ordcr for m a d to b f i t from
slats in tcrms of achieving shortu field sizes
or inaeasca in wing thrywillnecd to
bc w d i n eMcUtwithmorc~rhlflap8.
d) Reynolds' number effects on performaaac
wcrcfoundtobc~forturtrwithout
half modcl tcsthg
ne& to bc done at
larger scale. Alternatively, more extensive
testing of two-dim- airfoils may bc
used m wnjuaction with reliable model-to-
fullscale wing prcdiebionmethodr
e) Predictingthehighliftpertormanecofhigh
asped ratio, unswept wnga as uscd in thia
study can bc done with relatively simple.
lifting-linc procCdurCk
ACKNOWLEDGEMENT
Thc rcscarch progrpms diacusscd here wen 'OW
haded projects of de Havillpnd Inc and Thc
Program for Industry and Lsborotory Pnrtidp.tion
of the National R-ch council of Canada
ycan 1980 to 1987.

0 11%t/I
 0 0 I
32-6

FIGURE 3. MODEL OF 21 % THICK COMMUTER AIRFOIL

WITH FLAPS

FIGURE 4. MODEL OF 16% THICK ADVANCED AIRFOIL

WITH FLAPS AND SLAT
 32-7

Y O U L C CAMELS COlOUltURYTAUE

VULL I . O N T U Y N L
OUTLIYE

IUCTIOY I O X
AIR101L YODCL

STAT1 C C R I ¶ S U R I TRAVERSlNl
rims (noon a CEILIWOI WANE M O U I

FIGURE 5. I.A.R. TWO-DIMENSIONAL WlND TUNNEL

-m -6 o
INCIDENCE IDEORCEIlI
32-8

6.6 -
J 6-
-

1:
4.6

3
.
:

&No. 8 uman
MACH No. 0.16

1.6

- ( o o m 2 o a o ~ oI 1 -m o 10 20 30 40 60
MAIN f L A I DEfLECTION IdWn..l
f L U DEfLECTlON td.m.d

- 1iS CQYMYlER + U S COMMUTER

* i8S ADYYCED

klh - 16/11

MACH No. 0.16

-20
-1
O -lo 0 10 20 30
r -20 -10 0 10 20 SO
INCIDENCE I D E M L E I )
INCIDENCE IDEOREES)
-M P MIII LOOW + n. 8 MI LOCN ui
- SLAT OFF 3- SLAT ON
+ L t YII l i t Moa#
OPTIMUM LOCATION
OPTIYI(IE0 FOR 8 M11110a

WURE 10. ADV- NIML Ln AOURE 91. l l F w u l S N u Y m BRCl ON C o U y u T u l N W

p((.A*-
 32-9

.L...................... i............................... i i................................................................ i.........................

.......................... L............................ ./ ._ .................. J ................................................... i ...........................

-10 0 la 26 36 46
FLAP DEFLECTION ( d * g n * d

- OPTlMlSED AT OM Am 112 MODEL LOC" ZM

FIGURE 12. EFFECT OF REYNOLDS NUMBER AND FLAP LOCATION

ON COMMUTER AIRFOIL MAXIMUM LIFT COEFFICIENT

I
0 F

0 -1
-20 -10 0 m 20 30 -lo -10 0 10 20 50
INCIDENCE IDLOREEII INCIDENCE IDEOREEII

A R. No 2 Mlll -+ R. NO 0 Mlll -Re No 2 Mlll +- Ra NO 0 Ulll

rnurria. -*WII-O~~~~-DV~~K~DIMU -14. -llDllEllERLCTfflMV-lulFou

m* mhdm
 e e

REFLECTION PLANE MODEL

Commuter Airfoils - Low Speed Tunnel

FIGURE 16. REFLECTION PLANE MODEL

Advanced Airfoils - Low Speed Tunnel
 32-11

ESTIMATED

-16 -10 -6 0 6 10 16
INCIDENCE (deg)

- COMMUTER 1/2 MODEL+ ADVANCED 1/2 MODEL

;Y- 1.6x1.6M 1/2 MODEL -€I- 1.6X1.6M COMPLETE

FIGURE 17. ISOLATED FUSELAGE DRAGS

VaIlOua Instsacltionr

FIGURE 18.

-
REFLECTION PLANE MODEL MOUNTING SYSTEMS
32-12

3.6
SINOLE
SLOTTED

SLOTTED

Nscep. Off
RE NO. 1.5 Million 4
MACH No. 0.2
#
P
1.5.
IAR TESTS
R( No. 1 .S Million
MACH No. 0.2

-10 0 10 20 30 I .

INCIDENCE (DEQREES) -10 0 10 20 30 40 50

- MAIN FLAP DEFLECTION (dogreed

COMMUTER 0 dog
* COMMUTER 26 deg
+ ADVANCED 0 dog
-8 ADVANCED 26 deg
- COMMUTER AIRFOILS ADVANCED AIRFOILS
4 ADV DOUBLE SLOTTED

FIGURE le. n u MODELS - u n PERFORMANCE FIGURE 20. HALF -

MODELS U n PERFORMANCE
shpb sknsd ~ h g sklgla Md Doubb smod Flap#

3 4-

5 3.6 -
U
E
U
3-

__RE NO.1.5 Milli

5 2.6 -
31

0v
MACH No. 0.2

- 1-20 -10 0 10 20 30
II :.l
IAR TESTS
RE NO. 1.5 Million
MACH No. 0.2

11 ' I ' I ' I ' I ' I '

-10 0 10 20 30 40 60
INCIDENCE (DEOREES) FLAP DEFLECTION (degroer)

-t ADV NO SLAT 3610 * ADV SLAT 3610 - COMMUTER AIRFOILS + ADVANCED AIRFOILS
-8 ADV NO SLAT 36116 * ADV SLAT 36/16 * ADV'D SLAT

flGURE -
21. HALF MODELS U n PERFORMANCE
Sbrgl.mIl Darbl. sl0tt.d R.pr. s*1
FIGURE 22. -
HALF MODELS MAXIMUM UFT PERFORMANCE
Sinah skmd, DOI& Sknd OIXI S M
 32-13

3.5
I

3
0'
c 2.5
B
p 2
W
8
t3 lm5
1

0.5

I I 1 I I I I

- 0 deg MEASURED + 0 deg PREDICTED

4+ 26 deg MEASURED -B- 26 deg PREDICTED

FIGURE 23. COMPARISON WITH PREDICTIONS

Commuter Airfoils Lift -

-0.2

-0.4
M O ( (
I
LAA TESTS. 2m x 3m TUNNEL
& No. 1.5 M i i
MACH No. 0.2

-0.8
-16 -10 -6 0 6 10 1s 20 26
INCIDENCE (degreer)

- MODEL DATA W 4 + MODEL-ISOLATED FUSE

* COMPUTED WlNQ
ROUE 24. COMPARISON mm PRwcllONs
canmuwnwoih.pltcNno~Rgr~
 33-1

CHOICE AND OPTIMIZATION OF A HIGH-LIFT SYSTEM

FOR AN ADVANCED AMPHIBIOUS AIRCRAFT
M.A. Averardo
M . de Leo
V. Russo
F l i g h t Technology D.T.T.-
A L E N I A AERONAUTICA
80038 P o m i g l i a n o D'Arco -
Napoli - Italy

SUMMARY configurations u n t i l recent experimental in-

The d e s i g n h i s t o r y o f a f l a p s y s t e m f o r a n v e s t i g a t i o n s o f a n a i r c r a f t s c a l e model, l e d
advanced amphibious a i r c r a f t i s p r e s e n t e d i n t o t h e f i n a l c h o i c e o f t h e h i g h - l i f t device;
t h i s paper. A l l t h e most s i g n i f i c a n t phases d i f f e r e n t aspects a r e involved: general de-
o f t h e 2D t h e o r e t i c a l s t u d i e s a n d t h e e x p e - s i g n requirements and a l l t h e f a c t o r s which,
r i m e n t a l 20 a n d 3D i n v e s t i g a t i o n s w h i c h h a - from t h e aerodynamic v i e w p o i n t , concurred t o
ve a l l o w e d t o achieve t h e f i n a l geometry o f d e f i n e t h e f l a p system.
t h e h i g h - l i f t d e v i c e w i l l b e d e s c r i b e d and,
hence, t h e d e s i g n c r i t e r i a , t h e methods o f 1.1. Aerodynamic requirements and
a n a l y s i s , t h e c h o i c e s w i l l be p o i n t e d out. design criteria
S t a r t i n g from the preliminary design o f A m p h i b i o u s a i r c r a f t , more t h a n c o n v e n t i o -
s e v e r a l f l a p systems, t h e development o f n a l a i r c r a f t , need h i g h low-speed performan-
t h i s p r o j e c t needed a g r e a t d e a l o f numeri- ce: t a k e - o f f from water, scooping ( r e l a t i v e
c a l s t u d i e s and Mind-Tunnel t e s t s i n o r d e r t o t h e f i r e - f i g h t i n g v e r s i o n ) and a l i g h t i n g
t o s e l e c t , s t e p b y s t e p , t h e most e f f i c i e n t i n high sea-state, i n fact, are very c r i t i -
geometries and t o o p t i m i z e t h e f l a p c o n f i g u - c a l b a s i c c o n d i t i o n s which d e t e r m i n e t h e o-
r a t i o n s . So, d i f f e r e n t t e c h n i c a l a s p e c t s i n - p e r a t i v e f l i g h t range f o r t h i s k i n d o f a i r -
v o l v e d w i t h t h e s e a c t i v i t i e s w i l l be d i s c u s - c r a f t a n d , h e n c e , may make a n a m p h i b i o u s a
sed: t h e c h o i c e o f t h e f l a p t y p e s t o b e i n - s u c c e s s f u l one.
v e s t i g a t e d , r e l a t e d t o t h e s p e c i a l aerodyna- Therefore, h i g h aerodynamic e f f i c i e n c y i n
m i c r e q u i r e m e n t s o f an a m p h i b i o u s a i r c r a f t t a k e - o f f and s c o o p i n g and h i g h l i f t v a l u e s
and t o t h e needs o f o t h e r d e s i g n a r e a s i n l a n d i n g and a l i g h t i n g a r e e s s e n t i a l r e -
( s t r u c t u r e , weight, production); t h e geome- quirements f o r a f l a p system. I n p a r t i c u l a r ,
t r i c a l elements a f f e c t i n g t h e aerodynamic very important i s the f l a p e f f i c i e n c y . d u r i n g
performance o f a h i g h - l i f t device; general t h e t a k e - o f f f r o m w a t e r when, a s p o i n t e d o u t
problems connected w i t h t h e o r e t i c a l and ex- b y t h e diagram i n f i g . 1 , b o t h aerodynamic
p e r i m e n t a l s t u d i e s o f m u l t i - b o d y systems. and hydrodynamic d r a g a f f e c t t h e t a k e - o f f
run.
But i n d e s i g n i n g a wing h i g h - l i f t device,
Nomenclature t h e e f f e c t i v e n e s s from t h e aerodynamic p o i n t
o f v i e w i s o n l y one o f t h e e l e m e n t s t o be
ALPHA a i r f o i l angle of attack (deg) taken i n t o account: t h e needs r e l a t i v e t o
C a i r f o i l chord t h e s t r u c t u r e , w e i g h t , c o n t r o l systems must,
CL a i r f o i l l i f t coefficient i n f a c t , a l s o b e c o n s i d e r e d . Moreover, espe-
CLmax maximum l i f t c o e f f i c i e n t c i a l l y f o r amphibious a i r c r a f t , due t o t h e
of airfoil a d v e r s e e n v i r o n m e n t a l c o n d i t i o n s where o f t e n
CL' aircraft l i f t coefficient t h e y operate, c o s t and maintenance easiness
CD a i r f o i l drag coefficient p l a y a key r o l e i n d e f i n i n g t h e general cha-
Cm a i r f o i l pitching-moment r a c t e r i s t i c s o f a f l a p system.
coefficient I n t h i s p r o j e c t , s t r u c t u r a l needs t o g e t h e r
CP local pressure coefficient w i t h problems r e l a t i v e t o t h e c o n t r o l d e v i -
0 a i r c r a f t drag ces t o be f i t t e d i n t h e r e a r p a r t o f t h e
D( 1 variation of wing have s e n s i b l y a f f e c t e d , f o r i n s t a n c e ,
DF flap deflection (deg) t h e c h o i c e concerning t h e f l a p chord, w h i l e
Re R e y n o l d s number c o s t , m a i n t e n a n c e a n d w e i g h t h a v e b e e n some
t a i r f o i l thickness (percent c) of the determining factors i n directing the
ov overlap value (percent c) d e s i g n towards h i g h - l i f t systems i n v o l v i n g
V aircraft velocity simpler technical solutions.
X chordwise non-dimensional
co-ordinate 1.2. Design strategy
This section provides a synthesis of the
a c , t i v i t i e s performed f o r designing the high-
1. INTRODUCTION l i f t device, based on a s e r i e s o f t h e o r e t i -
The s t u d i e s o f t h e f l a p s y s t e m p l a y a k e y c a l and e x p e r i m e n t a l i n v e s t i g a t i o n s o f seve-
r o l e f o r the a i r c r a f t design; the operatio- r a l f l a p g e o m e t r i e s . The d i f f e r e n t p h a s e s o f
n a l features o f every type o f a i r c r a f t , in- t h i s s t u d y f o l l o w t h e l o g i c a l sequence ( d e -
deed, a r e s t r o n g l y a f f e c t e d b y t h e l o w - s p e e d p i c t e d i n f i g . 2 ) o f t y p i c a l works i n aerody-
p e r f o r m a n c e and, s o , by t h e a e r o d y n a m i c c h a - nami c s .
r a c t e r i s t i c s o f t h e wing h i g h - l i f t device. A f t e r a p r e l i m i n a r y and more g e n e r a l ana-
The way a f l a p s y s t e m f o r a n a d v a n c e d am- l y s i s , t h e a t t e n t i o n was f o c u s e d o n t w o f l a p
p h i b i o u s a i r c r a f t was d e s i g n e d ( a s shown i n s y s t e m s , h a v i n g t h e same ( o r s i m i l a r ) c h o r d
t h i s paper) represents, i n general, the ac- l e n g t h : d o u b l e - s l o t t e d ( v a n e - f l a p t y p e ) and
t i v i t i e s c o n c e r n i n g t h e d e v e l o p m e n t o f a new s i n g l e - s l o t t e d f l a p d e v i c e s . Each o f t h e m
f l a p g e o m e t r y and, i n p a r t i c u l a r , t h e e f - was, a t f i r s t , t h e o r e t i c a l l y s t u d i e d i n o r -
f o r t s made i n r e s e a r c h i n g a n e f f i c i e n t h i g h - der t o a t t a i n as e f f i c i e n t geometries as
l i f t system f o r a s p e c i f i c p r o j e c t . p o s s i b l e ; then, a 2-D M i n d - T u n n e l t e s t i n g o f
A complex t e c h n i c a l process, f r o m t h e e a r - t h e t w o t y p e s o f f l a p was p e r f o r m e d . T h r o u g h
l y numerical analyses of flapped a i r f o i l t h e same a p p r o a c h ( t h e o r e t i c a l a n a l y s i s a n d
33-2

subsequent e x p e r i m e n t a t i o n ) a s i n g l e - s l o t t e d and experimental s t u d i e s o f a f a m i l y o f

h i g h - l i f t system, c h a r a c t e r i z e d b y a l a r g e r s i n g l e - s l o t t e d f l a p s ( w i t h d i f f e r e n t chord
f l a p c h o r d a n d b y a new d e s i g n e d s l o t geome- e x t e n t s ) were c a r r i e d out; i n p a r t i c u l a r ,
t r y o f t h e m a i n - b o d y , was s t u d i e d f o r o b t a i - a n a l y s e s o f s y s t e m s w i t h a 3 0 % and, m o r e r e -
n i n g an improvement o f t h e a i r f o i l h i g h - l i f t c e n t l y , a 35% f l a p c h o r d t o o k p l a c e .
characteristics.
A t t h e end o f t h e s e 2-D i n v e s t i g a t i o n s , a 2.3. Problems in designing slotted-flap
g r e a t deal o f i n f o r m a t i o n about t h e p e r f o r - systems
mance a c h i e v a b l e by means o f d i f f e r e n t t y p e s Not o n l y t h e f l a p geometry, b u t a l s o t h e
o f f l a p s y s t e m was a v a i l a b l e ; o n t h e b a s i s s l o t c s ) c o n t o u r , t h e l i p l o c a t i o n , t h e com-
o f these d a t a and a l s o c o n s i d e r i n g t h e f a c - b i n a t i o n s o f gap and o v e r l a p ( f i g . 4 ) d e t e r -
t o r s which, beyond t h e aerodynamic design, mine t h e performance o f a s i n g l e o r m u t t i -
concern t h e a c t i v i t i e s o f o t h e r groups i n - s l o t t e d h i g h - l i f t d e v i c e ; such a system, i n -
volved i n t h i s project (structures, f l i g h t deed, may b e c o n s i d e r e d a s a w h o l e o f geome-
mechanics, p r o d u c t i o n , c o n t r o l systems) t h e t r i c a l c h a r a c t e r i s t i c s , each o f them d i r e c -
h i g h - l i f t d e v i c e was c h o s e n . t l y a f f e c t s t h e aerodynamic behaviour o f t h e
F i n a l l y , W i n d - T u n n e l t e s t s o f a 1:15 s c a l e a i r f o i l and, a t t h e same t i m e , a l s o a f f e c t s
model o f t h e a i r c r a f t were u s e f u l t o v e r i f y , t h e r o l e p l a y e d b y t h e o t h e r o n e s . T h i s ma-
i n three-dimensional flow conditions, the kes t h e d e s i g n o f s l o t t e d f l a p systems a ve-
performance o f t h e s e l e c t e d f l a p and t o op- r y c o m p l i c a t e d t a s k , as c o n f i r m e d , i n deve-
timize i t s configuration. l o p i n g t h i s p r o j e c t , b y t h e l a r g e number o f
t h e o r e t i c a l s t u d i e s and e x p e r i m e n t a l t e s t s
2. GEOMETRIES were needed f o r a c h i e v i n g a s u c c e s s f u l geo-
The h i g h - l i f t d e v i c e u n d e r s t u d y i s r e l a - metry, able t o provide the h i g h - l i f t p e r f o r -
t i v e t o a i r f o i l s d e r i v e d f r o m NACA 6 - s e r i e s mance r e q u i r e d .
wing s e c t i o n s , m o d i f i e d i n t h e r e g i o n s o f I n p a r t i c u l a r , v e r y d i f f i c u l t i s t h e ana-
l e a d i n g e d g e a n d t r a i l i n g edge, w i t h p e r c e n t l y s i s o f flapped geometries (concerning, f o r
t h i c k n e s s v a r y i n g f r o m 1 8 t o 15. F l a p p e d instance, d i f f e r e n t gap/overlap values) a t
c o n f i g u r a t i o n s o f two a i r f o i l s ( p e r c e n t t h i - i n c i d e n c e approaching t h e s t a l l i n g angle,
c k n e s s o f 16 and 18) were e s p e c i a l l y i n v e - because i n t h i s case, w h i l e n o a c c u r a t e
s t i g a t e d ; f o r e a c h one, a s m e n t i o n e d b e f o r e theoretical predictions are possible, the
(sec.l.2). s i n g l e and d o u b l e - s l o t t e d (vane- l o w v a l u e s o f l o c a l R e y n o l d s number i n t h e
f l a p ) systems were s t u d i e d . A p i c t u r e o f t h e s l o t ( s ) r e g i o n ( s ) make a l s o t h e e x p e r i m e n t a l
two types o f f l a p i s g i v e n i n f i g . 3 . i n v e s t i g a t i o n a c o m p l e x one.
Evaluations concerning the h i g h - l i f t per-
formance a c h i e v a b l e through such f l a p d e v i - 2.4. Flap setting
ces, based on p r e v i o u s e x p e r i e n c e s and l i t e - The c h o i c e o f t h e f l a p d e f l e c t i o n s r e l a t i -
r a t u r e d a t a about wing s e c t i o n s , l e d t o a ve t o the operative f l i g h t conditions of the
preliminary d e f i n i t i o n o f the chord length: a i r c r a f t i s a c r u c i a l phase f o r t h e d e s i g n
v a l u e s o f 30 p e r c e n t f o r v a n e - f l a p and o f o f a h i g h - l i f t system. I n g e n e r a l , t h e f l a p
30 t o 35 p e r c e n t f o r s i n g l e - s l o t t e d f l a p s e t t i n g i s t h e r e s u l t o f a compromise b e t -
systems. ween t h e n e e d o f a c h i e v i n g l i f t p e r f o r m a n c e
i n l a n d i n g as h i g h as p o s s i b l e and t h e r e -
2.1. Why a vane-flap quirements o f h i g h aerodynamic e f f i c i e n c y
Taking i n t o account t h e need o f a c h i e v i n g d u r i n g t a k e - o f f and t h e o t h e r low-speed
l i f t performance as h i g h as p o s s i b l e , s i n c e f l i g h t operations.
t h e p r e l i m i n a r y phases o f t h i s s t u d y t h e po- The s o l u t i o n o f s u c h a p r o b l e m becomes a
t e n t i a l i t i e s of several double-slotted f l a p v e r y d i f f i c u l t o n e when c o n s i d e r i n g f l a p d e -
s y s t e m s h a v e b e e n c o n s i d e r e d . However, d u e v i c e s whose r e a r w a r d movement i s d e f i n e d b y
t o t h e manufacturing and maintenance d i f f i - a c i r c u l a r path around a s i n g l e p i v o t , be-
c u l t i e s , i n a d d i t i o n t o weight and c o s t pe- c a u s e , i n t h i s c a s e , much m o r e c o m p l i c a t e d
n a l t i e s , r e l a t i v e t o t h e l i n k a g e s y s t e m when i s t o optimize the different configurations.
a r e l a t i v e m o t i o n between t h e f l a p elements In t h e case under s t u d y , d e f l e c t i o n a n g l e s
i s r e q u i r e d , t h e i n t e r e s t has been soon d i - o f 40 t o 45 d e g r e e s f o r s i n g l e - s l o t t e d d e v i -
r e c t e d t o w a r d s d o u b l e - s l o t t e d d e v i c e s whose c e s a n d o f 45 t o 50 d e g r e e s f o r v a n e - f l a p s
f o r e element (vane) and t h e m a i n ( a f t ) - f l a p w e r e t a k e n i n t o a c c o u n t a s s u i t a b l e maximum
d e f l e c t t o g e t h e r as a u n i t around a f i x e d v a l u e s , w h i l e d e f l e c t i o n s o f 15 a n d 3 0 d e -
p i v o t . Indeed, such a system, t h a n k s t o t h e g r e e s were c o n s i d e r e d as p o t e n t i a l f o r t a k e -
b o u n d a r y - l a y e r c o n t r o l p e r f o r m e d by t h e v a - o f f and approach, r e s p e c t i v e l y . O n t h e b a s i s
ne, a l l o w s t o a c h i , e v e , e s p e c i a l l y a t l a r g e o f these values, a p r e l i m i n a r y design o f
deflections, higher l i f t values than the g e o m e t r i e s was d e v e l o p e d ; t h e s u b s e q u e n t
ones o b t a i n a b l e t h r o u g h s i n g l e - s l o t t e d f l a p s phases o f t h e t h e o r e t i c a l s t u d y and t h e e x -
and, a t t h e same t i m e , d o e s n o t i n v o l v e t h e p e r i m e n t a l i n v e s t i g a t i o n i n v o l v e d a more de-
typical structural complexity of the multi- t a i l e d analysis of f l a p deflections, in or-
body d e v i c e s . der t o evaluate the a i r f o i l performance i n a
wide range o f c o n f i g u r a t i o n s .
2.2. Why a single-slotted flap
The i n v e s t i g a t i o n o f h i g h - l i f t s y s t e m s 3. THEORETICAL 2-D STUDIES
which i n v o l v e d s i m p l e s t r u c t u r a l s o l u t i o n s 3.1. Choice of the numerical code
has always been a p r i m a r y i n t e r e s t f o r t h i s Before s t a r t i n g on t h e t h e o r e t i c a l s t u -
p r o j e c t ; t h e r e f o r e , contemporaneously w i t h d i e s , a p r e l i m i n a r y i n v e s t i g a t i o n o f methods
t h e development o f t h e v a n e - f l a p , s e v e r a l f o r t h e a n a l y s i s of m u l t i - e l e m e n t geometries
s i n g l e - s l o t t e d f l a p g e o m e t r i e s were a n a l y - i n v i s c o u s f l o w was c a r r i e d o u t . T a k i n g i n t o
sed, l o o k i n g f o r a t t a i n i n g , b y means o f a n account, i n p a r t i c u l a r , t h e need o f estima-
a p p r o p r i a t e shaping, aerodynamic c h a r a c t e r i - t i n g t h e aerodynamic performance o f f l a p p e d
s t i c s ( e s p e c i a l l y v a l u e s o f t h e maximum l i f t a i r f o i l s a t h i g h incidence, i n presence,
c o e f f i c i e n t ) c l o s e t o t h e ones o b t a i n a b l e hence, o f l a r g e s e p a r a t e d r e g i o n s , s e v e r a l
t h r o u g h more c o m p l i c a t e d systems as m u l t i - c o d e s t1,21 w e r e c o n s i d e r e d , l o o k i n g f o r t h e
body f l a p s . O n t h i s s c e n a r i o , t h e o r e t i c a l m o s t s u i t a b l e one.
 33-3

The p r o g r a m u s e d i s a m u l t i - m e t h o d s c o d e o p e r a t i v e c o n d i t i o n s . Then, a d e t a i l e d s t u d y
[ 2 1 , p r o v i d i n g d i f f e r e n t s i n g u l a r i t y methods i n a wide range o f f l a p s e t t i n g and i n c i d e n -
f o r t h e i n v i s c i d s o l u t i o n and d i f f e r e n t i n - c e a n g l e s was p e r f o r m e d .
t e g r a l methods f o r t h e b o u n d a r y - l a y e r equa- Besides t h e v a l u e s o f l i f t , d r a g and
t i o n s . C o n c e r n i n g t h e p o t e n t i a l f i e l d , sym- pitching-moment c o e f f i c i e n t s , t h e pressure
metric d i s t r i b u t i o n s o f s i n g u l a r i t i e s (sour- d i s t r i b u t i o n s were t a k e n i n t o a c c o u n t a s
ces and v o r t i c e s ) on p l a n a r p a n e l s have been v e r y e f f e c t i v e means o f a e r o d y n a m i c i n s p e c -
p r e f e r r e d , o n t h e b a s i s a l s o o f p r e v i o u s nu- tion. I n fact, from the analysis of the
merical experimentations, t o other techni- p r e s s u r e p r o f i l e s over t h e whole a i r f o i l , as
q u e s a s D o u g l a s - N e u m a n n , Hess, G r e e n , e t c . w e l l as o v e r t h e s u r f a c e s o f each body and
The s o l u t i o n o f l a m i n a r a n d t u r b u l e n t b o u n - over small regions (slots, confluence o f
d a r y l a y e r s has been achieved, r e s p e c t i v e l y , boundary l a y e r s ) , i t nas p o s s i b l e t o i n v e -
t h r o u g h T h w a i t e s and Green methods, w i t h s t i g a t e i n d e t a i l t h e d i f f e r e n t geometries.
d i r e c t / i n v e r s e techniques. Other program F u r t h e r d a t a about t h e development o f t h e
f e a t u r e s a r e t h e t r a n s p i r a t i o n method f o r boundary l a y e r (as t h e l o c a t i o n o f t h e
t h e v i s c o u s - i n v i s c i d i n t e r a c t i o n and t h e l a m i n a r - t u r b u l e n t t r a n s i t i o n and f l o w sepa-
empirical c r i t e r i o n of Hichel-Smith f o r the r a t i o n r e g i o n s ) were a l s o c o n s i d e r e d f o r a
l a m i n a r - t u r b u l e n t t r a n s i t i o n . The S q u i r e - more a c c u r a t e aerodynamic s t u d y .
Young f o r m u l a i s u s e d f o r c a l c u l a t i n g t h e
drag c o e f f i c i e n t . 3.3.1. About the prediction of
airfoil stall
3.2. Geometrical Pre-Processing The e v a l u a t i o n o f t h e maximum l i f t c o e f f i -
The p a n e l i n g i s a v e r y i m p o r t a n t p h a s e f o r c i e n t o f a i r f o i l s t h r o u g h t h e o r e t i c a l me-
the theoretical study of bodies in viscous thods, a l r e a d y a v e r y c o m p l i c a t e d m a t t e r f o r
f l o w f i e l d ; i n f a c t , a good geometry s i m u l a - a s i n g l e b o d y L31, becomes a much m o r e d i f -
t i o n u i t h an a c c u r a t e p o i n t s s p a c i n g i s an f i c u l t o n e when c o n s i d e r i n g m u l t i - b o d y s y -
e s s e n t i a l requirement f o r a c o r r e c t analysis s t e m s i n . v i s c o u s f l o w . I n many c a s e s , i n -
o f t h e boundary l a y e r , e s p e c i a l l y w i t h r e - deed, i n p r e s e n c e o f l a r g e s e p a r a t e d r e -
g a r d t o t h e l a m i n a r - t u r b u l e n t t r a n s i t i o n and gions, i n a d d i t i o n t o t h e complex f l u i d dy-
f l o w s e p a r a t i o n phenomena. M o r e o v e r , t h e namic mechanisms c o n n e c t e d w i t h t h e b o u n d a r y
p r e d i c t i o n o f aerodynamic c h a r a c t e r i s t i c s layers i n t e r a c t i o n , no accurate p r e d i c t i o n
f o r m u l t i - b o d i e s i s s t r o n g l y a f f e c t e d by the o f t h e aerodynamic c h a r a c t e r i s t i c s i s p o s s i -
panels d i s t r i b u t i o n i n t h e r e g i o n ( s ) o f con- b l e . However, a s t h e t h e o r e t i c a l m e t h o d s
f l u e n t boundary l a y e r s , where t h e computa- used a l l o w t o w e l l s i m u l a t e t h e behaviour o f
t i o n becomes m o r e c o m p l i c a t e d . a s i n g l e - b o d y almost up t o t h e s t a l l , on t h e
Therefore, considering the geometrical basis of the results relative t o the a i r f o i l
c o m p l e x i t y o f t h e bodies t o be analysed, a c l e a n c o n f i g u r a t i o n i t was p o s s i b l e t o i n d i -
p r e l i m i n a r y s t u d y c o n c e r n i n g , f o r each body, r e c t l y e s t i m a t e t h e s t a l l i n g . a n g l e o f geome-
t h e c h o i c e o f t h e p a n e l i n g t y p e a n d t h e num- t r i e s w i t h t h e f l a p d e f l e c t e d . As c o n s e q u e n -
b e r o f c o r n e r p o i n t s was n e e d e d , d e p e n d i n g c e , t h e a n a l y s i s o f s u c h s y s t e m s was s t o p p e d
on t h e r e l a t i v e p o s i t i o n s o f t h e a i ' r f o i l e- a t i n c i d e n c e f a r enough f r o m t h e c r i t i c a l
lements and t h e a n g l e o f a t t a c k . For i n s t a n - v a l u e s o f t h e a n g l e o f a t t a c k , b u t so h i g h
c e , i n s t u d y i n g f l a p s y s t e m s a t Lou d e f l e c - a s t o o b t a i n , anyway, s i g n i f i c a n t i n f o r m a -
t i o n s (when s t r o n g e r i s t h e i n t e r a c t i o n o f t i o n on t h e e f f e c t i v e n e s s o f t h e d i f f e r e n t
t h e f l o w f i e l d s g e n e r a t e d b y each body) i t f l a p configurations.
was n e c e s s a r y t o i n c r e a s e t h e number o f p a -
n e l s f o r o b t a i n i n g a s a t i s f a c t o r y convergen- 3.4. Analysis of vane-flap systems
c e o f t h e i t e r a t i v e p r o c e s s and, so, m o r e D u r i n g t h e f i r s t phase o f t h e s t u d y t h e
accurate results. a t t e n t i o n was f o c u s e d o n d i f f e r e n t c o m b i n a -
A n example o f p a n e l i n g f o r t h e main-body/ t i o n s o f vane and m a i n - f l a p 1 4 1 . I n p a r t i c u -
v a n e - f l a p s y s t e m i s shown i n f i g . 5 ; t h e law a r , i t was t a k e n a s p e c i a l c a r e o f d e s i g n i n g
f o r d i s t r i b u t i o n o f panels (cosine o f c u r v i - t h e i r r e l a t i v e p o s i t i o n , as t h e e f f e c t i v e -
l i n e a r a b s c i s s a f o r t h e main-body and t h e ness o f t h e b o u n d a r y - l a y e r c o n t r o l s t r i c t l y
a f t - f l a p , constant c u r v i l i n e a r abscissa f o r d e p e n d s o n t h e way t h e s l o t b e t w e e n t h e t u 0
t h e vane e l e m e n t ) i s p o i n t e d o u t i n t h e s l o t f l a p elements d i r e c t s t h e h i g h - e n e r g y a i r
- e n t r y r e g i o n . T y p i c a l v a l u e s o f p a n e l s num- o v e r t h e upper s u r f a c e o f t h e a f t body.
b e r f o r t h i s geometry uere: 1 2 0 / 1 4 0 , 50/60, Some g e o m e t r i e s c h a r a c t e r i z e d by d i f f e r e n t
60/80 f o r t h e main-body, vane and a f t - f l a p , r e l a t i v e dimensions o f t h e two bodies u e r e
respectively. designed, i n o r d e r t o i n v e s t i g a t e t h e s e n s i -
I n general, f o r the comparative studies of t i v i t y o f t h e a e r o d y n a m i c f l o w f i e l d and,
v a n e - f l a p and s i n g l e - s l o t t e d f l a p systems so, t h e i n f l u e n c e o f s u c h m o d i f i c a t i o n s o n
(as w e l l as o f the a i r f o i l clean configura- t h e a i r f o i l performance. Concerning t h e f l a p
t i o n ) , t h e g e o m e t r i e s u e r e s i m u l a t e d i n such d e v i c e s shown i n f i g . 6 a , s e n s i b l e b e t t e r r e -
a way a s t o r e a l i z e , e s p e c i a l l y ( f o r m u l t i - s u l t s were a c h i e v e d t h r o u g h t h e system ha-
bodies) i n the s l o t ( s ) region(s), s i m i l a r v i n g t h e l o n g e r vane and t h e s m a l l e r main-
p a n e l s d i s t r i b u t i o n s r e l a t i v e l y t o t h e num- f l a p , as denoted by t h e p r e s s u r e d i s t r i b u -
b e r and t h e l e n g t h o f panels. tions (fig.6b) obtained with the f l a p defle-
c t e d 45 degrees a t h i g h i n c i d e n c e .
3.3. Analysis method Significant variations of the h i g h - l i f t
Since t h e p r i n c i p a l aim o f t h e t h e o r e t i c a l c h a r a c t e r i s t i c s a l s o o c c u r r e d when f l a p g e o -
s t u d y was t o i n v e s t i g a t e t h e h i g h - l i f t c h a - m e t r i e s r e l a t i v e t o d i f f e r e n t vane p o s i t i o n s
r a c t e r i s t i c s , p r i m a r y emphasis uas g i v e n t o u e r e studied, as t h e elements ( f i g . 7 a ) w i t h
t h e a n a l y s i s o f geometries a t l a r g e f l a p de- a r e l a t i v e r o t a t i o n o f two degrees around
f l e c t i o n s and h i g h .angles o f a t t a c k . F i r s t t h e l e a d i n g e d g e o f t h e v a n e . The c o m p a r i s o n
o f a l l , i t was t r y e d t o g e t r e s u l t s a t i n c i - of the respective pressure p r o f i l e s (fig.7b)
dence c l o s e t o t h e expected s t a l l i n g a n g l e shows a m o r e e f f i c i e n t b o u n d a r y - l a y e r
o f a t t a c k o f t h e a i r f o i l , i n o r d e r t o eva- c o n t r o l performed, a t l a r g e d e f l e c t i o n s ,
l u a t e t h e aerodynamic behaviour o f t h e d i f - t h r o u g h t h e g e o m e t r y IIAIl; i n f a c t , i n s p i t e
f e r e n t c o n f i g u r a t i o n s a t t h e most c r i t i c a l of very similar contributions t o the t o t a l
33-4

l i f t v a l u e by t h e t w o f l a p a r r a n g e m e n t s , t h e o f h i g h - l i f t performance, e s p e c i a l l y a t l a r -
d i f f e r e n t l e v e l s o f t h e chordwise pressure ge f l a p d e f l e c t i o n s , where t h e e f f e c t i v e n e s s
d i s t r i b u t i o n over a l l t h e upper s u r f a c e o f o f t h e b o u n d a r y - l a y e r c o n t r o l r a p i d l y goes
t h e main-body i n v o l v e d i f f e r e n t l i f t p e r f o r - down.
mance o f t h e a i r f o i l ( a b o u t 3 % ) . One o f t h e s i n g l e - s l o t t e d f l a p s s t u d i e d
The s t u d y o f t h e v a n e - f l a p r e l a t i v e t o a n d t h e v a n e - f l a p , b o t h d e f l e c t e d 45 d e g r e e s
several gap/overlap combinations ( r e l a t e d t o a r o u n d t h e same p i v o t , a r e c o m p a r e d i n
different positions of the f l a p rotation f i g . 1 0 a . The r e l a t i v e p r e s s u r e d i s t r i b u t i o n s
c e n t r e ) has a l s o p l a y e d a key r o l e i n d e f i - (fig.lOb) c l e a r l y denote t h e loss o f e f f e c -
n i n g t h e g e o m e t r y o f t h i s f l a p s y s t e m [SI. tiveness associated t o the single-body f l a p
Among t h e c o n f i g u r a t i o n s a n a l y s e d , t w o g e o - system; t h e s e n s i b l e l o w e r e x p a n s i o n o v e r
m e t r i e s , a l t h o u g h c h a r a c t e r i z e d by q u i t e t h e upper s i d e o f t h e a i r f o i l p o i n t s o u t t h e
d i f f e r e n t overlap values (fig.8a), gave t h e f u n c t i o n o f t h e vane element o f t h e d o u b l e -
same r e s u l t s i n t e r m s o f g l o b a l l i f t p e r f o r - s l o t t e d d e v i c e w h i c h e n e r g i z e s t h e f l o w and,
mance. T h e y a r e r e p r e s e n t a t i v e o f f l a p p o s i - so, d e l a y s s e p a r a t i o n .
tions involving substantial differences
a b o u t t h e s l o t - e n d g e o m e t r y and, so, a b o u t 3.5.2 Thirty-five per cent flap
t h e mechanism f o r t h e b o u n d a r y - l a y e r con- The s t u d y o f s i n g l e - s l o t t e d d e v i c e s w i t h a
t r o l . From t h e c o m p a r i s o n o f t h e r e l a t i v e l a r g e r f l a p c h o r d was c a r r i e d o u t t r y i n g t o
pressure p r o f i l e s (fig.6b). in particular improve t h e l i f t performance e s t i m a t e d f o r
two e f f e c t s can be observed: w h i l e t h e con- s i m i l a r geometries having a smaller chord
verging main s l o t r e l a t i v e t o t h e f o r u a r d l e n g t h and, so, t r y i n g t o a p p r o a c h o r , i f
f l a p p o s i t i o n allows t o better energize the possible, t o a t t a i n the theoretical results
f l o w over t h e upper s u r f a c e o f t h e main bo- r e l a t i v e t o t h e v a n e - f l a p system w i t h a
dy, t h e m o r e e x t e n d e d f l a p i n v o l v e s a l a r - smaller chord (0.30~).
g e r l i f t c o n t r i b u t i o n b y t h e vane, d u e t o a S i n c e t h e b a s i c g e o m e t r i c a l c o n d i t i o n was
stronger fluid acceleration in the region of t o n o t change e i t h e r t h e c u r v a t u r e o r t h e
c o n f l u e n t boundary l a y e r s . S i t u a t i o n s l i k e a i r f o i l t h i c k n e s s , t h e f l a p g e o m e t r i e s ana-
t h i s , d e p e n d i n g o n t h e c o m p l e x phenomena l y s e d resemble, f o r t h e r e a r p a r t , t h e p r o -
connected w i t h m u l t i - b o d y systems i n v i s c o u s f i l e s o f t h e v a n e - f l a p and s i n g l e - s l o t t e d
flow, uere f r e q u e n t l y observed d u r i n g the systems p r e v i o u s l y defined. Obviously, ac-
study concerning a wide g r i d o f p i v o t p o s i - c o r d i n g t o t h e problems ( d i s c u s s e d on
t i o n s . I t must b e s a i d , however, t h a t t h e sec.2.3) concerning the design o f s l o t t e d -
r e l a t i v e behaviour o f generic f l a p configu- f l a p d e v i c e s , t h e p r o j e c t o f t h e new geome-
r a t i o n s may s e n s i b l y c h a n g e , e s p e c i a l l y a t t r y i n v o l v e d a s e r i e s o f s t u d i e s about t h e
large deflections, by increasing the angle g e o m e t r i c a l elements which concur t o d e f i n e
o f a t t a c k ; so, m o r e i n t e r n a l f l a p p o s i t i o n s , a h i g h - l i f t system: t h e s l o t - e n t r y p r o f i l e ,
e v e n t h o u g h m o r e e f f e c t i v e a t Lou a n d medium t h e l i p e x t e n t and, a b o v e a l l , t h e g a p a n d
incidence, might cause a n t i c i p a t i o n o f a i r - o v e r l a p v a l u e s . I n t h i s case, t h e s l o t shape
f o i l s t a l l , due t o s t r o n g e r a d v e r s e p r e s s u r e o f t h e m a i n - b o d y was s e n s i b l y m o d i f i e d , d u e
gradients. t o t h e increment o f f l a p chord, w h i l e t h e
The l i f t c u r v e s c o n c e r n i n g t h e a i r f o i l e - l i p was moved t o w a r d s t h e n o r m a l t r a i l i n g -
q u i p p e d w i t h one o f t h e most e f f i c i e n t vane- edge p o s i t i o n f o r t a k i n g g r e a t e r b e n e f i t s o f
f l a p s y s t e m s a r e shown i n f i g . 9 a ; t o be no- the extended f l a p c o n f i g u r a t i o n .
ted the f l a p effectiveness i n progressively Concerning the p i v o t , a c a r e f u l a n a l y s i s
i m p r o v i n g t h e a i r f o i l l i f t p e r f o r m a n c e . Fur- was n e e d e d , i n o r d e r t o a t t a i n , f o r e a c h d e -
thermore, c o n s i d e r i n g a l s o t h e p i t c h i n g - f l e c t i o n connected w i t h t h e l a r g e backward
moment c o e f f i c i e n t ( f i g . 9 b ) . one c a n o b s e r v e f l a p movement, a n e f f i c i e n t b o u n d a r y - l a y e r
that the pressure centre (located, approxi- c o n t r o l through an a p p r o p r i a t e (converging)
mately, a t the quarter chord p o i n t w i t h f l a p s l o t . As a n e x a m p l e o f t h e s e n s i b l e v a r i a -
r e t r a c t e d ) s h i f t s rearward (about 0 . 0 2 ~ ) by t i o n s o f performance by changing the p i v o t ,
i n c r e a s i n g up t o 1 5 d e g r e e s t h e f l a p d e f l e c - pressure p r o f i l e s (fig.11) concerning d i f f e -
t i o n , w h i l e i t s h i f t s forward (about 0 . 0 3 ~ ) r e n t o v e r l a p v a l u e s i n d i c a t e t h a t more i n -
by v a r y i n g t h e d e f l e c t i o n f r o m 15 t o 45 d e - t e r n a l f l a p p o s i t i o n s provide, a t l a r g e de-
grees. flections, better results.

3.5. Analysis of single-slotted a) Comparison w i t h t h e v a n e - f l a p

flap systems I n fig.12, showing t h e f l a p system under
Two p h a s e s h a v e c h a r a c t e r i z e d t h e d e s i g n s t u d y and t h e v a n e - f l a p d e v i c e , t h e change
o f t h e s i n g l e - s l o t t e d d e v i c e s : t h e f i r s t one made i n t h e r e a r p a r t o f t h e m a i n - b o d y , t h e
concerned t h e p r o j e c t o f f l a p s w i t h chord variation o f the a i r f o i l curvature at high
v a r y i n g f r o m 3 0 t o 32 p e r c e n t o f a i r f o i l d e f l e c t i o n , t h e d i f f e r e n t c h o r d e x t e n t s may
chord and c o n f i g u r a t i o n s s i m i l a r t o t h e be observed.
v a n e - f l a p u n d e r s t u d y ; t h e s e c o n d o n e [61 O f g r e a t i n t e r e s t i s t h e r e l a t i v e beha-
was r e l a t i v e t o a s i n g l e - s l o t t e d s y s t e m w i t h v i o u r o f t h e two f l a p s a t d i f f e r e n t angles
more s i g n i f i c a n t m o d i f i c a t i o n s o f g e o m e t r y o f d e f l e c t i o n (see fig.13). U h i l e these sy-
and an i n c r e a s e d ( 0 . 3 5 ~ ) f l a p chord. stems i n v o l v e q u i t e s i m i l a r l i f t v a l u e s a t
h i g h d e f l e c t i o n s when, d u e t o l a r g e s e p a r a -
3.5.1. ThirtyIThirty-two per cent flaps t e d regions, the boundary-layer c o n t r o l
S e v e r a l g e o m e t r i e s a n d , f o r e a c h one, s e - t h r o u g h t h e vane p l a y s a k e y r o l e , a s e n s i -
veral configurations (relative t o different b l e i n c r e m e n t o f p e r f o r m a n c e by means o f t h e
g a p / o v e r l a p v a l u e s ) w e r e s t u d i e d . The f l a p s i n g l e - s l o t t e d g e o m e t r y was a t t a i n e d a t
p r o f i l e s were d e s i g n e d l i k e w i s e t h e vane- s m a l l a n d medium d e f l e c t i o n s , u h e n t h e a i r -
f l a p s y s t e m p r e v i o u s l y analysed; however, f o i l chord extent ( w i t h the f l a p deflected),
some l i t t l e m o d i f i c a t i o n s o f c u r v a t u r e w e r e t h e shape o f t h e s l o t , t h e l o c a l c u r v a t u r e
i n t r o d u c e d . The r e s u l t s o b t a i n e d , c o m p a r e d and, i n g e n e r a l , t h e e l e m e n t s w h i c h a l l t o -
u i t h t h e ones c o n c e r n i n g e q u i v a l e n t c o n f i - g e t h e r make a m u l t i - b o d i e s g e o m e t r y become
g u r a t i o n s ( w i t h t h e same g a p / o v e r l a p v a l u e s ) m o r e s i g n i f i c a n t . The p r e s s u r e d i s t r i b u t i o n s
o f t h e v a n e - f l a p , showed a s e n s i b l e d e c r e a s e (fig.14) r e f l e c t t h i s situation.
 33-5

The p o l a r a n d p i t c h i n g - m o m e n t c u r v e s d e n o - q u a l c h o r d ) s i n g l e - s l o t t e d s y s t e m s was v e r i -
t e s i m i l a r t r e n d s f o r t h e two f l a p s up t o f i e d ; afterwards, as p o t e n t i a l f o r a l t e r n a -
30 degrees o f d e f l e c t i o n : i n f a c t , n o s i g n i - t i v e s o l u t i o n , t h e extended (35% o f a i r f o i l
1 ficant variations of efficiency exist, while c h o r d ) s i n g l e - s l o t t e d f l a p was t e s t e d .
t h e d i f f e r e n c e s i n terms o f pitching-moment
coefficients are proportional t o the Lift 4.1. Investigation planning

'
values. On t h e contrary, a r a t h e r d i f f e r e n t The f i r s t p h a s e o f t h e e x p e r i m e n t a l s t u -
behaviour o f t h e s i n g l e - s l o t t e d f l a p can be d i e s concerned t h e i n v e s t i g a t i o n o f a i r f o i l s
observed a t l a r g e d e f l e c t i o n s (45 degrees), equipped w i t h t h e v a n e - f l a p and f a m i l i e s o f
when t h e l o c a t i o n o f t h e p r e s s u r e c e n t r e s i n g l e - s l o t t e d devices w i t h 30-32 per cent
s h i f t s f o r w a r d ( a b o u t O.O2*c) r e l a t i v e l y t o o f c h o r d e x t e n t . Through q u a s i - 2 D t e s t s i n
t h e v a n e - f l a p system. t h e U i n d Tunnels o f Naples U n i v e r s i t y and o f
Turin U n i v e r s i t y , t h e comparative s t u d i e s
b) Comparison w i t h s m a l l e r s i n g l e - s l o t t e d were performed, w h i l e a n e x p e r i m e n t a l a n a l y -
flaps s i s o f t h e v a n e - f l a p s y s t e m i n s t r i c t l y 2-D
I n t h e whole range o f f l a p d e f l e c t i o n s a c o n d i t i o n s took p l a c e i n t h e U i n d Tunnel o f
v e r y l a r g e improvement o f l i f t performance S t u t t g a r t U n i v e r s i t y [71.
was a t t a i n e d t h r o u g h t h e new s i n g l e - s l o t t e d The s e c o n d p a r t o f t h e e x p e r i m e n t a t i o n ,
f l a p , whose g e o m e t r y i s c o m p a r e d ( f i g . 1 5 a ) h e l d i n t h e Wind Tunnel o f N a p l e s U n i v e r s i -
w i t h a 3 0 % s i n g l e - b o d y f l a p . As p o i n t e d o u t t y ( 1 9 9 1 ) , was d e v o t e d t o t h e s i n g l e - s l o t t e d
by t h e p r e s s u r e d i s t r i b u t i o n s ( f i g . 1 5 b 1 , t h e g e o m e t r y w i t h 35% o f a i r f o i l c h o r d .
s t r o n g e r expansion over t h e upper s u r f a c e o f
t h e main-body i s r e s p o n s i b l e f o r t h e h i g h 4.2. Investigation methods
increase o f t h e t o t a l l i f t values achieved The i n d i c a t i o n f r o m t h e t h e o r e t i c a l s t u d y
through t h e extended system, w h i l e t h e f l a p about t h e most e f f i c i e n t c o m b i n a t i o n s o f gap
contributions are quite similar. a n d o v e r l a p , f o r e a c h f l a p g e o m e t r y , was u-
s e f u l t o p i l o t the experimental a c t i v i t i e s .
3.6. Concluding remarks I n f a c t , t h r o u g h U i n d - T u n n e l t e s t s , i n many
This series of theoretical studies provi- cases, t h e a i r f o i l aerodynamic c h a r a c t e r i -
ded a good p r e d i c t i o n o f t h e aerodynamic s t i c s r e l a t i v e l y t o some p a r t i c u l a r p i v o t
, performance o f a i r f o i l s equipped w i t h s i n g l e p o s i t i o n s were a n a l y s e d , t r y i n g t o o p t i m i z e
o r double-slotted flaps. I n fact, although s u c h c o n f i g u r a t i o n s and, t h u s , a v o i d i n g t e -
t h e a n a l y s i s o f multi-element geometries i s t i n g o t h e r f e a s i b l e f l a p arrangements.
a v e r y d i f f i c u l t one, t h e c o m p u t a t i o n me- For each h i g h - l i f t system and each c o n f i -
thods allowed t o evaluate the c h a r a c t e r i - g u r a t i o n , t h e a i r f o i l s were e x p e r i m e n t e d a t
s t i c s of h i g h - l i f t devices also a t condi- d i f f e r e n t f l a p deflections, i n order t o at-
t i o n s ( h i g h i n c i d e n c e and l a r g e f l a p d e f l e c - t a i n a c l e a r p i c t u r e o f performance f o r a l l
t i o n s ) w h e r e m o r e s i g n i f i c a n t and, a t t h e t h e o p e r a t i v e f l a p range.
same t i m e , m o r e c o m p l e x t h e s t u d y o f s u c h
, s y s t e m s becomes. 4.3. Preliminary tests of vane-flap and
Concerning, i n p a r t i c u l a r , t h e i n v e s t i g a - single-slotted systems
t i o n o f gap and o v e r l a p , a s p e c i a l t o p i c f o r The e a r l y W i n d - T u n n e l t e s t i n g c o n s i s t e d i n
the f l a p design, r a t h e r d e f i n i t e i n d i c a - d e t e r m i n i n g t h e most p o w e r f u l c o n f i g u r a -
t i o n s , depending on t y p e o f geometry ( s i n g l e t i o n ( s ) f o r t h e v a n e - f l a p geometry; s e v e r a l
o r d o u b l e - s l o t t e d ) , were o b t a i n e d . A c c o r d i n g arrangements o f t h e vane element were expe-
l
t o t h e t h e o r e t i c a l a n a l y s i s , indeed, w h i l e rimented, as w e l l as t h e p i v o t p o s i t i o n s
more e x t e r n a l p o s i t i o n s o f t h e v a n e - f l a p which, on t h e b a s i s o f t h e t h e o r e t i c a l s t u -
, t u r n e d o u t t o be b e t t e r , s i n g l e - s l o t t e d geo- dy, m i g h t have a l l o w e d t h e b e s t a i r f o i l p e r -
m e t r i e s needed, i n g e n e r a l , a more d e f i n i t e f ormance.
converging s l o t (connected w i t h l a r g e r over- From t h e t e s t s c o n c e r n i n g d i f f e r e n t r e l a -
l a p values) f o r a c h i e v i n g an e f f e c t i v e t i v e p o s i t i o n s o f t h e two f l a p elements,
1 bondary- l a y e r c o n t r o l .
From t h e w h o l e s p e c t r u m o f t h e i n v e s t i g a -
v a r i a t i o n s o f l i f t v a l u e s ( a b o u t 4%, a t
h i g h d e f l e c t i o n s ) q u i t e s i m i l a r t o t h e ones
t i o n s p e r f o r m e d , t h e v a n e - f l a p a p p e a r e d as a g i v e n by t h e t h e o r e t i c a l a n a l y s i s ( s e c . 3 . 4 )
v e r y e f f i c i e n t geometry f o r a t t a i n i n g t h e were found.
s p e c i a l requirements o f a f l a p system f o r Also the results obtained with regard t o
i a m p h i b i o u s a i r c r a f t ; i n f a c t , i t was n e c e s - t h e aerodynamic behaviour o f t h e a i r f o i l by
sary t o increase the chord o f s i n g l e - s l o t t e d varying the gap/overlap combinations c o n f i r -
f l a p s up t o 3 5 % f o r a c h i e v i n g ( a t l a r g e d e - med, f o r t h e m o s t p a r t , t h e t h e o r e t i c a l p r e -
~
f l e c t i o n s ) and improving ( a t s m a l l d e f l e c - d i c t i o n s . The c u r v e s o f f i g . 1 6 , f o r instan-
t i o n s ) the performance o f a i r f o i l s equipped ce, i n d i c a t e t h e v a r i a t i o n o f h i g h - l i f t p e r -
w i t h t h e v a n e - f l a p ( 0 . 3 0 ~ ) . I t was j u s t t h e formance a s s o c i a t e d t o d i f f e r e n t o v e r l a p va-
extended single-body f l a p a p o s s i b l e a l t e r - l u e s ; i n t h i s case, t h e more e x t e r n a l f l a p
n a t i v e t o t h e v a n e - f l a p system, as c o n f i r m e d position, although less e f f i c i e n t up t o high
~
by t h e s u b s e q u e n t e x p e r i m e n t a l s t u d i e s . incidence, a l l o w s t o s e n s i b l y improve t h e
Nevertheless, an i n t e n s i v e Uind-Tunnel ex- a i r f o i l aerodynamic c h a r a c t e r i s t i c s c l o s e t o
perimentation of a l l the h i g h - l i f t devices the s t a l l . This situation, typical of the
s t u d i e d was p l a n n e d , l o o k i n g f o r t h e i n f o r - vane-flap behaviour a t large deflections,
1 m a t i o n needed f o r c o n f i r m i n g and b e t t e r de-
f i n i n g the theoretical results.
has been d e c i s i v e f o r t h e c o n c l u s i v e d e f i n i -
t i o n o f t h e f l a p system.
I D u r i n g t h e same c a m p a i g n , some s i n g l e -
4. EXPERIMENTAL 2-D STUDIES s l o t t e d f l a p s were t e s t e d , n o t o n l y i n o r d e r
Somewhat d e f i n i t e r e s u l t s c o n c e r n i n g t h e t o determine other p o t e n t i a l wing h i g h - l i f t
most p o w e r f u l h i g h - l i f t systems had been a- d e v i c e s , but a l s o f o r e v a l u a t i n g t h e l o s s o f
c h i e v e d a t t h e end o f t h e t h e o r e t i c a l s t u - l i f t performance i n comparison w i t h the
d i e s . T h e r e f o r e , t h e e x p e r i m e n t a t i o n was a i - v a n e - f l a p system. Several l i f t curves r e l a -
med a t a n a l y s i n g e s p e c i a l l y t h e v a n e - f l a p t i v e t o t h e v a n e - f l a p and t o one o f t h e most
device, a f t e r i t s estimated higher p e r f o r - e f f i c i e n t s i n g l e - s l o t t e d f l a p s experimented
mance i n c o m p a r i s o n w i t h e q u i v a l e n t ( o f e - a r e shown i n f i g . 1 7 . This diagram i s repre-
33-6

sentative o f the d i f f e r e n c e of l i f t values and, t h e n , t h e e x p e r i m e n t a l i n v e s t i g a t i o n

achievable, a t l a r g e and s m a l l d e f l e c t i o n s , were aimed a t p e r f o r m i n g such a c o m p a r a t i v e
through s i n g l e o r double-slotted f l a p s (with a n a l y s i s ; t h e geometries and t h e r e l a t i v e
t h e same c h o r d ) , a s d e r i v e d , i n a c c o r d a n c e f l a p arrangements, as d e r i v e d f r o m t h e t h e o -
with the indication of the theoretical stu- r e t i c a l d e s i g n , were e x p e r i m e n t e d t h r o u g h a
dy, f r o m t h i s e x p e r i m e n t a t i o n . s e r i e s o f Wind-Tunnel t e s t s i n o r d e r t o ve-
r i f y a l l t h e t h e o r e t i c a l i n d i c a t i o n s and
4.4. Experimentation of the vane-flap predictions.
system (Part 2) According t o t h i s ltqualitativell method o f
Two a i r f o i l s w i t h a maximum p e r c e n t t h i - comparison between t h e t h e o r e t i c a l and expe-
c k n e s s o f 16 a n d 18, e q u i p p e d w i t h t h e v a n e - r i m e n t a l r e s u l t s , t h e Uind-Tunnel experimen-
f l a p d e v i c e , were t e s t e d i n t h e U i n d Tunnel t a t i o n has c o n f i r m e d f o r t h e most p a r t , as
o f S t u t t g a r t U n i v e r s i t y . The a i m o f t h i s pointed out i n the previous sections, the
c a m p a i g n was t o i n v e s t i g a t e t h e v a n e - f l a p theoretical conclusions r e f e r r i n g t o the
system i n s t r i c t l y 2-0 c o n d i t i o n s , a f t e r i t s d i f f e r e n t types o f f l a p t.ested, d i f f e r e n t
geometry had been d e f i n e d as r e s u l t o f theo- f l a p and s l o t s p r o f i l e s , d i f f e r e n t gap/
r e t i c a l s t u d i e s and p r e v i o u s experimenta- o v e r l a p v a l u e s and so on.
tions. A n a b s o l u t e c o m p a r i s o n , made p o s s i b l e b y
Several t e s t s , r e f e r r i n g t o d i f f e r e n t ex- t h e s t r i c t l y 2-D c o n d i t i o n s o f t h e Wind-
perimental conditions (techniques o f laminar Tunnel t e s t i n g ( s e e sec.4.4), i s presented
- t u r b u l e n t t r a n s i t i o n , R e y n o l d s number) were f o r t h e v a n e - f l a p system i n fig.20, where
p e r f o r m e d f o r b o t h a i r f o i l s . The m o s t i n t e - some e x p e r i m e n t a l r e s u l t s , o b t a i n e d f o r d i f -
r e s t i n g r e s u l t s a r e p r e s e n t e d i n f i g . 1 8 . As f e r e n t f l a p d e f l e c t i o n s , a r e compared w i t h
shown by t h e l i f t c u r v e s , t h e b e h a v i o u r o f t h e ones d e r i v e d f r o m t h e t h e o r e t i c a l a n a l y -
t h e two a i r f o i l s i s s i m i l a r , b o t h a t s m a l l s i s . F i r s t o f a l l , t h e v e r y good accordance
and l a r g e f l a p d e f l e c t i o n s , a l s o b y v a r y i n g between t h e t h e o r e t i c a l and experimental
t h e R e y n o l d s n u m b e r . The i n f o r m a t i o n o b t a i - l i f t curves concerning the a i r f o i l clean
ned, w i t h s p e c i a l r e g a r d t o t h e maximum l i f t c o n f i g u r a t i o n may b e o b s e r v e d ; m o r e o v e r , i n
c o e f f i c i e n t , were u s e f u l t o p r e d i c t t h e a i r - s p i t e o f an o v e r - e s t i m a t e o f a i r f o i l p e r f o r -
c r a f t low-speed performance achievable, i n mance w i t h t h e f l a p d e f l e c t e d , t h e q u i t e s i -
f u l l s c a l e , t h r o u g h t h i s f l a p system. m i l a r values of the curves gradient i n the
whole range o f d e f l e c t i o n s c o n f i r m e d t h e va-
4.5. Experimentation of the extended l i d i t y o f t h e t h e o r e t i c a l p r e d i c t i o n s . Re-
single-slotted flap. markable i s a l s o t h e accordance o f t h e theo-
The l a s t c a m p a i g n o f 2 - 0 U i n d - T u n n e l t e s t s r e t i c a l and experimental values o f p i t c h i n g -
concerned t h e experimental s t u d y o f t h e moment c o e f f i c i e n t : n o t o n l y c o n s i d e r i n g t h e
s i n g l e - s l o t t e d device with the increased a b s o l u t e data, b u t a l s o because o f t h e s i m i -
(35%) f l a p chord. l a r i t y o f the curves slope, which confirmed
T h i s s y s t e m has g i v e n v e r y i n t e r e s t i n g r e - t h e e s t i m a t e o f p o s i t i o n o f t h e aerodynamic
sults, confirming i t s greater potentiality, centre at different f l a p deflections.
a l r e a d y known f r o m t h e t h e o r e t i c a l a n a l y s i s ,
i n c o m p a r i s o n w i t h t h e v a n e - f l a p . As c l e a r l y 5.2. Choice of the flap system
denote t h e l i f t curves ( f i g . l 9 ) , the perfor- The t h e o r e t i c a l a n d e x p e r i m e n t a l 2 - 0 s t u -
mance a c h i e v e d a t s m a l l a n d medium d e f l e c - d i e s h a v e g i v e n t h r e e m a i n a n s w e r s . The
t i o n s i s s e n s i b l y b e t t e r t h a n t h e one o b t a i - f i r s t one i s t h e v a n e - f l a p , w h i c h c a n b e
ned through t h e d o u b l e - s l o t t e d system, w h i l e considered as t h e main o b j e c t o f t h e whole
with the f u l l deflected flap i t i s rather p r o j e c t : t h a t i s a f l a p system, w i t h o u t spe-
d i f f i c u l t t o analyse the behaviour o f the c i a l structural complications, able t o s a t i -
a i r f o i l equipped w i t h t h e two h i g h - l i f t de- s f y t h e a e r o d y n a m i c r e q u i r e m e n t s . The s e c o n d
vices, i n s p i t e o f d i f f e r e n t values o f the one i s t h e g a i n o f l i f t p e r f o r m a n c e a t t a i n a -
maximum l i f t c o e f f i c i e n t . Anyway, what a p - b l e through double-slotted devices w i t h re-
p e a r s w i t h e v i d e n c e i s t h e improvement o f ference t o s i n g l e - s l o t t e d f l a p s having the
the a i r f o i l l i f t performance obtainable same c h o r d . The t h i r d o n e i s t h e s i n g l e - b o d y
through t h i s s i n g l e - s l o t t e d f l a p , conside- f l a p w i t h t h e i n c r e a s e d chord: b o r n as a l -
r i n g t h e whole o p e r a t i v e range o f d e f l e c - ternative solution t o the vane-flap, t h i s
t i o n s . F u r t h e r m o r e , t h e new s y s t e m , c o m p a r e d s y s t e m h a s shown, i n d e e d , g r e a t p o t e n t i a l i -
t o t h e v a n e - f l a p geometry, does n o t i n v o l v e ties.
s i g n i f i c a n t p e n a l t i e s o f drag and p i t c h i n g - On t h e b a s i s o f s u c h r e s u l t s , r e j e c t e d t h e
moment c o e f f i c i e n t s . hypothesis o f s i n g l e - s l o t t e d f l a p s with
s m a l l c h o r d (up t o 30%), f r o m t h e a e r o d y n a -
5 . CONCLUSION OF TWO-DIMENSIONAL STUDIES mic p o i n t o f view no doubt e x i s t e d about t h e
The e n d o f t h e U i n d - T u n n e l 2 - 0 t e s t s h a s advantages connected with t h e extended
represented a m i l e s t o n e f o r t h e f l a p system s i n g l e - s l o t t e d system i n comparison w i t h t h e
d e s i g n , as t h e f o l l o w i n g phases c o n c e r n a v a n e - f l a p . I n f a c t , t h e remarkable improve-
s e r i e s o f c o r r e l a t i o n a c t i v i t i e s between t h e ment o f l i f t p e r f o r m a n c e i n a l m o s t t h e w h o l e
r e s u l t s o f t h e e x p e r i m e n t a l and t h e o r e t i c a l range o f f l a p d e f l e c t i o n s , t h e o r e t i c a l l y es-
s t u d i e s and, t h e n , t h e f i n a l c h o i c e o f t h e t i m a t e d a n d a l s o c o n f i r m e d by t h e U i n d -
h i g h - l i f t device. Tunnel 2-0 experimentation, c l e a r l y i n d i c a -
t e d such a h i g h - l i f t d e v i c e a s t h e most e f -
5.1. Comparative analysis of theoretical f i c i e n t , a l s o c o n s i d e r i n g p o s s i b l e d r a g and
and experimental results pitching-moment p e n a l t i e s i n three-
The d e s i g n o f t h e f l a p s y s t e m h a s b e e n b a - dimensional conditions.
s e d , a s i n g e n e r a l f o r a new g e o m e t r y d e - B u t , u h a t made v e r y d i f f i c u l t t h e c h o i c e
s i g n , on t h e s t u d y o f d i f f e r e n t c o n f i g u r a - o f t h e m o s t s u i t a b l e h i g h - l i f t s y s t e m was
tions, t r y i n g t o achieve, through the analy- t h e i n f l u e n c e o f o t h e r d e s i g n areas which,
s i s of t h e r e l a t i v e performance, t h e f l a p i n a d d i t i o n t o t h e aerodynamic c r i t e r i a , had
p r o f i l e , t h e s l o t shape, t h e p i v o t p o s i t i o n t o be considered. I n f a c t , problems connec-
w h i c h , a l l t o g e t h e r , make a n e f f e c t i v e h i g h - t e d with the structure, weight, c o n t r o l sy-
l i f t d e v i c e . The t h e o r e t i c a l s t u d y f i r s t stems, p r o d u c t i o n and m a i n t e n a n c e c o s t s ,
 33-7

etc., which had a l r e a d y s e n s i b l y a f f e c t e d , connected w i t h t h e aerodynamic s t u d y o f

a t the beginning, t h i s p r o j e c t i n determi- m u l t i - e l e m e n t g e o m e t r i e s . Only a p p l i c a t i o n s
n i n g f e a s i b l e types o f flap, a l s o played a t o s p e c i f i c p r o j e c t s , indeed, can p r o v i d e ,
key r o l e i n c o n c l u s i o n o f t h e t u o - i n g e n e r a l , t h e d a t a needed f o r d e s i g n i n g a
d i m e n s i o n a l s t u d i e s , when t h e f i n a l d e c i s i o n f l a p s y s t e m ; so, o n l y t h e o r e t i c a l s t u d i e s
a b o u t t h e f l a p s y s t e m was t a k e n . and e x p e r i m e n t a l i n v e s t i g a t i o n s , d e v o t e d t o
So, s t r u c t u r a l a n d m a n u f a c t u r i n g c o m p l e x i - a n a l y s e a g r e a t number o f f l a p c o n f i g u r a -
t i e s , weight, problems r e l a t i v e t o t h e a r - t i o n s , have a l l o w e d t o achieve t h e f i n a l
rangement o f s e v e r a l c o n t r o l d e v i c e s , were g o a l o f t h i s work.
a l l f a c t o r s , depending on t h e l a r g e r f l a p The f l a p s y s t e m c h o s e n i s t h e a n s w e r t o
c h o r d and a more complex l i n k a g e s y s t e m (due d i f f e r e n t needs: s t r u c t u r a l and m a n u f a c t u -
t o t h e more downward p i v o t l o c a t i o n ) u h i c h r i n g s o l u t i o n s i n accordance w i t h t h e gene-
p e n a l i z e d t h e e x t e n d e d s i n g l e - b o d y f l a p . The r a l design c r i t e r i a f o r t h i s type of airpla-
d e s i g n o f s u c h a s y s t e m , t h u s , was s t o p p e d ne, a i m e d a t r e d u c i n g t h e c o s t s o f p r o d u c -
a t t h i s point, without performing a 3-D tes- t i o n ; h i g h aerodynamic e f f e c t i v e n e s s i n t h e
ting. whole range o f o p e r a t i o n a l low-speed c o n d i -
As c o n s e q u e n c e , t h e v a n e - f l a p w h i c h , s i n c e t i o n s , i n accordance w i t h t h e requirements
t h e b e g i n n i n g o f t h i s study, had been c o n s i - o f a n 8 8 a d v a n c e d 1 1a i r c r a f t .
d e r e d as a v e r y p o w e r f u l s y s t e m f o r a c h i e -
v i n g t h e l i f t p e r f o r m a n c e needed and which
through t h e o r e t i c a l and experimental i n v e - Acknowledgements
s t i g a t i o n s c o n f i r m e d t h i s p o t e n t i a l i t y , was
c h o s e n a s h i g h - l i f t d e v i c e . T h i s was t h e The a u t h o r s u i s h t o e x p r e s s t h e i r own g r a -
f l a p u h i c h e q u i p p e d t h e 1:15 s c a l e m o d e l o f t i t u d e t o c o l l e a g u e s o f t h e Aerodynamics de-
t h e a i r c r a f t , d u r i n g t h e 3 - 0 Wind-Tunnel p a r t m e n t f o r t h e i r s u g g e s t i o n s and coopera-
t e s t s campaigns. t i o n . S p e c i a l t h a n k s t o Mr.M.Manco f o r h i s
c o n t r i b u t i o n t o t h e l a y - o u t o f t h i s paper,
6. EXPERIMENTAL 3-D STUDIES p a r t i c u l a r l y w i t h r e g a r d t o t h e d i a g r a m s and
The a i r c r a f t p e r f o r m a n c e a c h i e v a b l e , i n a the other pictures.
wide f i e l d o f aerodynamic c o n d i t i o n s ,
t h r o u g h t h e f l a p system s e l e c t e d have been
analysed on the b a s i s o f Uind-Tunnel t e s t s REFERENCES
[ 8 1 p e r f o r m e d o n a 1:15 p o w e r e d m o d e l o f t h e
a i r c r a f t (Turin U n i v e r s i t y ) . I n f a c t , expe- 1. Cebeci, T., Jau, J., V i t i e l l o , D.:
rimentations concerning d i f f e r e n t values o f A n I n t e r a c t i v e Boundary-Layer Approach
t h r u s t c o e f f i c i e n t , i n f r e e - a i r and ground- t o M u l t i - e l e m e n t A i r f o i l s a t High L i f t .
effect, both with f u l l operating propulsive A I A A 3 0 t h Aerospace Sciences Meeting,
system and i n one e n g i n e - o u t c o n d i t i o n , were Reno, Nevada, 1 9 9 2 .
c a r r i e d o u t f o r t h e whole f l a p s e t t i n g , a l s o
b y v a r y i n g t h e d e f l e c t i o n o f t h e o t h e r mova- 2. d e N i c o l a , C., C o i r o , D., L o s i t o , V.:
b l e surfaces (elevator, rudder, ailerons, A n e f f i c i e n t M u l t i - M e t h o d s C o m p u t e r Code
spoilers). f o r t h e P r e d i c t i o n o f t h e I n v i s c i d and
Before these experiments, a f l a p t e s t i n g Viscous Flow over Multi-Component A i r -
c o n c e r n i n g d i f f e r e n t p i v o t p o s i t i o n s was foils. - A I A A , 1 s t N a t i o n a l F l u i d Dyna-
p e r f o r m e d , i n t h e same W i n d T u n n e l a n d o n m i c C o n g r e s s , C i n c i n n a t i , O h i o , 1988.
t h e same a i r c r a f t m o d e l . The o b j e c t o f t h i s
p r e l i m i n a r y i n v e s t i g a t i o n was t o a n a l y s e , i n 3. C e b e c i , T., Jau, J., V i t i e l l o , D.:
three-dimensional conditions, the behaviour P r e d i c t i o n o f P o s t - S t a l l Flows on A i r -
o f t h e h i g h - l i f t s y s t e m by c h a n g i n g t h e g a p / foils. - I V Symposium o n N u m e r i c a l a n d
overlap combinations, looking f o r a configu- P h y s i c a l Aspects o f Aerodynamic Flows,
r a t i o n a b l e t o s a t i s f y t h e aerodynamic r e - CSULB, J a n u a r y 1989.
q u i r e m e n t s : h i g h e f f i c i e n c y a t s m a l l a n d me-
dium d e f l e c t i o n s ( t a k e - o f f and approach), 4. R U S S O , V., G u g l i o t t a , F., V i t i e l l o , V.:
h i g h v a l u e s o f CLmax a t medium a n d l a r g e d e - S t u d i o p e r l a D e f i n i z i o n e d i un S i s t e m a
f l e c t i o n s (approach and l a n d i n g ) . d i F l a p a Doppia Fessura e Confront0 con
The r e s u l t s c o n f i r m e d , i n g e n e r a l , t h e F l a p a S i n g o l a Fessura. - V I 1 Congress0
conclusions o f the two-dimensional studies; N a z i o n a l e A I D A A , N a p o l i , 1983.
i n p a r t i c u l a r (fig.211, a very interesting
r e s u l t i s t h e improvement o f v a l u e s r e l a t i v e 5. A v e r a r d o , M., A . : Study o f vane-flap
t o t h e maximum l i f t c o e f f i c i e n t a n d t h e configurations - I n t e r n a l ALENIA R e p o r t s
s t a l l i n g angle o f attack achievable, a t l a r - 5 6 7 9 0 0 1 3 ( 1 9 9 0 ) , 56790025 ( 1 9 9 0 ) .
ge d e f l e c t i o n s , b y means o f m o r e e x t e r n a l
f l a p p o s i t i o n s , i n accordance w i t h t h e theo- 6. A v e r a r d o , M., A., d e Leo, M.: Study o f a
r e t i c a l p r e d i c t i o n s and t h e i n d i c a t i o n s from new s i n g l e - s l o t t e d f l a p - I n t e r n a l A L E N I A
the two-dimensional experimental analysis R e p o r t s 5 8 7 9 1 0 1 7 ( 1 9 9 1 1 , 57691085 ( 1 9 9 1 )
( s e e sec.4.3, fig.16).
7. A l t h a u s , D., U u r z , U . : Uind-Tunnel
7. CONCLUSION t e s t s o n t h e SH316 a n d SH318 a i r f o i l s .
When t h e s t u d y o f t h e f l a p s y s t e m s t a r t e d , ( O r d e r e d b y ALENIA).
the best type o f h i g h - l i f t device, the best S t u t t g a r t , 1989.
f l a p p o s i t i o n s , t h e most s u i t a b l e v a l u e s f o r
t h e a n g l e s o f d e f l e c t i o n , were a l l unknown 8. E s p o s i t o , G., Manco, M.:
elements. b l i n d - T u n n e l t e s t s o n 1:15 powered model
Previous experiences concerning t h e o r e t i - o f Advanced A m p h i b i o u s A i r c r a f t .
c a l and/or experimental s t u d i e s of m u l t i - I n t e r n a l A L E N I A R e p o r t 57691101 (1991).
body systems and o t h e r d a t a d e r i v e d f r o m li-
t e r a t u r e o f wing s e c t i o n s were a b l e t o p r o - 9. A b b o t t , I . , H., v o n D o e n h o f f , A., E.:
v i d e a few p a r t i a l i n d i c a t i o n s a b o u t t h e a - Theory o f wing s e c t i o n s .
b o v e q u e s t i o n s and, i n g e n e r a l , t h e p r o b l e m s Dover P u b l i c a t i o n s Inc., New Y o r k , 1 9 5 9 .
33-8

GENERAL IN T H I S PROJECT

1. o o
D
.75
THEORETI U L - Preliminary & f i n i t i o n
of f l a p systam

.50
EXPER I MENTAL - Analysis of flap
system thragh
U-T tests

AND C M R E L A T I W
- Preliminary selection
of flap system

.25 AERODYNAMIC
/
/
/
! - D e f i n i t i m ( i n uhole or i n
pert) of neu flap s y s t e m
through e x p r i m n t a l andlor

.o I I I 1
-- theoretical aslysw

Choice of flap s p t m

Cptirnization of flap
.o 25 .50 .75 1.00v conf igurst ions
(pivot positions)
Fig.1 - Drag components of aircraft during
Redefinition
take-off from water up to lift-off of Ceomty(ies) 7
velocity. I
--------_______
,yes

GEDllETRV
- Corrluain definition
of f l a p system

ad

Fig.2 - Flow chart of standard activities

for the aerodynamic design.

-------
/' '@lo RApaaRDLIEOERWEll)

Fig.3 - Some of the flap systems studied;

retracted and deflected positions.

- - Sketch of the qeometrical factors

Fiq.4
(gap, overlap)-affecting the flap
position for a given deflection.

Fi 9.3 - (concluded) U

Fig.5 - Paneling of main-body/vane-flap

geometry: detail o f flap elements
and slot-entry region.
 33-9

'LIP
3 GEOMETRY B
* GEOMETRY A

Fig.6a - Different configurations of vane

and main-flap. Fig.6b - Theoretical pressure distributions.

LIP

w 3 GEOMETEY C
* GEOMETRY rl

7
Fig.7a - Vane-flap system: different
positions of the fore element. Fig.7b - Theoretical pressure distributions.

-5 -
CP
-4- 3 POSITION N o . 2
* POSITION N O . 1
-3 -

-2 -
-1 -

0-

1 -, 1 I

Fig.8a - Configurations o f vane-flap system Fig.8b - Theoretical pressure distributions

I for different overlap values. (detai 1 ) .
33-10

25
a
20

15

10

0
0 5 10 15 20 o( 25 25 .O -.25 -.50 -. 5
Cm
Fig.9 - Theoretical aerodynamic coefficients
of the airfoil equipped with the Fig.9 - (concluded) b) Pitching-moment
vane-flap system (0.30~). curves
a) Lift curves

CP
-8 - * SINGLE-SLOTTED
0 VANE-FLAP
-6 -
-4-

-2-

0-
-2J
I 1 I 1
Fig.lOa - Comparison of vane-flap and .o .25 .50
1.00 .75 x
single-slotted systems (0.30~). Fig.lOb - Theoretical pressure distributions.

CP
-8- * POSITION N O . 1
0 POSITION N o . 2
-6-

-4-

-2-

0-

2-1
I I I 1

Fig.lla - Different configurations o f .o .25 .50 :75 x 1.00

single-slotted flap (0.35~). Fig.llb - Theoretical pressure distributions.
 33-11

0 5 1015 o( 20
Fig.12 - Vane-flap 10.30cj (1) and single-
slotted (0.35~1 (2) systems; Fig.13 - Theoretical aerodynamic coefficients
L1, L2: respective lip positions. for vane-flap [0.30c] and single-
slotted [0.35c] systems.
a) Lift curves

4.0- 20-
CL o(
3.2- 15-

2.4- 10-

1.6- 5-

.EA 0-
I I I 1
.01 .02 .03 .04CD . 0 5 -.io -25 -.40 -.55
Cm
-.70
Fig.13 - (continued) b ) Polar curves Fig.13 - (concluded) c) Pitching-moment
curves

-10- -81 -2,

CP
-8-
* VANE-FLAP
0 SINGLE (35%)
-6 -
-4-

-2-

0-
* VANE-FLAP
2J 2-1 0 SINGLE (35%)
I I I I I I I
.O.25 .50 .75X 1.00 .O .25 .50 . 7 5 X 1.00
Fig.14 - Theoretical pressure distributions Fig.14 - (concluded) DF=15
of vane-flap (0.30~) and single-
slotted (0.35~) systems; DF-45.
33-12

-10-
1 L1 I I
CP 4 SINGLE (30%)
-8- 0 SINGLE (35%)

-6 -

\ \\ -4-

-2-

Fig.15a - Single-slotted flap systems; 0-

0.30~ (1) and 0.35~ (2).
L1, L2: respective lip positions. 2-

4.0
CL CL
3.2- 3.2

2.4
2.4-
1.6.
1.6-
.0‘

.8- .o- I I I I
I 1 I I I

-8 0 8 16 24 c( 3;
Fig.16 - Experimental lift curves concerning
rearward (1) and forward (2) posi- Fig.17 - Experimental lift curves for vane
tions of vane-flap geometry (0.30~). and single-slotted flaps (0.30~).

4.c 4.0
I I I
ct
3.2
451
CL
3.2
I 45 I
2.4 2.4

1.6 1.6

.8 .o
.o 0 w I
I
15 a 20
I I I I
-5 0 5 10 -5 0 .5 IO 15 o( 20
Fig.18 - Experimental lift curves for
airfoils equipped with vane-flap Fig.18 - (concluded) Re=1.5*10**6
system (0.30~); Re=5*10**6
 33-13

4.c,, I I I I

-7 0 '7 1'4 2 1 o( 28 -6 0 6 12 18 O! 24
Fig.19 - Experimental lift curves Fig.20 - Theoretical and experimental
for vane-aft (0.30~) and values of airfoil with vane-flap
single-slotted (0.35) flaps. system (0.30~); a) Lift curves

,3 . 1 -.l -.3 -.7

--.5 0 2 3 ov 4
Cm
Fig.20 - (concluded) b ) Pitching-moment Fig.21 - Variation of aerodynamic charac-
curves teristics of the aircraft versus
overlap values.
a) Maximum lift coefficient

r I I I b
0 1 2 3 ov 4
Fig.21 - (concluded) b ) Stalling angle
 RTD- 1

ROUND TABLE DISCUSSION

Dr. D. Woodward, DRA, Farnborough, U.K.

We have now reached t h e end o f t h e p r e s e n t a t i o n of papers, and now w e have come t o t h e time when w e
a t t e m p t t o p u l l t o g e t h e r a l l l e s s o n s t h a t have been l e a r n e d and demonstrated during t h e week. It i s my
p l e a s u r e t o c h a i r t h i s t e c h n i c a l d i s c u s s i o n s e s s i o n and i n t r o d u c e t o you your t e c h n i c a l e v a l u a t o r .

Before I do that I would l i k e t o t a k e a few minutes t o g i v e my p e r s o n a l impressions of what h a s happened

h e r e d u r i n g t h i s week. Unfortunately, I d i d n o t manage t o h e a r a l l t h e p r e s e n t a t i o n s because I w a s away
a t committee meetings, b u t n e v e r t h e l e s s t h e papers t h a t I d i d h e a r , mainly y e s t e r d a y a f t e r n o o n and
y e s t e r d a y evening and on t h e first day, g i v e me t h e f e e l i n g t h a t t h e Symposium has achieved m s t of t h e
o b j e c t i v e s that Lou Williams and I set f o r i t when we f i r s t put i t t o g e t h e r .

F i r s t l y , t h e papers y e s t e r d a y a f t e r n o o n and today have showed, I b e l i e v e , t h e many f a c e t s o f t h e o p e r a t i o n

that i n f l u e n c e t h e s e l e c t i o n of t h e h i g h - l i f t system. T h i s morning D r . F l a i g showed c l e a r l y t h e reasons
why t h e f l a p system on t h e A 320 had t o be redesigned f o r t h e mission f o r A 321. D r . Nark last n i g h t
i n t r o d u c e d t o u s a l l t h e o t h e r key f e a t u r e s t h a t have t o come t o g e t h e r i n o r d e r t o produce a s u c c e s s f u l
STOL a i r p l a n e . Dr. Mathews, y e s t e r d a y a f t e r n o o n , i n t r o d u c e d t h e important f e a t u r e s o f t h e mechanical
d e s i g n of t h e f l a p system. Mr. Averado introduced t o us t h i s morning t h e s p e c i a l f e a t u r e s a s s o c i a t e d w i t h
a n amphibious a i r c r a f t .

The CFD papers earlier i n t h e week demonstrated that w e now have t h e c a p a b i l i t y t o c a l c u l a t e t h e d e t a i l e d

flow around t h e h i g h - l i f t system; n o t y e t well enough t o dispense w i t h wind t u n n e l t e s t i n g , b u t w e l l
enough t o r e v e a l key f e a t u r e s o f t h e flow which are q u i t e d i f f i c u l t t o measure e x p e r i m e n t a l l y . I w a s
e x t r e m e l y s u r p r i s e d by t h e s i z e and shape of t h e s e p a r a t i o n r e g i o n beneath t h e s l a t cove. When we used t o
make c a l c u l a t i o n s o f p r e s s u r e d i s t r i b u t i o n s back i n t h e National High-Lift Program days, we c o u l d n ' t
c a l c u l a t e that s e p a r a t i o n r e g i o n so we u s e t o u s e e n g i n e e r i n g judgement and t h e f r e n c h curve t o draw what
w e thought w a s t h e s e n s i b l e shape. I have t o admit t h a t having seen t h e measurements by Dr. Almdaroglu
and t h e Navier Stokes c a l c u l a t i o n s , t h a t ye were a mile away from t h e shape that a c t u a l l y is t h e r e . It i s
much l a r g e r t h a n I would have thought i t was, and I found t h a t p a r t i c u l a r l y s t r i k i n g .

T h i r d l y , I was a l s o p l e a s e d t o see t h e g e n e r a l agreement t h a t t h e r e seems t o b e now on t h e importance and

t h e t y p e s o f s c a l e e f f e c t o n h i g h - l i f t systems and t h e importance o f t h e Mach number e f f e c t . This is a
message which we, from t h e 5 meter t u n n e l , have been g i v i n g o u t f o r some time now b u t i t is n i c e t o know
t h a t o t h e r people a r e g e t t i n g similar r e s u l t s and drawing t h e same conclusions.

So l e t us now f i n d o u t whether Don Whittley a g r e e s w i t h t h e t h i n g s t h a t I have s a i d . I am s u r e he h a s

much more t o s a y , much more d e t a i l e d and clear d i s t i n c t i o n s t o draw t h a n I have done. Don Whittley, who
i s your Technical Evaluator, began h i s t e c h n i c a l l i f e i n England, ( l i k e many Canadians), a t Saunders R o e
on t h e Isle o f Wight. Saunders Roe were w e l l known manufacturers o f f l y i n g b o a t s ; he t e l l s me h e used t o
r i d e on t h e towing c a r r i a g e over a s e a p l a n e tank making measurements - a n i n t e r e s t i n g experimental
technique which I have n o t had t h e p l e a s u r e of e x p e r i e n c i n g myself. He came t o Canada i n 1947 and j o i n e d
AVRO Canada. He worked on t h e C-102 j e t l i n e r , t h e CF-100 j e t f i g h t e r and continued t h e r e u n t i l 1960 which
c o i n c i d e d w i t h t h e c a n c e l l a t i o n of t h e Arrow p r o j e c t . I remember t h a t p a r t i c u l a r l y w e l l . It w a s about
t h e time when I was coming towards t h e end o f my a p p r e n t i c e s h i p a t d e Havilland o f H a t f i e l d i n England.
There were m n y people a year o r two ahead o f me who c r o s s e d t h e A t l a n t i c t o green f i e l d s t o work f o r AVRO
Canada on t h i s h i g h t e c h p r o j e c t and i t a l l went r a t h e r sour. I can remember one chap who came back. He
r e a c t e d v e r y q u i c k l y t o t h e c a n c e l l a t i o n and managed t o put h i s house on t h e market and s e l l i t about 3 o r
4 days b e f o r e everyone e l s e had recovered from t h e shock. H e managed t o r e c o v e r h i s money and set o f f
a g a i n , whereas o t h e r people, I heard, found i t much more d i f f i c u l t . So, t h a t was a v e r y t r a u m a t i c time, I
guess f o r AVRO Canada. He t h e n j o i n e d d e Havilland Canada and worked on powered l i f t f o r t h e rest o f h i s
career, working on t h e augmentor wing program and t h e e j e c t o r l i f t v e c t o r t h r u s t program, which involved
tests i n t h e Ames 40 x 80 and 80 x 120 t u n n e l . He h a s previous l i n k s w i t h AGARD, having been involved
w i t h t h e AGARD Advanced Aerosystems Study on AV-STOLdesignated AAS 14, and he worked on t h e f l i g h t
r e s e a r c h o f t h e augmentor wing J e t - S t o l a i r p l a n e which a l s o involved NASA Ames, and r e c e i v e d f o r that work
t h e McCurdy Award, which is t h e premier d e s i g n award f o r t h e Canadian Aerospace I n s t i t u t e . So h e i s w e l l
q u a l i f i e d t o e v a l u a t e t h e work that w e have been d o i n g h e r e t h i s week. So, now I i n v i t e you, Don
W h i t t l e y , t o t e l l u s whether we m d e a good j o b o f i t o r not.

Mr. D.C. Whittley, Canada

Mr. Chairman, Committee Members and Delegates. Contrary t o what has been suggested, i n f a c t I p l a n t o
keep my remarks q u i t e s h o r t making j u s t a few broad o b s e r v a t i o n s and t h e r e b y l e a v i n g a s much t i m e as
p o s s i b l e f o r t h e Round Table M s c u s s i o n . A s w e come t o t h e c l o s e o f 3 o r 4 days o f i n t e n s i v e d e l i b e r a t i o n
r e l a t i n g t o h i g h l i f t , one t h i n g I t h i n k is c l e a r l y e v i d e n t , we are involved n o t o n l y w i t h h i g h - l i f t , b u t
a l s o w i t h high-tech. It i s perhaps f a i r t o s a y t h a t t h e r e is more high-tech i n h i g h - l i f t t h a n i n any
o t h e r a s p e c t o f a i r c r a f t d e s i g n . T h e r e f o r e , p r o p o r t i o n a l l y i t demands a g r e a t d e a l of time and e f f o r t .
What then makes it so h i g h t e c h ?

F i r s t we have t h e u t t e r complexity o f t h e f l u i d dynamics involved. A t times i t seems a l m s t overwhelming,

does i t n o t ? Hence, t h e need f o r h i g h l y s o p h i s t i c a t e d t h e o r e t i c a l methods. Mny o f o u r s p e a k e r s made
r e f e r e n c e t o t h i s complexity i n t h e p a s t f e w days.

Secondly, t h e r e is t h e demanding requirement placed on e x p e r i m e n t a l f a c i l i t i e s and flow measuring d e v i c e s

i n o r d e r f o r r e s u l t s t o be a t a l l meaningful. Hence, t h e need f o r o u r high-tech f a c i l i t i e s . What t h e n
j u s t i f i e s t h e time and e f f o r t ? F i r s t , t h e r e i s t h e c r u c i a l importance o f g e t t i n g t h i s a s p e c t o f d e s i g n
c o r r e c t t h e f i r s t time around because t h e consequences o f a s h o r t f a l l , r e l a t i v e t o p r e d i c t i o n , can be
q u i t e harmful. Then t h e r e i s t h e knowledge that s u b s t a n t i a l b e n e f i t can result from improvements i n high
l i f t performance a n d / o r s i m p l i f i c a t i o n i n d e s i g n f o r t h e same performance. I t was Dr. Meredith who gave
u s some trade-off f i g u r e s i n t h i s r e s p e c t .
RTD-2

F i n a l l y , t h e r e i s t h e need t o reduce t h e c o s t and time o f t h e d e s i g n process, o r a t l e a s t keep them w i t h i n

bounds. It is a p p a r e n t t h a t improvements i n technology can h e l p i n a l l t h e s e r e s p e c t s .

Another broad o b s e r v a t i o n concerns t h e c l o s e i n t e r r e l a t i o n between t h e o r y and experiment a s s o c i a t e d w i t h

h i g h - l i f t technology. CFD can provide a n i n s i g h t t o t h e b a s i c f l u i d dynamics involved and t h e r e b y h e l p i n
t h e f o r m u l a t i o n of wind t u n n e l programs and i n t h e i n t e r p r e t a t i o n of wind t u n n e l r e s u l t s . It seems t o m e
i n t h i s s e n s e that CFD becomes almost a k i n t o flow v i z . S i m i l a r l y , advanced measuring t e c h n i q u e s can
d e t e c t unsuspected phenomena and i n t u r n , l e a d t o improvement i n t h e CFD code. I t h i n k i t was i n one o f
t h e q u e s t i o n and answer p e r i o d s that someone pointed o u t that w e a r e u n l i k e l y t o see b i g g e r and b e t t e r
wind t u n n e l s in t h e f o r e s e e a b l e f u t u r e . T h e r e f o r e , w e must make t h e b e s t o f what w e have. I b e l i e v e t h a t
t h e synergism whlch r e s u l t s from c l o s e a s s o c i a t i o n between t h e o r e t i c a l and experimental work i s t h e answer
t o t h i s p a r t i c u l a r problem. Also, j u s t y e s t e r d a y , t a l k i n g w i t h D r . Haines, he pointed o u t t h a t i t i s n o t
o n l y t h e d i s c i p l i n e s t h a t need t o i n t e r a c t : w e have t o e n s u r e t h a t t h o s e who understand t h e physics keep
c l o s e t o t h o s e who understand t h e mathematics. We may need t o i n t r o d u c e some f o r m 1 means, o r perhaps
should I s a y s t r u c t u r e d means t o make t h i s happen, r a t h e r t h a n j u s t g e t t i n g t o g e t h e r , as w e have i n t h i s
p a s t week, once i n a decade.

I n terms of o v e r a l l p e r s p e c t i v e we should b e reminded t h a t o u r H i g h - l i f t Symposium t h i s week has focussed

l a r g e l y on a p a r t i c u l a r segment o f h i g h - l i f t , namely, moderate t o h i g h a s p e c t r a t i o t r a n s p o r t a i r c r a f t .
True, w e had a b r i e f glimpse a t h i g h - l i f t as it relates t o combat a i r c r a f t , and i n t h a t r e s p e c t we were
reminded that t h i s i n v o l v e s more t h a n j u s t t a k e o f f and landing. S i m i l a r l y , e s p e c i a l l y last n i g h t , a peek
i n t o t h e r a t h e r d i v e r s e world, and p e c u l i a r world perhaps, of powered l i f t . W e might a l s o n o t e that w e
have n o t even d i s c u s s e d h i g h - l i f t as i t r e l a t e s t o t h e n e x t g e n e r a t i o n SST. No doubt t h i s and more would
make good s u b j e c t m a t t e r f o r some f u t u r e conference, b u t t h e p o i n t I r e a l l y wish t o make is t h a t
c o n c l u s i o n s drawn and recommendations made a s a r e s u l t of o u r meeting h e r e i n Banff would n o t n e c e s s a r i l y
h o l d t r u e a c r o s s t h e e n t i r e f i e l d . I myself happen t o have c e r t a i n r a t h e r f i x e d i d e a s about how one goes
a b o u t v e r i f i c a t i o n o f a powered-lift d e s i g n .

That o u r H i g h - l i f t Symposium h e r e i n Banff h a s been a g r e a t s u c c e s s is c e r t a i n l y w i t h o u t q u e s t i o n . We

have l e a r n e d o f marked p r o g r e s s i n CFD methods s i n c e B r u s s e l s 1984, which w a s t h e last h i g h - l i f t A G m D
symposium, and t h i s w a s a p p a r e n t w i t h r e s p e c t to b o t h v i s c i d - i n v i s c i d i n t e r a c t i o n methods and f u l l Navier
Stokes s o l u t i o n s , a l l c l e a r l y a t t h e c u t t i n g edge o f t h e h i g h - l i f t technology. W e have l e a r n e d o f t h e
i n c r e a s e i n t h e u s e o f p r e s s u r e d wind t u n n e l s t o i s o l a t e t h e effects o f Reynolds number and Mach number,
a s o u r Chairman has j u s t n o t e d ; o f improvements i n model t e s t i n g methods, a b e t t e r understanding o f
half-models f o r example; mre c a r e f u l c o r r e l a t i o n between t u n n e l and f l i g h t measurements; advanced
measuring t e c h n i q u e s t o probe t h e b a s i c f l u i d dynamics; t h e predominance o f v i s c o u s e f f e c t s and t h e need
f o r a much b e t t e r understanding. Apparently t h e r e have been a few s u r p r i s e s a l o n g t h e way s i n c e Brussels
'84 i n t h i s last r e s p e c t . We h e a r d about h i g h - l i f t technology i n t h e workplace and t h e extent t o which
i n d u s t r y has adopted new methods, and w e have seen that n o t h i n g i s l i k e l y t o r e p l a c e a 3-D test of t h e
f i n a l c o n f i g u r a t i o n a t r e l a t i v e l y h i g h Reynolds number, and so on. These d e l i b e r a t i o n s , i t would seem,
l e a d n a t u r a l l y t o two key q u e s t i o n s : f i r s t l y , how m t u r e is t r a n s p o r t h i g h - l i f t technology a s we f i n d i t
today? T h a t is t o s a y , i n terms o f i n d u s t r i a l u s e r a t h e r t h a n s c i e n t i f i c c u r i o s i t y . Are we i n f a c t now
a l m s t t h e r e , o r is t h e r e s t i l l a l o n g way t o go? For example, a t t h e Brussels meeting, I recall being
much impressed by a paper i n which M l l n e r and May d e s c r i b e d t h e CFD methods f o r h i g h - l i f t d e s i g n and
a n a l y s i s , which were a l r e a d y i n d a i l y use i n t h e b e i n g d e s i g n o f f i c e a t t h a t time.

F a i r l y comprehensive CFD methods were q u i t e w e l l e s t a b l i s h e d i n i n d u s t r y i n 1984. Since t h e n , over t h e

p a s t 8 y e a r s , CFD methods and more c a r e f u l t e s t i n g have become growing and important i n g r e d i e n t s o f
h i g h - l i f t d e s i g n procedures on a much broader f r o n t . C l e a r l y , t h e a d o p t i o n o f such t e c h n o l o g i e s t o
r e p l a c e p u r e l y e m p i r i c a l methods is a t r a n s i t i o n a l process and t a k e s p l a c e i n d i f f e r e n t ways in d i f f e r e n t
p l a c e s and t o s u i t d i f f e r i n g needs. I t h i n k t h e paper on t h e ATR 72 f l a p by P. Capbern o f A e r o s p a t i a l e
w a s a good example o f t h a t , and our f i n a l p a p e r today, by M. Averado s i m i l a r l y .

However, given that t h e r e are s u b s t a n t i a l performance g a i n s yet t o r e a p , and given that t h e r e remain
l u r k i n g r i s k s t o be e l i m i n a t e d , and given that t h e r e are f u r t h e r economies t o b e made i n terms o f d e s i g n ,
c o s t and time - all -
t h r e e most c e r t a i n l y t r u e t o some d e g r e e my second q u e s t i o n t h e n would be: where
should t h e emphasis and d i r e c t i o n l i e f o r t h e f u t u r e w i t h r e s p e c t to b o t h t h e o r y and experimentation?. A
few o f o u r s p e a k e r s have a l r e a d y expressed some o p i n i o n i n t h i s r e g a r d a l s o .

F i n a l l y , a word t o t h e a u t h o r s a b o u t my r e p o r t . I have been asked by t h e Executive Committee t o l a r g e l y

a v o i d a commentary paper by paper; that would simply make l i f e too easy f o r me, b u t r a t h e r provide an
o v e r a l l assessment of t h e t e c h n i c a l i s s u e s r e l a t i n g t o h i g h l i f t as a whole. I n view of t h i s , p l e a s e do
n o t f e e l spurned i f your p a r t i c u l a r paper i s n o t d e a l t w i t h i n my r e p o r t i n s p e c i f i c d e t a i l . So then,
back t o my two q u e s t i o n s and t h e n back t o t h e Chair.

F i r s t l y , "How mature i s o u r t r a n s p o r t h i g h - l i f t technology as we f i n d i t today?" and secondly, "Where

should t h e emphasis and d i r e c t i o n l i e f o r t h e f u t u r e w i t h r e s p e c t t o both t h e o r y and experiment?". I
hope, M r . Chairman, that some o f o u r time can be spent this morning i n response t o t h e s e two q u e s t i o n s .

M. L.J. Williams, NASA Washington, USA

Thank you. I f anyone would have any comments o r q u e s t i o n s , we w i l l be happy t o t a k e them.

Prof. D r . G.E.A. Meier, DLR, Gottingen, Germany

I would l i k e t o make some comments from t h e m i n t of view of a f l u i d dvnamicist who h a s n o t too much
e x p e r i e n c e i n t h e f i e l d o f t e c h n i c a l a p p l i c a t i o n s , b u t I would l i k e to-sumnarize my impressions on t h e
problems in o u r h i g h - l i f t d e v i c e s . F i r s t o f a l l , I see a main problem in t h e 3-D e f f e c t s which occur
because we have swept wing problems, swept wings w i t h end effects f o r a l l t h e d e t a i l s o f f l a p s , slats, and
s o on. Then w e have t h e 3-D s t r u c t u r e s o f t r a n s i t i o n which have n o t been mentioned h e r e , because i n a l l
o u r experiments w e have t o f o r c e t r a n s i t i o n and a l s o in t h e c a l c u l a t i o n , b u t we g e t a l s o 3-D e f f e c t s from
t h e t r a n s i t i o n t o turbulence. The f o u r t h t h i n g I would l i k e t o mention are a e r o e l a s t i c deformations which
have n o t been t a c k l e d too much, b u t which I t h i n k a l s o are a s e v e r e problem in experiments i n f l i g h t t e s t s .
 RTD-3

Another major regime o f problems is t h e flow s e p a r a t i o n . Here we have real d e f i c i t s in knowledge about
l o c a t i o n and shape o f s e p a r a t i o n bubbles and how t o determine them t h e o r e t i c a l l y . Then t h e r e i s t h e
i n f l u e n c e o f t r a n s i t i o n t o flow s e p a r a t i o n ; t h i s i n f l u e n c e i s n o t q u i t e c l e a r , e s p e c i a l l y t h e Reynolds
number problem which h a s been mentioned s e v e r a l times. Another p o i n t which h a s n o t come through v e r y much
is unsteady flow. I t h i n k a l o t o f t h e s e flow f i e l d s we have c o n s i d e r e d h e r e i n t h e conference are h i g h l y
unsteady and cause f l u t t e r and i n t h e t r a n s o n i c case a l s o b u f f e t . My f e e l i n g i s that this i s a l s o c a u s i n g
a t e c h n i c a l problem w i t h material f a t i g u e . Fina.lly, t h e o n s e t o f s t a l l and p o s t stall is a l s o something
which h a s t o b e taken up.
So from t h i s l i s t o f problems I have deduced w h a t w e have t o do, a l o t o f work f o r f l u i d dynamicists. We
have t o provide, I t h i n k , more knowledge and fundamentals i n unsteady compressible 3-D flows, w e have t o
f i n d good boundary l a y e r models f o r t h e s e complicated flow f i e l d s ; we have t o t a c k l e t h e 3-D s e p a r a t i o n
and attachment problems; w e have t o t h i n k over advanced geometries f o r h i g h - l i f t a p p l i c a t i o n s and w e have
a l s o t o make c o n s i d e r a t i o n s about high Reynolds number flows. F i n a l l y , we need more knowledge about j e t
and s w i r l i n t e r a c t i o n s w i t h a i r f o i l f l a p s , slats and s o on. T h i s is my l i s t o f problems, I have no
s o l u t i o n s , b u t I t h i n k t h a t i t i s worthwhile t h a t f l u i d dynamicists a c c e p t and d e a l w i t h t h e s e p r a c t i c a l
problems.

M r . L.J. Williams, NASA Washington, USA

That was v e r y good and a p p r o p r i a t e . It l o o k s l i k e a n o t h e r s u b j e c t f o r a symposium.

M. J. Bosquet, A e r o s p a t i a l e
I would l i k e t o share t h e f e e l i n g I had d u r i n g t h i s conference t a k i n g t h e m i n t of view o f a n a i r c r a f t
manufacturer. I t h i n k t h e h i g h - i i f t d e s i g n seems t o be l a r g e l y dominated by experimental work. But i t
seems t h a t t h e r e are two domains. The first one is r e p r e s e n t e d by t h e wings which have v e r y small o r zero
sweep a n g l e ; t h e o t h e r one by l a r g e sweep a n g l e wings. For t h e f i r s t domain ( s m a l l sweep a n g l e ) , i t seems
t h a t t h e 2-D c a l c u l a t i o n s have enough accuracy t o r e p r e s e n t t h e flow, and i n t h i s c a s e I g o t t h e f e e l i n g
that w h a t we c a l l t h e Navier Stokes method seems t o b e almost ready t o e n t e r i n t o t h e deaign team. I s a i d
a l m s t ready because i t seems t h a t f i r s t , t h e problem o f g r i d g e n e r a t i o n seems t o b e s o l v e d , t h e n
t u r b u l e n c e modelling is n o t completely s o l v e d , b u t i t seems t h a t i t is s u f f i c i e n t t o tune some
c o e f f i c i e n t s and t h i s could b e good enough. The t h i r d problem i s t h e computational time which seems n o t
t o b e so h i g h and could be s u f f i c i e n t l y small enough f o r i n d u s t r i a l a p p l i c a t i o n s . Then 1 expect t h a t we
w i l l s e e i n t h e n e a r f u t u r e Navier Stokes o r l a r g e coupling methods e n t e r i n t o t h e i n d u s t r i a l d e s i g n f i e l d .
I n t h e second domain w i t h l a r g e sweep I t h i n k t h a t 2-D c a l c u l a t i o n s are n o t s u f f i c i e n t enough because i n
t h e case of wings w i t h l a r g e sweep a n g l e , three dimensional boundary l a y e r c a l c u l a t i o n s are necessary.
Two-D c a l c u l a t i o n s are n o t v e r y I n t e r e s t i n g . So i n t h i s case probably we w i l l need f o r a l o n g t i m e v e r y
l a r g e wind t u n n e l f a c i l i t i e s and a l o t o f work i n t h e wind t u n n e l s . B u t i t seems t h a t i n t h i s case
coupling methods c o u l d b e o f i n t e r e s t .

M r . A.B. Haines, ARA, UK

A s Don W h i t t l e y s a i d , we have come a l o n g way s i n c e 1984, both in CFD and i n t h e understanding of t h e flow
over a wing w i t h h i g h - l i f t devlces. But D r . Woodward s a i d when h e was i n t r o d u c i n g t h i s s e s s i o n , that
t h e r e w a s a f a i r amount o f data a t h i g h - l i f t a v a i l a b l e t o study even b e f o r e we decided t o b u i l d a 5 metre
t u n n e l i n England and an F 1 t u n n e l In France. I t h i n k he posed t h e q u e s t i o n whether w e took t h e r i g h t
d e c i s i o n . A t t h a t time, w e were w e l l aware t h a t t h e r e were s t r o n g Reynolds number e f f e c t s up t o about 6
o r 7 m i l l i o n and that t h e r e were n o t only Reynolds number e f f e c t s , t h e r e were Mach number e f f e c t s so that
w e needed a p r e s s u r i z e d t u n n e l . We have seen plenty o f examples o f t h a t in t h e last t h r e e days. But w e
have a l s o s e e n examples o f s i g n i f i c a n t s c a l e e f f e c t s a t Reynolds numbers above 6 o r 7 m i l l i o n : scale
e f f e c t s which are n o t s o understandable and are n o t s o p r e d i c t a b l e . So, i n a n i d e a l world, w e need
t u n n e l s f o r even h i g h e r Reynolds numbers. I t h i n k i t was M r . Meredith y e s t e r d a y that w e have g o t t o f a c t
t h e f a c t t h a t w e m y n o t g e t t h e s e t u n n e l s , and t h e r e f o r e , w e must e x p l o i t CFD. I would l i k e t o make two
p o i n t s a b o u t whether t h a t d o e s meet t h e need.

F i r s t o f a l l , I t h i n k i f you are going t o b e l i e v e that, t h e CFD c a l c u l a t i o n s f o r any p a r t i c u l a r case

should be made f o r more t h a n one Reynolds number. The d i s a p p o i n t i n g t h i n g t o me i n t h i s conference is
t h a t when t h e r e have been CFD c a l c u l a t i o n s , t h e y have been made merely f o r one Reynolds number, e i t h e r a
model test Reynolds number o r i n one o r two c a s e s a f u l l - s c a l e Reynolds number. Now t o r e p e a t a
c a l c u l a t i o n f o r more t h a n one Reynolds number I am well aware means a s i g n i f i c a n t amount of work because
t h e t r a n s i t i o n mechanism a s David showed t h e f i r s t morning may b e v e r y d i f f e r e n t f u l l - s c a l e and model
s c a l e . You don't j u s t p r e s s a b u t t o n and s a y t h a t you w i l l r e p e a t t h e c a l c u l a t i o n p u t t i n g i n a d i f f e r e n t
R. You have g o t t o b e g i n your t h i n k i n g a l l over a g a i n and a s k w h a t is t h e t r a n s i t i o n mechanism, etc. I
t h i n k t h a t i f t h e CFD people d i d a f a i r number o f examples showing how results v a r i e d w i t h Reynolds number
w i t h f u l l understanding o f t h e p h y s i c s , w e would l e a r n q u i t e a l o t . I t i s a p i t y that a t t h e moment, t h e
CFD people have n o t shown any keenness t o d o t h i s . They have n o t recognized t h a t t o d o i t f o r more t h a n
one Reynolds number is a n important element, and that i s where w e are going t o have t o r e l y on CFD.in t h e
f u t u r e i f Mr. Meredith is r i g h t .

The second p o i n t is, "can w e c a l c u l a t e t h e real s i t u a t i o n ? " . Most o f t h e CFD w e have seen has been e i t h e r
f o r 2-D wings o r a t t h e b e s t f o r 3-D wings where t h e wings are simple and t h e flow i s reasonably q u a s i
two-dimensional. I t h i n k t h a t t h e CFD people should s t u d y one p a r t i c u l a r f i g u r e i n Herr F l a i g ' s paper
this morning which drew a t t e n t i o n t o t h r e e problem a r e a s on h i s a i r c r a f t . One w a s a t t h e wing r o o t , one
was c l o s e t o t h e wing-pylon-nacelle j u n c t i o n and one w a s near t h e t i p . I n none o f t h o s e areas w a s t h e
flow a n y t h i n g l i k e two-dimensional. I have met young a e r o d y n a m i c i s t s who have s a i d , "Ah, b u t i f t h e
a i r c r a f t d e s i g n e r w a s doing h i s job p r o p e r l y , t h e r e wouldn't be problem areas t h e r e . The whole wing would
behave l i k e q u a s i two-dimensional flow". But someone who s a y s that completely misunderstands t h e aims o f
a i r c r a f t d e s i g n . To h a v e t h e whole wing stall a t t h e same mment would b e c a t a s t r o p h e . These areas where
t h e flow i s n o t 2-D a r e n o t n e c e s s a r i l y problem a r e a s . They are a r e a s where by d e l i b e r a t e d e s i g n you
produce a premature s m l l o r a l a t e stall i n o r d e r t o get t h e c o r r e c t s t a l l d e v e l o p e n t over t h e wing. So
t h e s e are f e a t u r e s o f t h e flow that j u s t must be c a l c u l a t e d i f we are going t o do t h e j o b by CFD. Those
are t h e two p o i n t s I wanted t o make. W e are s t i l l a l o n g way t o go. We've gone from 1984 t o 1992; 8
y e a r s t o 2000. I t h i n k w e need a n o t h e r c o n f e r e n c e b e f o r e 2000.
RTD-4

Dr. D. Woodward
I just wanted t o come back a l i t t l e b i t on t h e comments m d e by P r o f e s s o r Meier. He mentioned in
p a r t i c u l a r unsteady flow. I t h i n k w e need n o t t o g e t o u r s e l v e s in a p o s i t i o n where we c r i t i c i z e too much
what we have done h e r e t h i s week.

We d i d n ' t s e t o u t t o cover v o r t e x flows and v o r t e x f l a p s and h i g h - l i f t due t o v o r t i c e s because w e had a

conference on that o n l y a b o u t a year ago. Then t h e o t h e r comment h e made was that w e h a d n ' t done enough
on t r a n s i t i o n . I suppose t h a t one c a n ' t d i s a g r e e - one h a s n e v e r done enough on t r a n s i t i o n . I t h i n k i t
i s worth h i g h l i g h t i n g t h e f a c t that w e have, during t h i s symposium and p r e v i o u s l y , i d e n t i f i e d t r a n s i t i o n
mechanism which were regarded a s somewhat academic 1 0 y e a r s ago -in p a r t i c u l a r t h e attachment l i n e
t r a n s r l t i o n problem - and shown that t h i s i s a mechanism which o c c u r s f r e q u e n t l y a t N g h Reynolds number
and has major e f f e c t s on t h e way t h e flow develops. So I t h i n k t h a t w e have gone some way i n covering h i s
comments on t r a n s i t i o n .

R. Bengelink, Boeing, S e a t t l e , USA

I would l i k e t o make a couple o f o b s e r v a t i o n s based on t h i n k i n g a b o u t many o f t h e p r e s m t a t i o n s of t h i s
week a l s o . A i r p l a n e s a r e designed by engineers. Engineering is n o t something w e l e a r n a t school. We
l e a r n t h e f o u n d a t i o n a l knowledge a t s c h o o l b u t e n g i n e e r i n g i s something t h a t w e l e a r n by p r a c t i c i n g i t
over t h e years. We l e a r n t h a t we are t o use t h e t o o l s that are a v a i l a b l e t o u s t o develop a n i n s i g h t i n t o
what i s going on and as a result come up w i t h an aerodynamic c o n f i g u r a t i o n t h a t d o e s t h e j o b b e t t e r t h a n
h a s been done before. While t h e r e c e r t a i n l y h a s n o t , a t least t h i s week, been a n adversarial r e l a t i o n s h i p
between t h o s e who are f o c u s s i n g on t h e CFD t o o l s and t h o s e who are f o c u s s i n g on t h e e x p e r i m e n t a l t o o l s ,
t h e r e c e r t a i n l y c o n t i n u e s t o be a n atmosphere o f competitiveness. I would l i k e t o s u g g e s t that given t h e
l e v e l of understanding t h a t w e have today w i t h each of t h o s e t o o l s t h a t w e have t o develop more o f a
teamwork a t t i t u d e . Today we do much b e t t e r a t b e i n g a b l e t o v i s u a l i z e t h e flow f i e l d t h a n i n c a l c u l a t i n g
f o r c e s w i t h CFD. From t h e wind t u n n e l s i d e o f t h i n g s i t is t h e o t h e r way around. The expensive and v e r y
d i f f i c u l t t e s t s a r e t h o s e where we are t r y i n g t o understand t h e flow f i e l d around t h e model, such as t h e
three-dimensional f e a t u r e s o f so-called two-dimensional t e s t i n g . I t h i n k t h a t i f both t h e computational
and t h e experimental groups saw themselves as members o f t h a t same team, c h a r t e r e d t o p r o v i d e t h e tools
n e c e s s a r y f o r a good e n g i n e e r t o b e a b l e t o d o h i s job p r o p e r l y and took t h a t t e a m approach; f o r example,
how d o we blend a n a b i l i t y t o understand t h e flow f i e l d u s i n g CFD w i t h t h e a b i l i t y t o understand what i s
happening t o t h e measured f o r c e s on t h e m d e l p r o p e r l y , w e m i g h t f i n d o u r s e l v e s a b l e t o p r o g r e s s more
r a p i d l y . U n t i l we g e t t h o s e flow models in o u r mind and a c t on themas engineers, we are n o t r e a l l y going
t o make t h e k i n d o f p r o g r e s s in developing a i r p l a n e s t h a t w e should.

Prof. Ir. E. Obert, Fokker A i r c r a f t BV, Netherlands

I would l i k e t o add a remark t o what t h e previous speaker s a i d . W h a t s t r u c k me t h e p a s t couple o f days is
that t h e people who are working on CFD a n d are v e r y much involved in what is happening in t h e boundary
l a y e r o f t e n refer t o t u r b u l e n c e models. That means t h e y p r i m a r i l y Consider t u r b u l e n t boundary l a y e r s .
We, as e n g i n e e r s , a r e i n t e r e s t e d i n t h e q u e s t i o n : When does t h e flow s e p a r a t e ? I t i s assumed that
s e p a r a t i o n w i l l be more p r e d i c t a b l e due t o a b e t t e r r e p r e s e n t a t i o n of t h e flow p h y s i c s by t h e s e t u r b u l e n c e
models. T h a t means CFD s p e c i a l i s t s talk about t h e question: When d o e s t h e t u r b u l e n t boundary layer
s e p a r a t e ? This s u g g e s t s t h a t C ~ m xis p r i m a r i l y determined by t u r b u l e n t boundary l a y e r s e p a r a t i o n ,
t r a i l i n g edge s e p a r a t i o n .

I am aware t h a t many o f t o d a y ' s t r a n s p o r t a i r c r a f t which have v e r y l a r g e wing c h o r d s and consequently even

i n t h e s t a l l o p e r a t e a t a f a i r l y h i g h Reynolds number r e a c h C p a x through t r a i l i n g - e d g e s e p a r a t i o n .
However, I f i n d i t d i f f i c u l t t o b e l i e v e that l e a d i n g edge s t a l l , wNch we f r e q u e n t l y see i n t h e wind
t u n n e l i s completely i r r e l e v a n t f o r full-scale a i r c r a f t . Having been on a number o f s t a l l tests myself on
t u f t e d F28's and FlOO's, my f e e l i n g is that l e a d i n g edge stall is s t i l l an important p h y s i c a l phenomena
f o r f u l l - s c a l e a i r c r a f t . I a m aware t h a t t h i s m t t e r i s v e r y d i f f i c u l t t o t a c k l e , even m r e d i f f i c u l t
t h a n answering t h e q u e s t i o n , "When d o e s t r a i l i n g edge s e p a r a t i o n occur?", but I s t i l l t h i n k that we should
be aware t h a t t h a t i s a n area t h a t d e s e r v e s a t t e n t i o n .

M r . B. E l s e n a a r , NLR, Netherlands
I f I might c o n t i n u e a l i t t l e b i t a l o n g t h i s l i n e . I missed b a s i c e x p e r i m e n t a l c o n t r i b u t i o n s o f t h e
p h y s i c s o f h i g h - l i f t problems. I am t h i n k i n g about t h i n g s l i k e bubble b u r s t s , about l o c a l s e p a r a t i o n s ,
about mixtng o f wakes and boundary l a y e r s . W e need b u i l d i n g block experiments in t h a t area t o understand
t h e physics. I d i d n ' t n o t i c e any paper that r e a l l y addressed t h e s e t o p i c s . In a d d i t i o n t o t h i s I would
l i k e t o mention that I t h i n k t h a t i n view o f t h e v e r y complicated physics e s s e n t i a l t o h i g h - l i f t , I would
r a t h e r see s o l u t i o n s o f Navier Stokes e q u a t i o n s f o r i s o l a t e d p a r t s o f t h e flow f i e l d , t o l o o k i n t o d e t a i l
t o t h e bubble d e v e l o p e n t s , t o s e p a r a t i o n , t o t h e mixing o f l a y e r s r a t h e r than a p p l y i n g Navier Stokes
codes t o complete multi-element a i r f o i l c o n f i g u r a t i o n s and t o look a t t h e f i n a l r e s u l t s i n terms o f
p r e s s u r e d i s t r i b u t i o n . I t h i n k t h a t t h e b u i l d i n g block p a r t was a l i t t l e b i t n e g l e c t e d in t h i s conference

M r . L.J. Williams
I agree with t h a t . We were hoping t o get more papers a l o n g that l i n e .

M r . P. Capbern, A e r o s p a t i a l e , Toulouse, France

In t h e s e t h r e e days we h a v e seen a l o t of comparisons between c a l c u l a t i o n and experiment, b u t t h e r e i s
something which appeared v e r y c l e a r l y . &erybody t r i e d t o compare C~maxl e v e l from t h e i r c a l c u l a t i o n s
t o experiment, b u t we have h a r d l y s e e n any r e s u l t of comparison f o r drag. I t h i n k t h a t i f C p x is
o f t e n t h e key i s s u e f o r a p r o j e c t , t h e r e are some p r o j e c t s , and t h e y are n o t so rare, where t h e l i f t t o
d r a g r a t i o i n t h e second segment climb is mre important t h a n C ~ m a xi t s e l f . So t h e r e is perhaps a
message f o r people involved in method development that t h e y should t r y t o v a l i d a t e t h e i r methods f o r drag
p r e d i c t i o n . I know i t i s a v e r y hard j o b , b u t we should t r y t o d o that e s p e c i a l l y w i t h a far f i e l d
c o n d i t i o n and n o t o n l y by i n t e g r a t i n g t h e p r e s s u r e f i e l d and t h e s k i n f r i c t i o n f i e l d on t h e shape i t s e l f .
I am convinced t h a t some i n t e r e s t i n g i n v e s t i g a t i o n s could be achieved and t h e r e f o r e , i t w a s a message I
would l i k e t o send t o t h o s e people.
 RTD-5

Mr. L.J. Williams

1 t h i n k i t is a v e r y good message and a challenge.
D. A s h i l l , DRA, Bedford, UK
Barry Haines has a l r e a d y h i g h l i g h t e d t h e importance o f understanding scale e f f e c t s a t high Reynolds
numbers, and w e have had a number o f p r e s e n t a t i o n s during t h e l a s t few days on a t t a c h m e n t l i n e t r a n s i t i o n
and r e - l a m i n a r i z a t i o n i n d i c a t i n g t h e i r s i g n i f i c a n c e .

I would l i k e t o mention a n o t h e r p o s s i b l e scale e f f e c t which i s t h a t even a f t e r t h e flow becomes t u r b u l e n t

c l o s e t o t h e l e a d i n g edge, t h e t u r b u l e n t boundary layer w i l l l o c a l l y have q u i t e a low Reynolds number, l e t
u s s a y based on wmentum t h i c k n e s s . This is a n o t h e r p o t e n t i a l source o f u n p r e d i c t a b l e scale e f f e c t s which
I I t h i n k people need t o t a k e i n t o account, e i t h e r t h e o r e t i c a l l y o r i n i n t e r p r e t i n g t h e i r wind t u n n e l t e s t
data.
Prof. J.W. S l o o f f , NLR, Netherlands
I would l i k e t o add a l i t t l e t o what s e v e r a l o f t h e preceeding s p e a k e r s have s a i d a b o u t t h e physics o f t h e
flow; t h a t r e l a t i v e l y l i t t l e a t t e n t i o n was paid t o t h a t , and a l s o t o P a t A s h i l l ' s remark on t h i n g s that
may happen a t t h e l e a d i n g edge. It would seem, given t h e fact t h a t l i t t l e a t t e n t i o n w a s p a i d t o t h e
p h y s i c s o f h i g h - l i f t systems, that i f we understand e v e r y t h i n g about t h e v a r i o u s mechanisms o f stall. I
am n o t q u i t e convinced t h a t that is t h e case. For example, I am aware o f a t least three d i f f e r e n t t y p e s
o f l e a d i n g edge s t a l l mechanisms: s h o r t bubble and l o n g bubble b u r s t i n g , and a t h i r d v a r i a n t which I call
t u r b u l e n t l e a d i n g edge stall. I b e l i e v e t h e n o t i o n of t h e l a t t e r was i n t r o d u c e d a t t h e former h i g h - l i f t
meeting o f t h e F l u i d Dynamics Panel. There w e have a s i t u a t i o n w i t h a laminar s e p a r a t i o n bubble followed
by r e a t t a c h m e n t , t r a n s i t i o n a t o r around r e a t t a c h m e n t , and a t u r b u l e n t boundary l a y e r w i t h a v e r y s t e e p
a d v e r s e p r e s s u r e g r a d i e n t r i g h t a f t e r t h e bubble i n which t h e f r i c t i o n c o e f f i c i e n t goes through zero
l o c a l l y when t h e a n g l e o f a t t a c k is i n c r e a s e d . It i s v e r y d i f f i c u l t t o d i s c r i m i n a t e between t h i s k i n d o f
t u r b u l e n t l e a d i n g edge s t a l l and a bubble b u r s t i n g type s t a l l . I have n o t seen a n y t h i n g o f such n a t u r e a t
t h i s symposium. Moreover, I t h i n k , now t u r n i n g t o CFD, t h a t i t w i l l t a k e q u i t e some time b e f o r e CFD codes
w i l l be a b l e t o d i s c r i m i n a t e between t h e v a r i o u s types o f l e a d i n g edge s t a l l , o r o t h e r types of stalls f o r
that m a t t e r . The d e t a i l s o f t h e flow mchanisms, i n t h e l e a d i n g edge area i n p a r t i c u l a r , are such t h a t ,
a p a r t from t h e problems w i t h l o c a l t u r b u l e n c e and t r a n s i t i o n modelling, we a l s o have a r e s o l u t i o n problem
i n t h e sense t h a t w e w i l l need very, v e r y small meshes in o r d e r t o r e s o l v e a l l t h e phenanena. CFD f o r
h i g h - l i f t , even i n two dimensions, w i l l n o t be mature u n t i l w e are a b l e t o d i s c r i m i n a t e between t h e
v a r i o u s t y p e s o f p h y s i c a l phenomena.

Another p o i n t that I would l i k e t o make o r b r i n g t o your a t t e n t i o n is t h e p o s s i b i l i t y o f e x p l o i t i n g t h e

t h i r d dimension as I c a l l it. On three dimensional wings w e have part-span f l a p s and slats, w i t h a l o t o f
v o r t i c e s coming o f f a l l t h e edges. These may g i v e rise t o , e.g., h i g h e r induced d r a g l e v e l s than w e would
l i k e . It h a s occurred t o me t h a t t h e r e may be a p o i n t i n i n v e s t i g a t i n g whether f l a p t r a c k f a i r i n g s and
t h e l i k e could s e r v e t h e d u a l purpose o f covering t h e gear and mechanisms and of promoting a more o r d e r l y
t y p e o f flow a t t h e f l a p edges. For example i n such a way that t h e y l e a d t o a lower l e v e l o f induced
d r a g , i n t h e same s e n s e as a w i n g l e t works a t t h e t i p o f a wing. One might, perhaps a l s o g e t a l i t t l e
i n c r e a s e i n C~max,because you have smaller l o s s e s a t t h e f l a p edges. I would l i k e t o see some
r e s e a r c h of t h i s n a t u r e i n t h e f u t u r e . I have t h e f e e l i n g that t h e r e may be some b e n e f i t i n t h e r e .

Mr. L.J. Williams

I t h i n k t h a t t h a t i s a v e r v good comment. Carrving t h a t even f u r t h e r I would l i k e t o see some work t h a t
combines t h e h i g h - l i f t s y s f e i w i t h t h e wake vorfex-problem, t r y i n g t o e l e v i a t e t h a t w h i l e g e t t i n g lower
induced d r a g .

Dr. D. Woodward
There are two comments I would l i k e t o make. One i s r e l a t e d t o what Prof. Obert was saying. It i s
c e r t a i n l y t r u e that some f u l l - s i z e aircraft a r e a s s a i l e d with a l e a d i n g edge s t a l l behavior. The B r i t i s h
Aerospace Hawk c e r t a i n l y has a l e a d i n g edge stall. This has been shown by doing o i l f l o w s i n f l i g h t and
when t h e a i r p l a n e l a n d s you can s t i l l see t h e o i l f l o w i n t h e bubble. The second i n v o l v e s a n o t h e r
i m p o r t a n t d e t a i l o f t h e flow on which we h a v e n ' t s e n a n y work - and t h i s is r e l a t e d t o t h e
c o m p r e s s i b i l i t y e f f e c t on h i g h - l i f t s y s t e m . I f you a c t u a l l y do t h e c a l c u l a t i o n i n 2-D o f t h e e x t e r n a l
flow around a h i g h - l i f t a i r f o i l , you f i n d t h a t t h e s u p e r s o n i c patch, which we a l l have i d e n t i f i e d from t h e
s u r f a c e p r e s s u r e d i s t r i b u t i o n , i s o n l y some 10 t o 12 boundary l a y e r t h i c k n e s s e s high. This is v e r y
d i f f e r e n t from t h e t r a n s o n i c shock wave boundary l a y e r i n t e r a c t i o n and t h e r e must be a much ' l a r g e r
i n t e r a c t i o n o f t h e boundary layer on t h e shock p a t t e r n . I n a d d i t i o n , o f c o u r s e , i t is normally a shock
i n t e r a c t i o n w i t h a laminar boundary and n o t a t u r b u l e n t one. This is q u i t e a n unknown f l o w f i e l d and
since t h e key f e a t u r e i s v e r y small, I d o n ' t know how t h e experimentmight be conducted.

M r . F. Kafyeke, Bombardier Canadair, Canada

I would l i k e t o make two comments from an a i r p l a n e manufacturer's p o i n t of view. F i r s t , I would l i k e t o
know i f t h e experimental r e s e a r c h t h a t has been shown h e r e , t h e n a t i o n a l h i g h - l i f t program i n t h e UK, o r
t h e G a r t e u r program in Europe, w i l l e v e n t u a l l y f i n d i t s way i n t o d a t a s h e e t s which can h e l p d e s i g n e r s in a
p r e l i m i n a r y d e s i g n phase o f a n a i r p l a n e . Secondly, can someone t a l k about t h e e f f e c t s o f t h i n g s l i k e slat
g a p s and s t e p s and t h e s e a l i n g o f slats and t h e e f f e c t o n t h e o v e r a l l performance o f t h e real a i r p l a n e . I
wonder i f some r e s e a r c h h a s been done i n t h a t area.

Mr. L.J. Williams

That sounds l i k e a good q u e s t i o n f o r Dr. Woodward t o a d d r e s s .
RTD-6

Dr. D. Woodward
Unfortunately, I t h i n k that t h e answer i s v e r y s i m p l e , i t i s n o t o t h e f i r s t q u e s t i o n . For t h e reason
t h a t d e s p i t e a l l t h e work we d i d , i t is s t i l l n o t enough t o form w h a t one would b e happy t o s e e a s a d a t a
s h e e t on which people m i g h t b a s e p r e l i m i n a r y d e s i g n s . We a c t u a l l y s p e n t q u i t e a l o n g time t r y i n g t o
a n a l y s e t h e drag, f o r i n s t a n c e . Now, b e a r i n g i n mind, o f c o u r s e , that most o f t h e work that we d i d was on
t h a t end-plate model, we faced very g r e a t d i f f i c u l t y i n t r y i n g t o e x t r a c t meaningful d r a g data from t h a t
model. I d o n ' t know whether you had a chance t o read t h e w r i t t e n paper, b u t w e s e t o u t o r i g i n a l l y s a y i n g
that t h e r e was n o problem about t h i s , w e can d o a wake t r a v e r s e t h e same as we d o i n 2-D flow, and t h i s
w i l l t e l l u s what t h e d r a g is. Unfortunately, because o f t h e l a r g e t r a i l i n g v o r t i c i t y , t h e drag
c a l c u l a t e d t h i s way v a r i e s downstream v e r y r a p i d l y a s t h e wake g e t s s t r e t c h e d o u t , s o we gave up t h a t i d e a
and t h e n w e decided to t r y t o s u b t r a c t t h e induced drag from t h e f o r c e b a l a n c e measurements. Some people
b e l i e v e d that t h e y could do t h a t and o t h e r s were more s k e p t i c a l . I d o n ' t t h i n k t h a t w e r e a l l y managed t o
g e t a n y t h i n g o t h e r t h a n comparative d a t a on t h e d r a g i n p a r t i c u l a r .

When i t comes t o maldmum l i f t , maybe you could d o something, b u t you need t o d o a l o t more v a r i a t i o n s in
f l a p and slat d e s i g n t h a n we did. W e d i d two slat chords and a r a n g e o f Krugers, b u t t h e r e a r e a l o t o f
o t h e r q u e s t i o n s t h a t s t i l l were n o t answered by t h a t d a t a set and w e were doing that work f o r 6 t o 7
years. I t slowed down towards t h e end because t h e funding was reduced, b u t i t s t i l l took a l o n g t i m e t o
g a t h e r . A l a r g e r data s e t would r e q u i r e a correspondingly l a r g e r investment of time and money. I would
have t o pass t h e q u e s t i o n on s e a l i n g t o e i t h e r someone from Boeing o r Airbus because I do not know t h e
answer t o t h a t . Anybody g o t a n y comments?
No one is prepared t o say. I t h i n k t h a t s e a l i n g i s a g r e a t problem, b u t I don't have any v e r y p r a c t i c a l
experience o f i t .

Prof. Dr. Ir. J.L. Van Ingen D e l f t I l h i v e r s i t y , Netherlands

I d o n ' t have a comment but a q u e s t i o n . The q u e s t i o n of t h e preceding speaker was, " w i l l t h e s e b a s i c

t h i n g s f i n d t h e i r way i n t o data sheets". A q u e s t i o n t o t h e d e s i g n e r s might be, "do you need data s h e e t s
i n t h e f u t u r e o r can we come as f a r as p r o v i d i n g computer programs which are reasonably a c c u r a t e t o be
used a s a design tool". To me, my f i r s t r e a c t i o n is that t h e s e data s h e e t s a r e something o f t h e past, b u t
I am n o t a d e s i g n e r , s o I t h i n k t h e y should r e a c t t o t h a t .

Prof. Ir. E. Obert, Fokker A i r c r a f t

Although h e s i t a t i n g t o react, b e i n g a r e p r e s e n t a t i v e from one o f t h e s m a l l e r companies, I am s t r i c t l y
speaking f o r myself, n o t f o r t h e whole d e s i g n community. I t h i n k that w e should d i s t i n g u i s h between
d i f f e r e n t phases. Before a n a i r c r a f t , o r a program, g e t s a f u l l - s c a l e developnent go-ahead, one needs, i n
t h e p r e - f e a s i b i l i t y / f e a s i b i l i t y s t u d y phase, t o be a b l e t o look a t many d i f f e r e n t c o n f i g u r a t i o n s t o go
through a l a r g e number o f d e s i g n l o o p s in a f a i r l y s h o r t time. I t h i n k t h e g e n e r a l complaint concerning
many computer programs is that t h e y are too cumbersome f o r day-to-day use. I n p a r t i c u l a r , when you want
t o h a v e a c c u r a t e r e s u l t s , you have t o i n c o r p o r a t e many d e t a i l s i n t h e r e p r e s e n t a t i o n of t h e
c o n f i g u r a t i o n . Consequently, yes, I t h i n k d a t a s h e e t s i n one way or a n o t h e r ( d a t a b a s e s , data
c o m p i l a t i o n s ) w i l l be r e q u i r e d i n t h e f u t u r e as well, b u t i n t h e f i r s t p l a c e f o r t h e i n i t i a l d e s i g n
phase. When you c o n t i n u e i n t h a t development p r o c e s s , you w i l l c o n c e n t r a t e more and more on a s p e c i f i c
c o n f i g u r a t i o n t h a t you want t o optimize and then of c o u r s e you w i l l have t o go t o more e l a b o r a t e processes
l i k e using complicated computer programs and complicated wind t u n n e l models.

Dr. J. McCroskey, NASA Ames Research Center, USA

Just a s h o r t comment t o respond to.David Woodward's q u e s t i o n about experimental techniques f o r l o o k i n g a t
t h e s e l i t t l e bubbles and shock wave boundary l a y e r i n t e r a c t i o n s . Many o f you w i l l remember that maestro
of flow v i s u a l i z a t i o n , Henri Werl6 a t ONERA, who, about 25 y e a r s ago, i n t r o d u c e d t h e i d e a t o study l e a d i n g
edge bubbles i n a small w a t e r t u n n e l by t a k i n g t h e l e a d i n g edge o f a n a i r f o i l and t e r m i n a t i n g i t w i t h a
blown f l a p , t o be a b l e t o g e n e r a t e h i g h s u c t i o n on t h e l e a d i n g edge. With t h i s c o n f i g u r a t i o n , he was a b l e
t o increase t h e s i z e o f t h e model by approximately a f a c t o r o f 10. I have thought many times that t h i s
should b e a u s e f u l technique i n o t h e r t y p e s o f i n v e s t i g a t i o n s . I proposed i t f o r dynamic stall
i n v e s t i g a t i o n s on a i r f o i l s , b u t i t seems t o me that t h i s would be a t e c h n i q u e o f use i n a compressible
high subsonic wind t u n n e l .

M r . A. B. Haines
Sorry t o appear a g a i n , b u t I wanted t o respond t o t h e gentleman i n t h e f r o n t who wanted evidence on t h e
e f f e c t s o f caps and s e a l i n g o f slats. There was q u i t e a l o t o f work done a b o u t 1965 i n t h e development o f
t h e Super VC 1 0 on t h a t s u b j e c t . Broadly speaking, of c o u r s e , a g r e a t d e a l of work h a s been done by
v a r i o u s people on s t e p s , r i d g e s , e p s , by t e s t i n g t h e s e on t h e walls o f wind t u n n e l s i n c l u d i n g t h e 8 f o o t
t u n n e l a t DRA Bedford. The r e s u l t s on t h e slats was c o n s i s t e n t w i t h t h e s e g e n e r a l results provided you
took account o f t h e f a c t that t h e l o c a l Mach numbers in t h e c r u i s e could b e v e r y h i g h , in t h e neighborhood
o f where t h e back end of t h e slat f i t t e d on t h e upper s u r f a c e . Also, you could do q u i t e a l o t o f damge
t o t h e wave drag, s o i t i s n o t j u s t a c a s e o f a s e r i o u s e f f e c t on t h e v i s c o u s drag. You can q u i t e modify
t h e d e v e l o p e n t o f t h e s u p e r c r i t i c a l r e g i o n and s o t h e i s s u e is important. But you won't f i n d t h e d a t a ,
because n o one wants t o admit t o having done i t badly. When t h e slat i s a good f i t , t h e e f f e c t s a r e
trivial.

P r o f . S. De Ponte, P o l i t e c n i c o , I t a l y
Everybody h a s d i s c u s s e d computation of two-dimensional flows, b u t I have n e v e r seen a two-dimensional
s e p a r a t i o n bubble, f o r example. I have seen s t r a i g h t s e p a r a t i o n l i n e s , s t r a i g h t a t t a c h m e n t l i n e s , b u t
three-dimensional flow i n s i d e t h e bubble. I have n e v e r seen a two-dimensional s e p a r a t i o n . So I wonder
why we compare c a l c u l a t i o n w i t h measurements i n wNch w e suppose t h e flow is two-dimensional, b u t we
measure three-dimensional flow in r e a l mesurements. How can we compare t h e s e f a c t s o r how can
two-dimensional c a l c u l a t i o n s be r e l i a b l e ?
 RTD-7

Mr. B. E l s e n a a r , N U
The q u e s t i o n of two-dimensionality has been worrying a working group o f AGARD t h a t had t o s e l e c t t e s t
cases f o r CFD v a l i d a t i o n d u r i n g t h e l a s t couple o f days. But t h e problem I t h i n k is probably more s e v e r e
in t r a n s o n i c s than i t is i n subsonic flow. That i s my p e r s o n a l view and I would l i k e t o e x p l a i n i t a
l i t t l e b i t . One t h i n g i s that two-dimensional s e p a r a t i o n bubbles, laminar s e p a r a t i o n bubbles, can be
e a s i l y v i s u a l i z e d on a two-dimensional a i r f o i l . That i s n o t a problem, t h e y are t h e r e . From t h e p i n t o f
view o f t h e t u r b u l e n t s e p a r a t i o n t h a t s t a r t s on t h e f l a p o r on t h e main a i r f o i l , i t ' s p o s s i b l e t o keep t h e
flow two-dimensional i n my view w i t h side-wall blowing or side-wall s u c t i o n . I watched numerous
experiments (I am o n l y familiar w i t h t h e blowing technique that we a p p l y a t NLR) where t h e blowing was
such t h a t you could a c t u a l l y see t h e s t a l l o f t h e a i r f o i l . You could v i s u a l i z e t h i s by p u t t i n g t u f t s on
t h e f l a p , and on t h e main a i r f o i l and you could r e a l l y see t h e complete a i r f o i l s t a l l a t a c e r t a i n point.
So, I t h i n k that two-dimensionality can be r e a l i z e d i n a n experiment. There i s one c r i t i c a l remark t h a t I
would l i k e t o make. T h a t i s t h e u s e o f f a i r i n g s and b r a c k e t s t o a t t a c h t h e slat and t h e f l a p t o t h e main
a i r f o i l . They i n t r o d u c e wakes and t h e s e wakes w i l l i n t r o d u c e a kind o f displacement i n t h e boundary l a y e r
and in t h e wake downstream. To d e r i v e drag from two-dimensional a i r f o i l s by means o f t h e wake-rake
t e c h n i q u e , t h e r e f o r e might be more cumbersome and mor.e troublesome. The d e t e r m i n a t i o n of d r a g f o r
two-dimensional a i r f o i l s is, I t h i n k , b o t h from t h e t h e o r e t i c a l p o i n t and from t h e e x p e r i m e n t a l p o i n t a
r a t h e r d i f f i c u l t t h i n g t o do. It i s s t i l l one o f t h e b a s i c q u e s t i o n s t h a t remains open f o r t h e f u t u r e .

Prof Ir. E. Obert, Fokker, Netherlands

The q u e s t i o n of why we are i n t e r e s t e d i n r e s e a r c h on two-dimensional a i r f o i l s and how good w e can r e a l l y
r e a l i z e two-dimensional flow should be coupled t o t h e q u e s t i o n o f what you are l o o k i n g f o r . From t h e
p o i n t of view of t h e d e s i g n e r you a r e l o o k i n g f o r maximum accuracy i n t h e p r e d i c t i o n of c e r t a i n t a r g e t s ,
c e r t a i n g o a l s . It on h i g h a s p e c t r a t i o wings w i t h n o t too much sweep, a knowledge o f two-dimensional
flow, how imperfect i t may be, can h e l p you i n s t r e n g t h e n i n g your b e l i e f i n your f i n a l p r e d i c t i o n , t h e n I
t h i n k t h e d e s i g n e r s are q u i t e happy. I f t h e r e are d i s c r e p a n c i e s between r e a l flow and t h e o r y , we a c c e p t
t h a t , b u t i t i s b e t t e r t h a n nothing.

Mr. B. Eggleston, de Haviland, Canada

Just coming back t o t h e comment a b o u t l a r n e f l a r , b r a c k e t s . as a u s e r o f large f l a D b r a c k e t s . I am
n a t u r a l l y s e n s i t i v e . We have done some & e r e s t i n g work h e r e i n f a c t we t e s t e d ;he models.flaps up w i t h
t h e b r a c k e t s i n p l a c e . Then t o get a h a n d l e on t h e penalty due t o t h e f l a p b r a c k e t s , w e pt i n s e r t s i n
and r e p e a t e d t h e test. They are s i g n i f i c a n t . There i s no doubt that i f you want good two-dimensional
flow you don't l e a v e t h o s e l a r g e f l a p b r a c k e t s on. I n t e r e s t i n g l y enough though, you may s t i l l want t o
have some o f t h e g p s t h a t are p r e s e n t i n t h e f l a p system when i t i s r e t r a c t e d . There can be minor
leakages through t h e gaps o f t h e f l a p system t h a t c a n attach flow r i g h t a t t h e back. You can see
S i g n i f i c a n t changes in drag. How you set t h a t arrangement up w i t h o u t having b r a c k e t s i s an i n t e r e s t i n g
c h a l l e n g e and you might t h i n k about how you a r e going t o do i t .

@
I am a l i t t l e s u r p r i s e d t h a t w e haven't heard more o f a plea f o r b e t t e r turbulence modelling. I t h i n k
t h a t i t i s f a i r t o s a y that i f we had a p e r f e c t t u r b u l e n c e model, which i s n ' t r e a l l y i n s i g h t , a l l we need
t o do i s put i t i n t o t h e Navier Stokes e q u a t i o n s and w e can s o l v e mre o r l e s s anything. The numerical
methods are coming a l o n g v e r y w e l l . To me i t seems that one o f t h e key a r e a s where t h e fundamental people
have t o f o c u s more a t t e n t i o n i s on t u r b u l e n c e models. Most o f what w e saw i n t h i s meeting w a s e i t h e r
mixing l e n g t h type models, Baldwin and Lomax and so on, o r K-epsilon models, that i s , two-equation
models. I t h i n k t h a t many of u s are aware t h a t even t h e two-equation m d e l s have q u i t e s e r i o u s
shortcoinings; t h e y a r e l i m i t e d t o c e r t a i n c a t e g o r i e s o f flow and you have t o tune t h e c o n s t a n t s and s o
on. I would put i n a p l e a , a s Dr. Elsenaar d i d , f o r more b u i l d i n g block t y p e o f experiments whose focus
i s on improving t h e t u r b u l e n c e models.

Another comment on t h e matter o f two-dimensional o r three-dimensional s e p a r a t e d flows. I t o o worry about

t h e n o t i o n o f a two-dimensional s e p a r a t i o n bubble. We know that, a t l e a s t i n a l o t o f c a s e s and depending
somewhat on a s p e c t r a t i o s , c a v i t y flows tend t o be i n h e r e n t l y three-dimensional, even i f w e suck t h e
end-walls and a l l t h e rest. Nevertheless, I t h i n k two-dimensional experiments o r nominally
two-dimensional experiments a r e t h e way t o go i n developing t u r b u l e n c e models, a t l e a s t i n i t i a l l y . I f we
c a n ' t d o i t in two-dimensional flows, we c e r t a i n l y c a n ' t do t o o w e l l in t h r e e .

D r . D. Woodward
Can I j u s t make a comment on t h e s e p a r a t i o n bubble i s s u e . With some r e s e a r c h done a t Queen Mary College, -
London; s t a r t i n g w i t h a n unswept s h o r t s e p a r a t i o n bubble which was induced on a p l a t e by an e x t e r n a l
p r e s s u r e d i s t r i b u t i o n and then t h i s p l a t e was swept around t o 30 degrees. I n f a c t , t h e flow developnent
normal t o t h e l e a d i n g edge is e x a c t l y t h e same i n both cases.

So, a t l e a s t f o r a s h o r t s e p a r a t i o n bubble, t h e a d d i t i o n of sweep, i n f a c t d o e s induce s i g n i f i c a n t flow

a l o n g t h e bubble in t h e reattachment r e g i o n , b u t t h e a c t u a l s i z e and t h e shape o f t h e bubble i s n o t
changed; i t i s e x a c t l y t h e same normal t o t h e l e a d i n g edge. This u n d e r l i n e s t h e f a c t t h a t
two-dimensionality may a c t u a l l y be a n a b s t r a c t concept which d o e s n ' t e x i s t p r e c i s e l y , p e d a n t i c a l l y i n
r e a l i t y , b u t n e v e r t h e l e s s i s a v e r y u s e f u l e n g i n e e r i n g concept.

M r . L.J. Williams
A t t h e r i s k of g i v i n g a n ' a n s w e r t h a t some o f u s my n o t want t o h e a r , I would l i k e t o a s k Ted Nark i f what
h e h a s seen the-last-few days would have changed t h e d e s i g n o f t h e YC 14.
RTD-8

Dr. T.C. Nark, Bceing, USA

I am a f r a i d I have n o information t h a t a d d r e s s e s t h e kind o f problems a s s o c i a t e d w i t h t h a t powered l i f t
configuration. For example, o u r half-model t e s t i n g w a s developmental i n n a t u r e . W e used it f o r
i n v e s t i g a t i n g d i f f e r e n c e s between c a n d i d a t e c o n f i g u r a t i o n s , r e c o g n i z i n g t h a t t h e r e might be some e r r o r i n
t h e a b s o l u t e v a l u e . Understanding t h a t w e had a d e s i g n s c h e d u l e , we d i d t h e b e s t we could w i t h t h e t o o l s
a v a i l a b l e a t t h e time. I f t h e t a s k had been t o d e s i g n t h e "best" powered l i f t system, I would have needed
b e t t e r e n g i n e e r i n g t o o l s t h a n were t h e n a v a i l a b l e . By u s i n g t e s t i n g r a t h e r t h a n a n a l y s i s I accepted t h e
inadequacies o f t e s t i n g and made u s e o f t h e b e n e f i t s . That was t h e time i n t h e program when i f you d i d n ' t
g e t t h e job accomplished today, t h e r e would b e no a i r p l a n e tomorrow. You n.ow have t h e c a p a b i l i t y o f
developing new t o o l s t h a t would have made my job e a s i e r i n t h e past and w i l l m k e t h e same job e a s i e r i n
the future.

p r o f . Dr. Ir. J.L. Van Ingen, D e l f t I h i v e r s i t y

I wanted t o comment on t h e i s s u e o f 2-D s e p a r a t i o n bubbles. I t h i n k I have watched more low speed
s e p a r a t i o n bubbles i n my l i f e t h a n anybody else i n t h e audience. Indeed, i f you look a t a s e p a r a t i o n
bubble f i l l e d w i t h smoke, even i f i t i s nominally 2-D, you see 3-D flows. O f course, where you see i t i s
in t h e low momentum p a r t s o f t h e bubble, and t h e s e a p p a r e n t l y are n o t determining t h e wera11 shape. I f
you u s e a 2-D c a l c u l a t i o n f o r a s e p a r a t i o n bubble, you g e t more o r l e s s t h e r i g h t answers.

Prof. Ir. E. Obert

D r . Woodward mentioned a t t h e beginning o f t h i s conference t h a t t h e r e a r e c e r t a i n phenanenae which have
misunderstood in t h e p a s t o r have been d e s c r i b e d i n a c e r t a i n e r r o n e o u s way which t h e n are passed on from
g e n e r a t i o n t o g e n e r a t i o n f o r decades, f o r example, t h e f a c t t h a t a l e a d i n g edge s l o t o r a f l a p s l o t
r e - a e r g i z e s a boundary l a y e r . I would l i k e t o s h a r e an e x p e r i e n c e that I w e n t through when I was
c o l l e c t i n g m a t e r i a l f o r my l e c t u r e n o t e s . That concerns t h e s t a l l i n g c h a r a c t e r i s t i c s o f h i g h a s p e c t r a t i o
swept wings. From o u r i n i t i a l e x p e r i e n c e s and o b s e r v a t i o n s in t h e l a t e 1940's w e have come up w i t h t h i s
p i c t u r e t h a t h i g h l y swept s l e n d e r o r high a s p e c t r a t i o swept wings produce t i p stall, i f you have more o r
less t h e same a i r f o i l s e c t i o n and n o t too much t w i s t . This is caused by a n outward flow o f t h e boundary
l a y e r . Consequently, t h e t i p stalls m c h earlier t h a n p u would expect on t h e b a s i s o f t h e e q u i v a l e n t
two-dimensional flow s i t u a t i o n . I found an NACA r e p o r t from t h e 1 9 5 0 ' s (NACA RM A52A10) where tests on
t h e following h a l f model were r e p o r t e d .

-
Oirnensionr of tAe rarnispon wing fuselope models including tAe oritice riotion
locotions ond the tnist distribution lor the cornbemj, Iwisted wing.

I
 RTD-9

L2-

/
/-- ---------. \

.-
LO

/
/
/
/

,/
1
!A "\\, \

I Id 1 - I I I'\I
.8 /' 'plain' wing
.$
.t! b
3

*.
a
e
.4
/I I H
,y I I I I I I
i i
I
I
..?* I

I l l 1 I &7:
I

Relation between section and wing maximum lift coefficient. (I)

0 4 8 I2 I6 20 24 (for v*.l67J
Angle of atfack, a, deg

(aJ Plain wing.

Fkjwre 8.- Comparisons of locaf lift curves on the wing models with
-
those derived from two dimensions/ data and span loading theory.
RTD-10

R e l a t i o n between s e c t i o n and wing maximum l i f t c o e f f i c i e n t . (11)

A s you s e e , t h e r e were a number o f wing s e c t i o n s equipped w i t h p r e s s u r e t a p p i n g . This wing w a s t e s t e d i n

two shapes, one as a f l a t wing w i t h NACA 64A010 a i r f o i l s e c t i o n and a n o t h e r one based on a h i g h l y cambered
a i r f o i l s e c t i o n NACA 64 A810, a d e s i g n l i f t c o e f f i c i e n t o f 0.8 and h i g h l y t w i s t e d . If one now l o o k s a t
t h e CL-alpha c u r v e s f o r t h e v a r i o u s s e c t i o n s one s e e s that a t t h e t i p t h e CL-alpha curve i s v e r y
s i m i l a r t o t h e e q u i v a l e n t curve i n sheared two-dimensional flow whereas f u r t h e r inboard t h e CL-alpha
c u r v e s c o n t i n u e t o much higher angles-of-attack. I n o t h e r words, t h e t i p s t a l l is n o t caused because t h e
t i p stalls earlier t h a n what one would expect, b u t t h e t i p s t a l l o c c u r s . b e c a u s e t h e i n b o a r d wing keeps
a t t a c h e d flow much l o n g e r than you would e x p e c t , a p p a r e n t l y by some s o r t of boundary l a y e r removal on t h e
inboard wing. I t h i n k m s t o f us s t i l l t e l l o u r younger c o l l a b o r a t o r s and s t u d e n t s t h e o r i g i n a l s t o r y .
That i s w h a t I wanted t o s h a r e w i t h you concerning h i g h l i f t c h a r a c t e r i s t i c s on swept wings.

M r . L.J. Williams
Thank you f o r t h a t i n s i g h t . Now are t h e r e any o t h e r comments? I would j u s t l i k e t o wrap i t up with a
c o u p l e o f comments. Although I a g r e e w i t h David t h a t w e achieved t h e o b j e c t i v e s t h a t w e had i n mind when
t h i s Symposium was conceived two and a h a l f y e a r s ago, w e never i n t e n d e d - t h a t t h i s Conference be anything
more than a s t a t u s r e p o r t on t h e development o f h i g h - l i f t system aerodynamic technology. W e never
expected t h a t i t would be t h e f i n a l s a y on t h e m a t t e r . I t h i n k w i t h t h a t i n mind I would l i k e t o s e e t h i s
be a Symposium t h a t happens about every f i v e y e a r s so that w e could follow t h e developments i n h i g h - l i f t
technology and perhaps even speed i t up. One o f t h e t h i n g s t h a t we would l i k e t o have s e e n i n t h i s
Symposium a l i t t l e more than we d i d a r e some new i n n o v a t i v e i d e a s . With that i n mind I would l i k e t o a s k
DDn Whittley t o g i v e us a f e e l i n g f o r what Zurakowski would have s a i d i f h e w a s w i t h u s today.

M r . D.C. Whittley
I happen t o be s h a r i n g a few p a s t experiences w i t h Lou Williams t h e o t h e r day a b o u t t h e v e r y f a m u s test
p i l o t - Janus Zurakowski. I d o n ' t know how many o f you know o f Janus Zurakowski. Does that name mean
anything? J u s t a few. He was a P o l i s h f i g h t e r p i l o t d u r i n g t h e Battle o f B r i t a i n and a f t e r t h e war h e
became A s s i s t a n t Chief Test P i l o t a t G l o u c e s t e r A i r c r a f t . As such he was d o i n g a l o t o f f l i g h t t e s t i n g on
t h e G l o u c e s t e r Meteor, which you may remember was one o f t h e e a r l y t w i n e n g i n e j e t f i g h t e r a i r c r a f t .
Zurakowski invented one o f t h e o n l y new manoeuvres s i n c e t h e Immelmann t u r n . T h i s was a c t u a l l y w r i t t e n up
i n TIME magazine. H i s a e r o b a t i c s were known as Zurabatics. The a e r o b a t i c manoeuvre h e performed w a s
known as t h e Zura cartwheel. H e would t a k e t h e Meteor a i r c r a f t and climb v e r t i c a l l y t o z e r o speed: once
t h e a i r c r a f t reached z e r o speed he would c u t one engine and t h e a i r p l a n e would do a c a r t w h e e l , go i n t o a
s p i n and come down, a l l w i t h i n t h e immediate v i c i n i t y of t h e a i r f i e l d . T h i s was performed a t t h e
Farnborough Airshow on a number of occasions. He l e f t t h e U.K. and j o i n e d t h e t e a m o f AVRO Canada, f l y i n g
CF 1 0 0 ' s and he was t h e p i l o t t o f l y f i r s t t h e CF 105 s u p e r s o n i c delta-wing f i g h t e r . He was always a v e r y
i n g e n i o u s kind o f a guy and v e r y i n t e r e s t e d i n t h e t e c h n i c a l background t o t h e f l y i n g program. He had
b r i g h t b l u e e y e s and a deep s t a r e and a ruddy complection. Very o f t e n he would come i n t o my o f f i c e and
w i t h a t w i n k l e In h i s eye and i n a v e r y high p i t c h e d P o l i s h broken a c c e n t he would s a y , "anything new o r
funny". This r a t h e r t i c k l e d Lou, as I had been s h a r i n g w i t h him t h e i d e a that we h a d n ' t r e a l l y looked a t
v e r y much h e r e i n Banff which i s r e a l l y new o r funny (such as t h e v o r t e x f l a p , f o r i n s t a n c e ) , b u t maybe
one o f t h e s e days w e can.
 RTD-11

M r . L.J. Williams
Maybe t h e i d e a s a r e o u t t h e r e b u t people a r e j u s t too embarrassed t o b r i n g them forward i n t h i s p a r t i c u l a r
s e t t i n g . Maybe i n t h e bar l a t e r . A t t h i s p o i n t I would l i k e t o thank Kazik i n p a r t i c u l a r f o r s u g g e s t i n g
t h i s a s a venue f o r t h i s Symposium. I t h i n k a l o t o f t h e success o f t h e Symposium i s due t o t h e venue,
t h e H o t e l i s p e r f e c t , t h e s i z e of t h e conference room is p e r f e c t , t h e accommodations were e x c e l l e n t , and
i t was amazing t h a t even w i t h t h e o u t s t a n d i n g s c e n e r y and t h e r e s t a u r a n t s i n t h e area, I would have t o s a y
t h a t t h e f a c t t h a t w e have had a f u l l a t t e n d a n c e f o r t h e whole Symposium v a l i d a t e d t h e d e d i c a t i o n of t h e
a t t e n d e e s more t h a n a n y t h i n g e l s e . A t t h i s p o i n t I would l i k e t o t u r n i t over t o our FDP Chairman.

Prof. Ir. J . W . S l o o f f
Thank you, Lou. Ladies and Gentlemen, t h i s is t h e end o f t h r e e and a h a l f v e r y i n t e r e s t i n g days, It is
my duty t o b r i n g i t , w i t h some r e l u c t a n c e , t o a n o f f i c i a l end. I t h i n k t h e days were i n t e r e s t i n g i n t h e
t e c h n i c a l a s w e l l as t h e r e c r e a t i o n a l s e n s e t h a t Lou a l r e a d y h i n t e d a t .

T h i s meeting and i t s s u c c e s s would n o t have been p o s s i b l e without t h e hard work and h e l p o f s e v e r a l

people. F i r s t o f a l l , I a l s o would l i k e t o thank our Canadian h o s t s f o r i n v i t i n g u s h e r e and i n
p a r t i c u l a r t h e l o c a l , b u t n o t so v e r y l o c a l , c o o r d i n a t o r , D r . Kazik Orlik-Ruckemann and h i s co-panel
member a l s o from Canada, Prof. Dick Kind f o r s e l e c t i n g t h i s s i t e , Banff, and i t s Park Lodge w i t h i t s
e x c e l l e n t f a c i l i t i e s . Kazik and Dick, p l e a s e convey o u r thanks t o t h e Canadian N a t i o n a l Delegates and
thank you p e r s o n a l l y f o r b r i n g i n g u s h e r e and exposing u s t o t h e beauty o f t h e Canadian Rockies and i t s
N a t i o n a l Parks. W e c e r t a i n l y w i l l remember t h e i c e f i e l d s , Lake Louise, t h e e l k and t h e moose and t h e
s q u i r r e l s and t h e chipmunks.

Switching t o t h e primary reason f o r o u r presence h e r e , t h a t i s o u r meeting on High-Lift Aerodynamics, I

would l i k e t o compliment t h e Program Committee w i t h t h e s u c c e s s of t h e meeting. I n p a r t i c u l a r , of c o u r s e ,
Dr. David Woodward and Dr. Lou W i l l i a m s f o r t h e i r i n i t i a t i v e s and e f f o r t s . We a l s o thank Dr. Whittley f o r
h i s on-the-spot t e c h n i c a l e v a l u a t i o n of t h e meeting, which i s a v e r y hard t h i n g t o do. We a r e c e r t a i n l y
l o o k i n g forward t o your w r i t t e n report on t h e meeting. Thank you v e r y much.

Thanks are a l s o due t o a number o f people t h a t have a s s i s t e d u s t o run t h e meeting smoothly. A t t h e desk
we were a s s i s t e d by two charming l a d i e s , Mrs. Alison Day and Mrs. Sarah Pearson. W e were a s s i s t e d by a
team of i n t e r p r e t e r s who v e r y o f t e n , I am s u r e , had t h e d i f f i c u l t j o b o f keeping up w i t h t h e pace of some
o f o u r speakers. I would l i k e you t o applaud a l s o f o r Teresa Beauregard, Susan Ourion, Carol Savard and
Madeleine T e r r i e n . There i s a l s o t h e t e c h n i c i a n f o r t h e audio v i s u a l a i d s . W e thank J a s o n Penner f o r
s e r v i n g our eyes and ears. L a s t b u t n o t l e a s t I would l i k e t o thank our Executive, Winston Goodrich and
h i s s e c r e t a r y , Anne-Marie R i v a u l t f o r t h e i r i n d i s p e n s a b l e and as u s u a l h i g h l y s u c c e s s f u l e f f o r t s t o r u n
t h e meeting smoothly.

F i n a l l y , I would l i k e t o draw your a t t e n t i o n t o some of t h e f u t u r e a c t i v i t i e s o f t h e Fluid Dynamics

Panel. Our next meeting i s i n Winchester on "Computational and Experimental ssessments of Jets i n
Cross-Flow". W e have a f a l l meeting i n Brussels next y e a r on "Wall I n t e r f e r e n c e Support and I n t e r f e r e n c e
i n Flow Measurements" and we have a number of S p e c i a l Courses planned t h a t you can a l s o see on t h i s
vu-graph. I n t h i s r a p i d l y changing world t h e mission and t a s k of NATO w i t h AGARD i n i t s wake, a r e i n t h e
p r o c e s s o f r e o r i e n t a t i o n and are s u b j e c t o f d i s c u s s i o n . I am s u r e t h a t you w i l l a g r e e w i t h me t h a t AGARD
has a v e r y good r e a s o n t o c o n t i n u e i t s work i n t h e s p i r i t o f t h i s p a s t week. Perhaps w i t h a l i t t l e
r e o r i e n t a t i o n h e r e and t h e r e , and perhaps a l i t t l e r e d i r e c t i o n o f emphasis o r a t t e n t i o n , but c e r t a i n l y not
i n a d i s r u p t i v e way. You can s u p p o r t AGARD by d i s s e m i n a t i n g t h e i n f o r m t i o n on , t h i s vu-graph and t h e
message o f AGARD i n g e n e r a l . I might add t h a t a s a r e s u l t of t h e changing world, w e w i l l , i n a l l
p r o b a b i l i t y , have a t our f a l l meeting i n B r u s s e l s , t h e f i r s t c o n t r i b u t i o n from a s c i e n t i s t from t h e former
Soviet Union. That i s something t h a t , I am s u r e , we are a l l l o o k i n g forward t o .

T h i s r e a l l y i s t h e end. Thank you a l l f o r your p a t i e n c e and f o r your personal c o n t r i b u t i o n s t o t h e

s u c c e s s of t h i s Symposium. I hope t h a t you will be a b l e t o s t a y a f e w days l o n g e r to f u r t h e r enjoy t h e
Rockies. I wish you a l l a s a f e journey home.
 REPORT DOCUMENTATION PAGE
1. Recipient’s Reference 2. Originator’s Reference 3. Further Reference 4. Security Classification
of Document
AGARD-CP-5 15 ISBN 92-835-0715-0 UNCLASSIFIED/
UNLIMITED

8. Author(s)/Editor(s) 9. Date
Various September 1993

10. Author’dEditor’s Address 11. Pages

Various
508

12- DistributionStatement There are no restrictions on the distribution of this document.

Information about the availability of this and other AGARD
unclassified publications is given on the back cover.
13. Keywords/Descriptors
High lift devices wing slots
Aerodynamics Boundary layer control
Design Model tests
Flaps - control surfaces Wind tunnel tests
Leading edge slats

14. Abstract
The AGARD Fluid Dynamics Panel (FDP) sponsored a Symposium on “High-Lift System
Aerodynamics” which was held 5th-8th October 1992 in Banff,Alberta, Canada. This report
includes the 32 technical papers developed for the Symposium along with an edited transcript of
the Round Table Discussion held at the end which addressed issues raised during the Symposium.
In addition, the Technical Evaluation Report - aimed at assessing the success of the Symposium
in meeting its objectives and its relevance to the technical needs of the aerospace community - is
also included in this document.

The Symposium objectives were to address:

(1) how, given the role and configuration of an aircraft, the most appropriate high-lift system
can be selected,
(2) how, having selected a type of high-lift system, an efficient detailed design can be produced,
and
(3) the experimental and analysis techniques which are necessary to explore and enhance the
performance of a high-lift system.

Thus, although the aerodynamics of high-lift systems was the dominant theme, the very pertinent
aspects of weight, simplicity, reliability, and structural and mechanical integrity were an integral
part of the Symposium, and were treated in the papers presented. In this sense, this Symposium
has attempted to take a broader view of the high-lift system than has been taken by similar
conferences in the past.
 0
2

-
C
F
P
n n n
d
o
w
v
h)
u
rcn

n n n
d
o
w h) r w
w w u y
 NATO -@-OTAN
7 RUE ANCELLE * 92200 NEUILLY-SUR-SEINE DIFFUSION DES PUBLICATIONS
FRANCE AGARD NON CLASSIFIEES
TBIecopie (1)47.38.57.99 TBlex 610 176

Aucun stock de publications n'a existi 1AGARD. A partir de 1993, AGARD ditiendra un stock limit6 des publications associkesaux cyclesde
confirences et cours spiciaux ainsi que les AGARDographieset les rap orts des groupes de travail, organisis et publiis 1partir de 1993inclus.
P
Les demandesderenseignementsdoiventCtreadresskes1AGARDpar ettreou parfax1I'adresseindiquieci-dessus.Veurllez nepastkl&phoner.
La diffusion initialede toutes les publications del'AGARD est effectukeaupres des pays membres de I'OTAN par I'intermidiairedes centres
de distribution nationaux indiquis ci-dessous. Des exemplaires su Iimentaires euvent parfois itre obtenus auprbs de ces centres (1
I'exception des Etats-Unis). Si vous souhaitez re evoir toutes les pulEcations de I'lGARD, ou simplement cella qui concernent certains
Panels, vous pouvez demander a itre inch sur lafiste d'envoi de I'un de ces centres. Les publications de I'AGARD sont en vente aupres des
agences indiquies ci-dessous, sous forme de photocopie ou de microfiche.
CENTRES DE DIFFUSION NATIONAUX
ALLEMAGNE ISLANDE
Fachinformationszentrum, I Director of Aviation
Karlsruhe c/o Flugrad
D-7514 Eggenstein-Leopoldshafen2 Reykjavik
BELGIQUE ITALIE
Coordonnateur AGARD-VSL Aeronautica Militare
Etat-Major de la Force Airienne Uhficio del Delegato Nazionale all'AGARD
Quartier Reine Elisabeth Aeroporto Pratica di Mare
Rue d'Evere, 1140 Bruxelles 00040 Pomezia fRoma)
CANADA LUXEMBOURG
Directeur du Service des Renseignements Scientifiques Voir Belgique
Ministitre de la DBfense Nationale
Ottawa, Ontario K1A OK2 NORVEGE
DANEMARK Norwegian Defence Research Establishment
Danish Defence Research Establishment Attn: Biblioteket
Ryvangs All6 1 P.O. Box 25
P.O. Box 2715 N-2007 Kjeller
DK-2100 Copenhagen 0 PAYS-BAS
ESPAGNE Netherlands Delegation to AGARD
INTA AGARD Publications)
k
Pintor osales 34
28008 Madrid
National Aerospace Laboratory NLR
P.O. Box 90502
1006 BM Amsterdam
ETATS-UNIS PORTUGAL
National Aeronautics and Space Administration Forqa Airea Portuguesa
Langley Research Center Centro de DocumentaGHo e Informa@o
M/S 180 Alfragide
Hampton, Virginia 23665 2700 Amadora
FRANCE
O.N.E.R.A. (Direction) ROYAUME UN1
29, Avenue de la Dividon Leclerc Defence Research Information Centre
92322 Chitillon Cedex Kentigern House
65 Brown Street
GRECE Glasgow G2 SEX
Hellenic Air Force
Air War College TURQUIE
Milli Savunma Bagkanlib MSB
Scientific and Technical Library
Dekelia Air Force Base
Dekelia, Athens TGA 1010
ARGE Daire Ba$kanliB ( RG )
Ankara
a 2
Le centre de distribution national des Etats-Unis (NASNLangley) ne detient PAS de stocks des publications de I'AGARD.
D'kentuelles demandes de photocopies doivent Ctre formulies directement auprb du NASA Center for Aerospace Information
(CASI) l'adresse suivante:
AGENCES DE VENTE
NASA Center for ESA/Information Retrieval Service The British Library
Aerospace Information (CASI) European Space Agency Document Supply Division
800 Elkridge Landing Road 10, rue Mario Nikis Boston Spa, Wetherby
Linthicum Heights, MD 21090-2934 75015 Paris West Yorkshire LS23 7BQ
United States France Royaume Uni
Les demandes de microfiches ou de photocopies de documents AGARD (y compris les demandes faites aupres du CASI) doivent comporter
la dtnomination AGARD, ainsi quelenumiro de sine d'AGARD (par exemple AGARD-AG-3 15 Des informations analogues,telles que
RI)
le titre et la bate de publication sont souhaitables. Veuiller noter qu'il y a lieu de spkcifier AGA -R-nnn et AGARD-AR-nnn lors de la
commande des rapports AGARD et des rapports consultatifs AGARD respectivement. Des rifirences bibliographiques completes ainsi
que des risum6s des publications AGARD figurent dans lesjoumaux suivants:
Scientifique and Technical Aerospace Reports (STAR) Government Reports Announcements and Index (GRA&I)
publi6 par la NASA Scientific and Technical publii par le National Technical Information Semce
Information Program Springfield
NASA Headquarters (J'IT) Virginia 22 161
Washington D.C. 20546 Etats-Unis
Etats-Unis (accessible igalement en mode interactif dans la base de
donnies bibliographiques en ligne du NTIS, et sur CD-ROM)

M
Imprime'par Specialised Printing Services Limited
40 Chigwell Lane, Loughton, Essex IGlO 312
 r

NATO -&- OTAN

7 RUE ANCELLE 92200 NEUILLY-S.UR-SEINE DISTRIBUTION OF UNCLASSIFIED
FRANCE AGARD PUBLICATIONS
Telefax (1)47.38.57.99 * Telex 6 IO 176

AGARD holds limited quantities of the publications that accompanied Lecture Series and Special Courses held in 1993 or later, and of
AGARDographs and Working Group reports published from 1993onward. For details, write or send a telefax to the address given above.
Please do not telephone.
AGARD does not hold stocks of publications that accompanied earlier Lecture Series or Courses or of any other publications. Initial
distributionofall AGARDpublicationsis made toNATOnations throughtheNationalDistributionCentres listed below.Furthercopiesare
sometimes available from these centres except in the United States). If you have a need to receive all AGARD publications, or just those
b
relating to one or more specificAGARD anels, they may be willing to include you (or yourorganisation)on their distribution list. AGARD
publications may be purchased from the Sales Agencies listed below, in photocopy or microfiche form.
NATIONAL DISTRIBUTION CENTRES
BELGIUM LUXEMBOURG
Coordonnateur AGARD - VSL See Belgium
Etat-Major de la Force Akrienne
Quartier Reine Elisabeth NETHERLANDS
Rue d‘Evere, 1140 Bruxelles Netherlands Delegation to AGARD
National Aerospace Laboratory, NLR
CANADA P.O. Box 90502
Director Scientific Information Services 1006 BM Amsterdam
Dept of National Defence
Ottawa, Ontario KIA OK2 NORWAY
DENMARK Norwegian Defence Research Establishment
Danish Defence Research Establishment Attn: B:blioteket
Ryvangs All6 1 P.O. Box 25
P.O. Box 27 15 N-2007 Kjeller
DK-2 100 Copenhagen 0 PORTUGAL
FRANCE Forsa ABrea Portuguesa
O.N.E.R.A. (Direction) Centro de DocumentaqHoe InformaCio
29 Avenue de la Division Leclerc AJfragide
92322 Chitillon Cedex 2700 Amadora
GERMANY SPAIN
Fachinformationszentrum INTA AGARD Publications)
Karlsruhe
D-75 14 Eggenstein-Leopoldshafen2
k
Pintor osales 34
28008 Madrid
GREECE TURKEY
Hellenic Air Force Milli Savunma Ba$kanli&(MSB
Air War College ARGE Daire Baskanli& (ARGli ) b

Scientific and Technical Library Ankara

Dekelia Air Force Base
Dekelia, Athens TGA 1010 UNITED KINGDOM
ICELAND Defence Research Information Centre
Director of Aviation Kentigern House
c/o Flugrad 65 Brown Street
Reykjavik Glasgow G2 8EX
ITALY UNITED STATES
Aeronautica Militare National Aeronautics and Space Administration (NASA)
Ufficio del Delegato Nazionale all’AGARD Langley Research Center
Aeroporto Pratica di Mare M/S 180
00040 Pomezia (Roma) Hampton, Virginia 23665
The United States National Distribution Centre (NASNLangley) does NOT hold stocks of AGARD publications.
Applications for copies should be made direct to the NASA Center for Aerospace Information (CASI) at the address below.
SALES AGENCIES
NASA Center for ESNInformation Retrieval Service The British Library
Aerospace Information (CASI) European Space Agency Document Supply Centre
800 Elkridge Landing Road 10,rue Mario Nikis Boston Spa, Wetherby
Linthicum Heights, MD 21090-2934 75015 Paris West Yorkshire LS23 7BQ
United States France United Kingdom
Requests for microfichesor photocopies of AGARD documents (includingrequests to CASI) should include the word ‘AGARD’and the
B
AGARD serial number (for example AGARD-AG-315 .Collateral information such as title and publication date is desirable. Note that
AGARD Reports and Advisory Reports should be speci ed as AGARD-R-nnn and AGARD-AR-nnn, respectively.Full bibliographical
references and abstracts of AGARD publications are given in the following journals:
Scientific and Technical Aerospace Reports (STAR) Government Reports Announcements and Index GRA&I)
published bv NASA Scientific and Technical nuhlished hv !he Natinnal Technical Infnrmarinn &ervice
lnfot
NAI
Was
Unit

ISBN 92-835-0715-0 ( 1
7 .
c